cs.CL [Total: 27]
cs.CV [Total: 56]
cs.MM [Total: 1]
eess.IV [Total: 3]
cs.CR [Total: 1]
cs.IR [Total: 1]
cs.AI [Total: 6]
cs.RO [Total: 2]
cs.LO [Total: 1]

cs.CL [Back]

[1] When Vocabulary Comprehension Fails Clinical Reasoning: Evaluating Therapy Bots’ Safety Risks for Generation Alpha cs.CL | cs.AI | cs.CYPDF

Manisha Mehta, Virendra Mehta

TL;DR: 本文评估了面向Alpha世代（2010-2024年出生）的心理健康对话AI系统的安全风险。研究发现，尽管主流大语言模型（如Claude、GPT-4o、Llama-3.1）能理解76-82%的词汇，但其临床风险评估准确率仅为64-72%，存在10-14个百分点的词汇理解与临床推理差距，且该差距在表达模糊时进一步扩大。研究识别出六种典型的失败模式，并指出现有轻量级缓解措施无效，仅高成本的人工介入架构能达到人类治疗师水平。

Details

Motivation: 鉴于有13.1%的美国青少年使用生成式AI寻求心理健康建议，且已发生多起与AI聊天机器人交互相关的青少年死亡事件，亟需系统评估这些依赖大语言模型的对话系统对青少年特有的夸张语言、反讽式积极、快速语义漂移和语境多义性等沟通模式的安全性。

Result: 在构建的两个基准测试（64个经认证的青少年心理健康表达和75组多轮对话）上评估发现，模型存在显著的词汇理解-临床推理差距（10-14个百分点，p<.001），而人类治疗师的差距仅为3个百分点（p=.22）。该差距在架构间表现一致，并在模糊性高时从7个百分点扩大至18个百分点。基线34%的漏报率估计导致每年146,880次危机被遗漏。

Insight: 论文的创新点在于构建了针对青少年心理健康沟通模式的专用评估基准，并系统性地量化了AI在词汇理解与临床风险评估之间的关键差距。客观来看，其识别出的六种具体失败模式（如反讽掩饰、最小化接受、非正式风格偏见等）为理解AI在复杂、非标准语言环境下的局限性提供了清晰框架，强调了面向青少年的心理健康AI必须采用人工介入架构、定期专项验证并建立监管框架。

Abstract: Conversational AI systems have become informal mental health support resources for Generation Alpha (Gen Alpha, born 2010-2024), with 13.1% of U.S. adolescents (5.4 million) using generative AI for mental health advice. While these systems, from therapy apps to general chatbots, rely on large language models trained on extensive psychological literature, their safety for youth communication patterns characterized by hyperbolic language, ironic positivity, rapid semantic drift, and contextual polysemy remains unvalidated. Following multiple adolescent deaths linked to AI chatbot interactions, systematic evaluation is critical. We present two benchmarks: (1) 64 Gen Alpha mental health expressions validated by native speakers (ICC=0.72) and clinicians (kappa=0.78); (2) 75 multi-turn conversations (780 turns) with paired Standard/Gen Alpha versions. Across evaluations of LLM architectures underlying therapy apps and general chatbots - Claude, GPT-4o, Llama-3.1 - models understand 76-82% of vocabulary but correctly calibrate only 64-72% of clinical risk, creating a 10-14 percentage point (pp) vocabulary-comprehension gap (p<.001, d>0.48) absent in human therapists (3pp, p=.22). The gap is architecturally consistent and widens with ambiguity (7pp -> 18pp). We identify six failure patterns: sarcasm masking (29pp), minimization acceptance (43pp), informal style bias (24pp), risk-stratified ambiguity (19pp), semantic drift (19pp), context-dependent violence (7pp). Patterns compound; three or more yield 94% miss rates. Lightweight mitigations fail; only heavy scaffolding achieves human performance (6.4x cost). With 34% baseline miss rate yielding 146,880 estimated annual missed crises, we recommend mandatory human-in-the-loop architectures, quarterly youth-specific validation, transparent performance disclosure, and regulatory frameworks for youth-facing mental health AI.

[2] Beyond Prompt Engineering: A Systematic Analysis of Prompt Lexical Sensitivity and Its Impacts on Quality cs.CL | cs.AIPDF

Qipeng Xie, Zi Liang, Jiafei Wu, Yufei Chen, Weizheng Wang

TL;DR: 本文系统分析了大型语言模型（LLM）对提示词表面词汇变化的极端敏感性，并揭示了提示性能稳定性的缩放定律：平均任务性能越高，其在不同提示扰动下的方差越低、鲁棒性越强。研究识别了驱动鲁棒性的两个核心语言因素：领域特定术语和显式行动指令，并基于此开发了一个自动提示优化代理，在代码生成任务中将性能方差降低了40.7%，同时保持或提升了平均性能。

Details

Motivation: 针对LLM对提示词微小词汇变化表现出的性能剧烈波动问题，超越黑盒优化和粗粒度模板方法，旨在通过大规模、细粒度的机制分析来理解并提升提示的稳定性。

Result: 在包含132,000个提示变体的数据集上进行实证评估，所提出的自动提示优化方法在代码生成任务中将性能方差降低了40.7%，同时保持或改进了平均性能。

Insight: 创新点在于首次进行了大规模、n-gram词元级别的提示稳定性机制分析，并揭示了性能与鲁棒性之间的缩放定律。客观来看，其提出的通过注入领域锚定和操作约束来系统重构查询的自动化方法，为构建鲁棒的提示工程提供了统计基础和可解释的框架。

Abstract: Large Language Models (LLMs) exhibit extreme sensitivity to surface-level prompt variations, in which minor lexical changes can trigger disproportionate performance fluctuations. Moving beyond black-box optimization and coarse-grained templates, we present the first large-scale, n-gram token-level mechanistic analysis of prompt stability, leveraging a dataset of 132,000 prompt variants. Our investigation reveals a fundamental Scaling Law of Prompt Performance Stability: higher average task performance is strongly associated with lower variance and greater robustness across prompt perturbation. We identify two core linguistic drivers underlying this robustness: (1) Domain-Specific Terminology, which tightly anchors semantic boundaries, and (2) Explicit Action Directives, which formalize reasoning trajectories. Together, these elements constrain the model’s interpretative space, effectively ``locking in’’ more deterministic generation behavior. Building on these insights, we introduce an automated Prompt-Refining Agent that systematically restructures input queries by injecting domain anchoring and operational constraints. Empirical evaluation shows that our approach reduces performance variance by 40.7% in code generation task, while preserving or improving mean performance. These findings provide a statistically grounded and mechanistically interpretable framework for achieving robust prompt engineering.

[3] Inhibitory Attention for Clinical Long-Context Reasoning: Characterizing and Mitigating Lost-in-the-Middle Effects in EHR Processing cs.CL | cs.AIPDF

Sanjay Basu

TL;DR: 这篇论文针对电子健康记录（EHR）处理中大型语言模型存在的‘迷失在中间’（LitM）效应，提出了‘临床迷失在中间’（CLitM）问题。作者使用MedAlign数据集首次系统性地描述了该问题，并引入了查询条件化临床抑制（QCCS）这一轻量级查询条件化选择门方法，以改善长上下文推理。实验表明，在关键的中段信息处理上，QCCS显著优于传统的检索方法（如BM25、密集检索）和完整上下文输入。

Details

Motivation: 动机在于解决临床场景中一个关键问题：电子健康记录（EHR）通常超过10万tokens，而大型语言模型在处理长上下文时，位于中间位置的重要信息（如病历的核心事实）检索可靠性显著下降，这被称为‘迷失在中间’（LitM）效应，在临床应用中可能导致严重后果。

Result: 在2,196个指令-响应对和六个语言模型的评估中，观察到准确率在峰值（59.5%，位于20-30%十分位）和谷值（37.6%，位于70-80%十分位）之间存在21.9个百分点的差距，67.8%的参考答案位于EHR时间线的10-90百分位之间，即CLitM低谷区域。使用Qwen2.5-7B-Instruct模型（16k上下文）在83个保留指令上的评估显示，QCCS在LLM-as-judge评分下优于所有比较方法：对于中间位置指令，QCCS达到16.7%的准确率，而BM25为3.3%，交叉编码器为0.0%，密集检索为0.0%，完整上下文为6.7%；总体QCCS达到25.3%，而仅检索方法最高为3.6%。

Insight: 论文的创新点在于首次系统性地定义了‘临床迷失在中间’（CLitM）问题，并提出了查询条件化临床抑制（QCCS）这一轻量级、查询对齐的上下文选择机制。关键洞察是：在临床长上下文推理中，与查询对齐的上下文选择（即使不总是检索到黄金证据句子）比传统的基于召回率的检索方法（如BM25）更能预测指令遵循的准确性，这挑战了单纯依赖检索召回率作为性能指标的传统观念。

Abstract: Electronic health records now routinely exceed 100,000 tokens per patient. Yet large language models exhibit the lost-in-the-middle (LitM) effect: information near the center of a long context is retrieved less reliably than information near the edges. In clinical use this is not benign: the single most consequential fact in a note can sit at its center. We term this the clinical lost-in-the-middle (CLitM) problem, give its first systematic characterization using MedAlign, and compare context-selection strategies as remedies. Across 2,196 instruction-response pairs and six language models, we observe a 21.9 percentage-point gap between peak accuracy (59.5%, 95% CI [46.3, 71.0], 20-30% decile) and trough accuracy (37.6% [23.2, 52.5] at 70-80%); 67.8% of reference answers fall between the 10th and 90th percentiles of the EHR timeline, inside the CLitM trough. We introduce Query-Conditioned Clinical Suppression (QCCS), a lightweight query-conditioned selection gate, and evaluate it against BM25, BM25 with section-header filtering, dense retrieval, and cross-encoder reranking (N=83 held-out instructions). With Qwen2.5-7B-Instruct (16k context), QCCS outperforms all five comparators under LLM-as-judge scoring: for middle-position instructions QCCS reaches 16.7% versus BM25 3.3%, cross-encoder 0.0%, dense 0.0%, and full context 6.7%; overall QCCS reaches 25.3% versus at most 3.6% for retrieval-only comparators. This advantage is not explained by retrieval recall: at k=20, BM25 retrieves the gold evidence sentence in 98.8% of instructions (QCCS 34.9%), yet retrieval arms stay at most 2.6% accurate even when they retrieve it, whereas QCCS reaches 25.0% even when it does not. In this proof-of-concept evaluation, query-aligned context selection predicts EHR instruction-following accuracy better than gold-sentence retrieval recall.

[4] How to Train a Real-World Silicon Concierge? Internalizing Complex Business Workflow to Only OneModel cs.CL | cs.AIPDF

Chang Liu, Chaoyang Ning, Dayi Jiang, Enrui Gu, Fang Ran

TL;DR: 这篇论文提出了OneModel，一种将复杂业务流程内化到单一模型参数中的新范式，以替代传统工业智能体中依赖多个模块化组件的流水线系统。通过持续预训练和逻辑编译监督微调，模型将碎片化的业务规则转化为统一的注意力空间内的直观推理。在实际金融系统中部署后，OneModel在降低延迟和提高智能解决率方面取得了显著效果。

Details

Motivation: 传统工业智能体依赖模块化流水线，导致系统复杂、错误传播和延迟高，难以处理流畅的用户意图。论文旨在解决这些系统在延迟、准确性和复杂性之间的权衡问题。

Result: 在线A/B测试显示，端到端延迟从18.7秒降低到8.0秒（减少超过50%），智能解决率从64.3%提升到83.3%，在真实世界金融服务系统中实现了性能突破。

Insight: 创新点在于从外部工作流转向内化知识表示的范式转变，通过模型参数直接整合复杂业务逻辑和标准操作程序，用统一的注意力机制替代脆弱的工程逻辑，为工业智能体提供了可扩展的蓝图。

Abstract: Traditional industrial agents rely on modular pipelines, including Router, Retriever, Planner, Executor, Responder, Reviewer, and other components. These systems often fracture into a labyrinth of ad-hoc patches, leading to cascading errors and high latency. We propose OneModel, an applicable paradigm shift from external workflows to internalized knowledge representation. Unlike modular systems that slice fluid user intents into static steps, OneModel consolidates complex business logic and SOPs directly into the model parameters. Through Continual Pre-training (CPT) and logic-compilation SFT, we transform fragmented business rules into intuitive model reasoning within a unified attention space. Deployed in our global financial service system, OneModel effectively breaks the trade-off between latency, accuracy, and complexity. Online A/B testing demonstrates an end-to-end latency reduction of more than 50 percent, from 18.7 seconds to 8.0 seconds, while the Intelligent Resolution Rate (IRR) increases from 64.3 percent to 83.3 percent. The results show that OneModel can replace brittle engineering logic with internalized cognitive intuition, offering a scalable blueprint for transitioning industrial agents from complex, error-prone workflows to unified model architectures.

[5] The Divergence Hypothesis: Unmasking Lexical Interference and Label Bias in Mental Health NLP cs.CL | cs.AIPDF

Moustafa Yehia Hassan

TL;DR: 本文提出了一个名为TSS（Triple-Stream Stress probe）的多通道诊断框架，用于揭示和量化计算心理健康（CMH）分类器中因标注来源（人工标注 vs. 远程监督）不同而导致的词汇干扰和标签偏差问题。该框架将文本分解为词汇、形态句法和心理语言学风格三个通道进行分析，并引入了源自计量经济学的差异中的差异统计量DoD进行标签来源审计。

Details

Motivation: 计算心理健康分类器在分布变化下性能下降，因为人工标注和远程监督流程奖励了不同的语言信号，导致模型学习到标注来源特定的捷径特征，而非真正的心理健康信号。

Result: 在四个英文数据集（N=12,906）上，TSS揭示了词汇干扰效应：在风格通道中添加词汇特征会显著降低模型在人工标注数据上的Macro-F1（平均下降0.072，p<10^-4），但在自动标注数据上则不会。DoD统计量（如DoD(BC-A) = 0.0374, p=0.0032）和平台分层分析（如Twitter-only DoD）均证实了该模式。干预性掩蔽实验表明，在破坏内容词后，风格通道仍能保留约95-99%的性能。

Insight: 创新点在于提出了一个可分解文本信号的多通道诊断框架TSS和用于标签来源审计的DoD统计量，能够有效识别和量化标注偏差与词汇干扰，为评估NLP模型泛化能力前的捷径学习提供了审计工具，而非直接用于临床筛查。

Abstract: Computational mental health (CMH) classifiers often degrade under distribution shift because human annotators and distant-supervision pipelines reward different linguistic signals. We introduce TSS (Triple-Stream Stress probe), a multi-channel diagnostic framework that decomposes text into (A) lexical character n-grams, (B) a small, mostly content-free morpho-syntactic channel, and (C) a 154-feature psycholinguistic style channel. Across four English datasets (N=12,906), TSS reveals a lexical interference effect: adding lexical features to the style channel reduces Macro-F1 on human-labeled data (mean drop 0.072, p<10^-4) but not on auto-labeled data. We propose Degree of Divergence (DoD), a difference-in-differences statistic adapted from econometrics for label-source auditing, with instance-level bootstrap inference; the headline estimate is DoD(BC-A) = 0.0374, 95% CI [0.0097, 0.0651], p=0.0032. A platform-stratified Twitter-only DoD (which removes the Reddit vs. Twitter contrast) reproduces the pattern with bootstrap inference: DoD-Tw(BC-A) = +0.096 (p<0.001) and DoD-Tw(AC-A) = -0.089 (p<0.001). Interventional masking (pos_only) retains ~95-99% of Channel C’s performance after destroying content words on human datasets, indicating that the style channel does not rely primarily on lexical surface form. TSS is positioned as a diagnostic audit framework, not a clinical screening tool: it flags label-source-specific shortcut learning before generalization claims are made.

[6] Self-Speculation for Faster Reasoning Models cs.CLPDF

Ravisri Valluri, Tung Nguyen, Aditya Grover

TL;DR: 本文提出了一种名为SSR（Self-Speculation for Reasoning Models）的无训练自推测解码方法，旨在加速大语言模型在涉及规划和多步决策的复杂任务上的推理生成。该方法利用模型自身的思维链（CoT）作为推测源，将部分CoT的答案分布作为草稿模型，完整CoT分布作为验证模型，通过接受长草稿前缀和引入后缀解码来利用草稿与最终响应之间的语义和词汇重叠，从而显著降低生成延迟。

Details

Motivation: 大语言模型在处理需要长推理链的复杂任务时，生成延迟高，这严重影响了在延迟敏感和交互式应用（如语音助手、编码代理）中的用户体验。现有加速方法通常关注于令牌级生成，未能有效利用推理工作流的结构特性。

Result: 在多个结构化和长文本生成任务上对SSR进行评估，结果表明，对于Qwen3.5和Gemma-4等流行的开源模型，SSR在总生成延迟上实现了高达24.1%的相对提升。

Insight: 核心创新点在于提出了一种无训练的自推测解码框架，其利用同一模型在不同推理预算（部分CoT vs. 完整CoT）下的输出分布作为草稿和验证器，并观察到部分CoT响应与完整响应之间存在显著的语义和词汇重叠。在此基础上，通过接受长草稿前缀和引入后缀解码（利用草稿种子化后缀缓存），进一步利用非连续的重叠部分，从而在保持推理质量的同时实现大幅加速。这是一种巧妙利用模型自身推理过程内部结构进行加速的方法。

Abstract: Large language models (LLMs) are deployed for increasingly complex tasks involving planning and multi-step decision making, but high-quality performance on these tasks often requires generating long reasoning traces. This is a poor fit for latency-sensitive and interactive applications like voice assistants or coding agents, where generation latency can strongly affect user experience. Existing acceleration methods typically focus on token-level generation, without utilizing the structure of reasoning workflows. We introduce SSR: Self-Speculation for Reasoning Models, a training-free self-speculative decoding method that leverages the chain-of-thought (CoT) as a source of speculation. SSR uses the partial-CoT answer distribution as the drafter and the full-CoT distribution as the verifier, deriving both from the same model at different reasoning budgets. This builds on the observation that later partial-CoT responses often exhibit greater semantic and lexical overlap with the full-budget response. Due to this overlap, SSR can accept long draft prefixes at once, leading to large speedups on structured and long-form generation tasks. To further exploit draft-response overlap beyond the contiguous prefix accepted by standard speculative decoding, SSR also incorporates suffix decoding, using the draft to seed a suffix cache and recover useful spans beyond the accepted prefix, further reducing latency on tasks with high lexical overlap between the draft and the final response. We evaluate SSR on multiple structured and long-form generation tasks where it is most useful, and demonstrate a relative improvement of up to 24.1% on total generation latency for popular open-source models such as Qwen3.5 and Gemma-4.

[7] Toward Auto-Research: Mining Falsifiable Research Ideas from Paper Knowledge Graphs with Categorical Structure cs.CL | cs.AIPDF

Yuchen Wang, Zhongzhi Luan

TL;DR: 本文提出了一种基于范畴论的自动化研究想法生成方法，通过将论文建模为包含类型化研究实体和关系的小范畴，并利用部分函子候选来构建跨论文的桥梁，从而克服了现有LLM方法将论文视为扁平对象的局限性。该方法采用三层算法（范畴签名聚类、函子保持门和六轴LLM合理性判断）进行实现，并在数万篇全文解析的论文语料上进行了评估。

Details

Motivation: 现有基于大语言模型的自动化研究想法生成系统存在结构性弱点，它们将构思简化为自由文本重组、随机论文配对或嵌入相似性检索，忽略了论文中类型化的问题-方法-指标-主张关系结构，而这些正是研究人员进行跨领域类比推理时实际使用的要素。

Result: 在四种消融条件下对包含数万篇全文解析论文的语料库进行评估，范畴门以大约17:1的比例过滤跨领域候选想法，同时被接受想法的定量可证伪率始终保持在83%以上；每个被拒绝的候选想法都保留了其每轴理由，因此该门控机制也充当了日志层而非静默过滤器。

Insight: 创新点在于将范畴论（特别是组合与恒等箭头的概念）引入研究想法生成，将论文建模为小范畴以捕获类型化实体和关系，并设计了一个包含函子保持门的三层算法来评估跨领域类比的结构一致性，这为自动化研究提供了更严谨的、基于结构推理的框架。

Abstract: Automated research-idea generation systems built on large language models (LLMs) share a structural weakness: they reduce ideation to free-text recombination, random paper pairing, or embedding-similarity retrieval. The three approaches fail in the same way: each treats a paper as a flat object, a string or a vector, and so quotients away the typed problem-method-metric-claim arrows a researcher actually uses when reasoning about a cross-domain analogy. We recover the missing structure with the minimal piece of category theory that a typed graph alone does not provide: composition, together with identity arrows, which makes it possible to ask whether a proposed analogy preserves relation chains. Concretely, each paper $p$ is modelled as a small category $C_p$ whose objects are extracted typed research entities and whose morphisms are the relations the paper asserts; a cross-paper bridge from $p$ to $q$ is then a partial functor candidate $F: C_p -> C_q$ that preserves object kinds and covered relation classes. We instantiate the model as a three-layer algorithm: categorical signature clustering, a functor-preservation gate, and a six-axis LLM plausibility judge. Evaluated on a corpus of tens of thousands of full-text-parsed papers under four ablation conditions, the categorical gate filters cross-domain candidates at roughly a 17:1 ratio while the quantitative-falsifier rate of accepted ideas stays above 83% throughout; every rejected candidate is retained with its per-axis rationale, so the gate doubles as a logging layer rather than a silent filter.

[8] Multilingual Verifier Bias in RLVR: Benchmark, Rollout Diagnosis, and the Cross-Lingual Selection Bottleneck cs.CL | cs.LGPDF

Chenyu Zhou, Qiliang Jiang, Xu Zhou

TL;DR: 本文研究了在多语言环境中，强化学习与可验证奖励（RLVR）方法中精确匹配验证器存在的语言偏见问题。作者提出了一套可复用的审计协议，包括验证器鲁棒性测试、策略诊断和语言条件奖励误差度量，并在MGSM和MATH-500数据集上进行了实验。研究发现，精确匹配验证器在不同语言（如日语、英语、中文）上会产生显著不同的假阴性奖励噪声，并揭示了跨语言选择瓶颈的存在。

Details

Motivation: 动机在于揭示并解决RLVR方法在多语言数学推理任务中的一个关键假设失效问题：即精确匹配验证器作为语言中立的奖励函数，在实际中会因格式和脚本的变体而产生依赖于语言的假阴性奖励噪声，从而影响模型训练效果。

Result: 在MGSM数据集上，使用k=8采样时，精确匹配验证器在Qwen3-4B、Qwen3-8B和Llama-3.1-8B-Instruct模型上对不同语言（日语、英语、中文）的可信正确答案的拒绝率存在显著差异，例如Qwen3-8B在日语上的假阴性率高达0.642，而英语和中文分别为0.122和0.073。在MGSM250数据集上，提出的目标局部聚合规则在不使用可信标签的情况下，能缩小55-78%的平均选择差距。

Insight: 创新点在于提出了一套系统性的多语言RLVR奖励审计协议，并定位了奖励误差的主要机制在于最终答案的接口模型。核心洞察是揭示了跨语言选择瓶颈的存在，并表明多语言RLVR奖励在优化前必须按语言和答案接口进行审计，这对多语言模型训练具有重要操作指导意义。

Abstract: Reinforcement learning with verifiable rewards (RLVR) is a standard recipe for training large language models on mathematical reasoning, where an answer verifier serves as a language-neutral reward function. We show that this assumption fails in multilingual settings: an exact-match verifier turns format and script variation into language-dependent false-negative reward noise. We introduce a reusable protocol for auditing multilingual RLVR rewards: a verifier-robustness suite, a rollout-diagnosis procedure, and language-conditioned reward-error metrics for Japanese, English, and Chinese answers. On MGSM rollouts with k=8, the exact-match proxy rejects trusted-correct answers at sharply different rates by language across Qwen3-4B, Qwen3-8B, and Llama-3.1-8B-Instruct; for Qwen3-8B, the false-negative rate reaches 0.642 on JP against 0.122 on EN and 0.073 on CN. A plain-numeric probe localizes the mechanism to the final-answer interface: an interface model drives reward-error VLB to zero while the residual accuracy gap is unchanged. We then expose a cross-lingual selection bottleneck: on MGSM250 rollouts, a target-local aggregation rule using no trusted labels closes 55-78% of the average selection gap, and over 95% of repairs require genuine cross-lingual support. The bottleneck replicates on a 483-problem MATH-500 set. A controlled training audit shows that rule-GRPO raises trusted accuracy while the reward-error VLB stays high. The unifying message is operational: multilingual RLVR rewards should be audited by language and by answer interface before they are optimized.

[9] Hadith computational science in the age of large language models: a critical narrative review cs.CL | cs.AIPDF

Md. Ashraful Haque, Riasat Islam

TL;DR: 本文对圣训计算科学在大型语言模型时代的发展进行了批判性叙事综述，分析了该领域如何被Transformer模型、检索增强流程和LLMs重塑。研究发现进展不均衡：数据资源扩展、分割任务成熟、叙述者和来源验证问题得到更好形式化，LLM辅助工作流支持语料库规模增强、多语言访问和基于事实的评估。但进展仍受限于狭窄语料库、基准可比性弱、合成到真实迁移差距、叙述者身份解析、预处理脆弱性、有限可重复性和稀疏专家验证等问题。

Details

Motivation: 现有文献综述记录了圣训计算科学领域的增长，但缺乏对哪些进展方法稳健、哪些仍受限于基准、哪些未解决问题限制学术使用的批判性分析。本文旨在填补这一空白。

Result: 研究发现重要差距存在于主流基准之外：非经典和晦涩语料库、注释和解释文献、与古兰经和先知传记的跨源链接，以及面向教法学证据支持的问题。

Insight: 论文主张圣训计算不应仅作为孤立模型性能评估，而应视为需要知识整合、来源追溯和专家监督的证据基础设施问题。在此基础上，提出了使该领域方法更强、对伊斯兰学术更有用的研究议程。

Abstract: We examine how hadith computational science is being reshaped by transformer models, retrieval-grounded pipelines, and large language models (LLMs). Recent reviews document growth in the literature, but they do not yet provide a critical account of which advances are methodologically robust, which remain benchmark-bound, and which unresolved problems still limit scholarly use. We address this gap through a critical narrative review that combines critique of existing reviews, paper-level appraisal of representative original studies, and synthesis of Islamic scholar and domain-expert perspectives on authenticity, authority, and responsible use. We find uneven progress. Data resources have expanded, segmentation tasks have matured, narrator and source-verification problems are better formalized, and LLM-assisted workflows now support corpus-scale enrichment, multilingual access, and grounded evaluation. At the same time, progress remains constrained by narrow corpora, weak benchmark comparability, synthetic-to-real transfer gaps, narrator identity resolution, preprocessing fragility, limited reproducibility, and sparse expert-grounded validation. We show that important gaps lie beyond dominant benchmarks: non-canonical and obscure corpora, commentary and explanatory literature, cross-source links with Qur’an and seerah, and fiqh-facing evidence support. We argue that hadith computation should be assessed less as isolated model performance than as an evidence infrastructure problem requiring knowledge integration, provenance, and expert supervision. On this basis, we define a research agenda for making the field methodologically stronger and more useful to Islamic scholarship.

[10] Trilingual Topic Modeling of Sri Lankan Parliamentary Debates cs.CL | cs.AIPDF

Himath Dhanapala, Haren Daishika, Himandhi Kuruppu, Sithija Seneviratne, Ashini Kavindya

TL;DR: 该论文提出了一种端到端框架，用于处理斯里兰卡议会辩论（汉萨德）这一包含僧伽罗语、泰米尔语和英语的三语语料库。该框架利用基于大语言模型（LLM）的文本提取，结合多语言嵌入和基于密度的聚类流程进行主题建模，并探索了一种混合语义-词汇扩展方法BiTopic以提升可解释性。

Details

Motivation: 斯里兰卡议会辩论语料库包含多语言、代码混合内容，且受复杂PDF格式和黏着语形态影响，无法被标准自然语言处理（NLP）流程处理。论文旨在克服这些挑战，实现跨语言的、无监督的主题建模。

Result: 该方法应用于2017-2026年间的19,553篇演讲，成功识别出30个宏观主题，其聚类纯度（BCP）达到0.673。主题的时间轨迹与2019年复活节袭击、2022年经济危机等重大国家事件无监督对齐。相比之下，传统LDA方法因跨语言碎片化问题在此语料上失效。

Insight: 创新点在于结合LLM进行复杂PDF文本提取，并采用多语言嵌入与密度聚类进行跨语言主题建模。提出的BiTopic混合扩展方法提升了主题可解释性并有效处理了噪声数据，为处理多语言、代码混合及形态复杂的语料提供了可行的无监督解决方案。

Abstract: Sri Lankan parliamentary debates (Hansards) constitute a trilingual corpus of speeches in Sinhala, Tamil, and English, including code-mixed content, yet remain inaccessible to standard NLP pipelines due to layout-complex PDFs, multilingual scripts, and agglutinative morphology. We present an end-to-end framework that addresses these challenges through LLM-based text extraction followed by a multilingual embedding and density-based clustering pipeline for topic modeling. A hybrid semantic-lexical extension, BiTopic, is further explored to improve interpretability and recover speeches otherwise discarded as noise. Applied to 19,553 speeches spanning 2017-2026, the pipeline recovers 30 macro-topics achieving a cluster purity (BCP) of 0.673, whose temporal trajectories align unsupervised with major national events including the 2019 Easter Sunday attacks and the 2022 economic crisis. Traditional LDA fails on this corpus due to cross-lingual fragmentation, whereas the proposed approach successfully identifies thematic structure across all three languages without supervision.

[11] An ambiguity taxonomy for evaluating large language model performance on clinical registry abstraction: a multi-site prospective study cs.CLPDF

James Matheson, Betsy Castillo, Andrew Y. Shin, David Scheinker

TL;DR: 本研究评估了大型语言模型（LLM）在未处理的电子病历（EMR）数据上执行临床注册表抽象任务的性能。通过在美国心脏病学会国家心血管数据注册表（ACC NCDR）上的多中心前瞻性研究，发现LLM的准确率远低于人工抽象员，且随着问题模糊性和所需临床推理水平的增加而显著下降。

Details

Motivation: 评估LLM直接从原始电子病历数据中提取信息以完成临床注册表抽象任务的可行性，解决自动化临床数据抽象中因数据模糊性和复杂临床推理带来的挑战。

Result: 在验证研究中，LLM答案与共识答案完全匹配的比例为87%，部分匹配为2%，不匹配为9%；平均问题级准确率为91.5%，但随模糊性类别从药物/事件标志（96%）到事件时间（62%）而递减。

Insight: 论文创新性地提出了一个基于模糊性和临床推理需求的六类分类法（如药物/事件标志、临床解释、事件时间等），用于系统评估LLM在复杂临床任务中的性能退化；客观分析表明，这种分类框架有助于揭示LLM在处理需要深层临床推理的模糊信息时的局限性，为未来改进提供了明确方向。

Abstract: Objective: To evaluate large language model (LLM) performance on unprocessed electronic medical record (EMR) data for clinical registry abstraction. Methods: We evaluated LLM performance answering registry questions for the American College of Cardiology National Cardiovascular Data Registry (ACC NCDR). In a pilot study at an academic medical center, the model identified candidate data sources for each registry question and experienced abstractors used these results to define question-specific document sets. In a validation study at a second center with a second ACC NCDR registry, the LLM answered questions using the question-specific document sets. Before reviewing any output, two abstractors independently established the ground truth and assigned each question to one of six categories, ordered by the ambiguity and clinical reasoning required to resolve it: Medication/Event Flag, Binary Clinical Presence, Administrative, Quantitative Laboratory/Physiologic, Clinical Interpretation, and Event Timing. Results: The analytical sample comprised 9,430 abstractor answers reconciled to 4,715 consensus answers (501 pilot; 4,214 validation). In the pilot, candidate data sources per question averaged between 14.6 (SD 13.9) for demographics and 89.2 (SD 56.1) for history and risk factors. In validation, human inter-rater agreement was approximately 98% while 87% of LLM answers exactly matched consensus, 2% partially, and 9% did not. Mean question-level accuracy was 91.5% (SD 13.4%) across 157 questions with at least 20 answers, and declined as ambiguity increased, from 96% for Medication/Event Flag to 62% for Event Timing questions. Conclusions: LLMs answering clinical registry questions on unprocessed EMR data achieved far lower accuracy than human abstractors. LLM accuracy fell steadily as ambiguity and the level of required clinical reasoning increased.

[12] Decoupled Vision-Language System for Multimodal Understanding and Generation cs.CL | cs.CVPDF

Yifan Xu, Baochen Xiong, Xiaoshan Yang, Donglin Di, Yaowei Wang

TL;DR: 本文提出了一种名为Libra的新型多模态大语言模型架构，该架构通过解耦视觉与语言系统，并设计动态路由的交叉模态桥接模块，实现了统一的多模态理解与生成能力。

Details

Motivation: 为了解决现有MLLMs在模态建模与跨模态交互耦合导致的学习效率与性能瓶颈，旨在设计一种能同时支持高质量多模态理解与生成的统一架构。

Result: 实验表明，Libra在仅理解任务（Libra-1）和统一理解与生成任务（Libra-2）的多个基准测试中均取得了强劲性能，实现了理解与生成任务的相互促进。

Insight: 核心创新在于通过解耦的自模态建模与动态路由的跨模态交互模块（如Switch Attention/FFN），使各模态能学习独特表征，同时保持高效的跨模态理解，其架构设计及在分词、位置编码和监督信号上的改进具有借鉴意义。

Abstract: We introduce a new architecture design for multimodal large language models (MLLMs), Libra, capable of both multimodal understanding and generation. Libra architecture contains one vision system and one language system, connected by cross-modal bridges. This design decouples self-modal modeling and cross-modal interaction, enabling each modality to learn its unique representations while maintaining effective cross-modal comprehension. The decoupling is mainly achieved in a switch attention module and a switch FFN module, which dynamically routes the computation flow for self-modal modeling and cross-modal interaction scenarios. We evaluate the effectiveness in two important settings: \textbf{Libra-1} for the understanding-only image-to-text setting, and \textbf{Libra-2} for unified image-to-text understanding and text-to-image generation. In addition to the architecture design, we discuss various improvements on tokenization, positional encoding, and supervision. Experiments demonstrate that the dedicated Libra design enables mutual improvements on multimodal understanding and generation, achieving strong performance on both understanding and generation benchmarks.

[13] Poly-InstructTTS: Learning In-the-Wild Expressive Speech Synthesis from Open-Ended Instructions cs.CL | cs.AIPDF

Junhui Zhang, Qianhui Xu, Qingxiang Guo, Dawei Yang, Ling Miao

TL;DR: Poly-InstructTTS是一个从开放式自然语言指令中学习生成富有表现力语音的文本转语音模型。它利用野外视听数据构建了一个大规模指令标注语料库，并采用无提示GPT结合基于属性的思维令牌和流匹配模块来生成语音，同时支持针对特定说话人的微调。

Details

Motivation: 现有TTS模型虽然自然度高，但难以通过自然语言指令精细控制语音的表达方式，本文旨在解决这一挑战。

Result: 实验表明，Poly-InstructTTS在指令遵循和表达力方面表现出色，并在扩展的InstructTTSEval基准测试中进行了评估。

Insight: 创新点在于构建了大规模多模态指令标注数据集，并设计了结合属性思维令牌的无提示GPT架构与流匹配模块，实现了从开放指令到细粒度语音风格的控制与迁移。

Abstract: While recent text-to-speech (TTS) models achieve high naturalness, controlling fine-grained expression via natural-language instructions remains challenging. We introduce Poly- InstructTTS, which learns expressive speech from open-ended instructions using in-the-wild audiovisual data. We build a scalable multi-modal pipeline to construct a 1,000-hour instruction-annotated corpus covering 1,000+ fine-grained emotions and styles. The framework uses a prompt-free GPT with attribute-based thinking tokens, followed by a flow-matching module that injects timbre from a reference audio. We also present a speaker fine-tuning procedure to transfer instruction control to specific speakers while preserving persona. We further extend InstructTTSEval with broader tasks. Experiments show that Poly-InstructTTS delivers strong performance in instruction adherence and expressiveness. Audio demos and the expanded testset are available on our project page.

[14] Ansari: A Retrieval-Grounded Islamic AI Assistant – Architecture, Deployment, and Lessons from 140,000 Conversations cs.CL | cs.AI | cs.CYPDF

M Waleed Kadous, Amr Elsayed, Abdullah Al Nahas, Ashraf Haress

TL;DR: 本文介绍了Ansari，一个基于检索的伊斯兰AI助手，通过代理检索循环从经过认证的伊斯兰语料库（如《古兰经》、圣训集、法学百科全书和注释来源）中检索信息来回答宗教问题，以避免事实捏造和价值偏差。该系统自2023年6月以来已处理超过14万次跨25种语言的对话，并部署于多个平台。论文还通过多项评估展示了其性能，并总结了在信仰或价值观敏感领域部署LLM的通用经验教训。

Details

Motivation: 通用大语言模型在回答宗教问题时存在两个严重风险：事实捏造（如虚构《古兰经》经文或圣训）和微妙的价值偏差，因此需要开发一个能够提供可靠、基于权威来源的伊斯兰内容助手。

Result: Ansari在多项评估中表现优异：在零样本的机构考试中取得良好成绩，在斋月期间通过人工验证，并在公开的IslamicMMLU排行榜上领先于前沿模型，同时在伊斯兰法律推理基准（IslamicLegalBench）上具有竞争力，并能有效抵抗错误前提。

Insight: 创新点在于采用代理检索循环架构，将LLM与经过认证的伊斯兰语料库严格结合，确保回答基于检索内容并附带引用；客观分析认为，该系统强调了在价值观敏感领域，检索增强虽必要但不足，系统提示既是技术工具也是神学产物，且模型缺乏社区参与仍是关键挑战。

Abstract: General-purpose large language models (LLMs) are increasingly used to answer religious questions, but for Islamic content they carry two serious risks: factual fabrication (inventing Qur’anic verses or hadith) and subtle value misalignment. We present Ansari, a deployed, retrieval-grounded Islamic AI assistant that has handled more than 140,000 conversations across 25+ languages since June 2023. Ansari is built around an agentic retrieval loop: a tool-using language model issues searches against authenticated Islamic corpora – the Qur’an, hadith collections, a multi-volume jurisprudence (fiqh) encyclopedia, and exegetical (tafsir) sources – and answers only on the basis of what it retrieves, with citations attached for verification. We describe the system’s architecture (the agent loop, the retrieval tools, the corpora, and the system prompt that encodes editorial and theological policy), its multi-platform deployment (web, mobile, WhatsApp, and as a Model Context Protocol server and an Agent Skill), and what 140,000 real conversations reveal about how Muslims actually use such a tool. We report results on several complementary evaluations – zero-shot performance on accredited institutional exams, a human-rated validation during Ramadan, and two independent, externally run benchmarks on which Ansari currently tops the public IslamicMMLU leaderboard ahead of frontier models and is competitive on Islamic legal reasoning (IslamicLegalBench) while strongly resisting false premises – and draw out lessons that generalize beyond Islam to any faith- or values-sensitive deployment of LLMs: grounding is necessary but not sufficient, the system prompt is a theological as much as a technical artifact, and the absence of community in how models are formed remains a hard gap.

[15] ImmigrationReason: A Structured Dataset of U.S. Immigration Appeals for Legal Reasoning Research cs.CLPDF

Amirhossein Afsharrad, Seyed Shahabeddin Mousavi

TL;DR: 本文介绍了ImmigrationReason数据集，这是一个基于美国公民及移民服务局行政上诉办公室12,375份非先例决定构建的大规模结构化数据集，用于法律推理研究。该数据集包含法律框架、证据充分性分类、裁决者批评引文、引用和最终处置等信息，并通过多模态流程和专家验证确保质量。

Details

Motivation: 现有法律NLP资源主要集中于联邦判例法并进行粗粒度分类，而忽略了绝大多数政府决策发生的行政裁决领域，因此需要构建专门针对行政上诉的结构化数据集以填补这一空白。

Result: 数据集提取质量通过结合两种独立模态并由Opus 4.7进行对比提示裁决的三阶段流程进行验证，并在500条记录的样本上由领域专家核实，记录了近9,000条AAO识别的法律错误实例。

Insight: 创新点在于构建了首个专注于美国移民行政上诉的大规模结构化数据集，并设计了结合多模态和大型语言模型裁决的质量验证流程，为结果预测、裁决错误分析和高风险监管领域的智能体设计等研究提供了新资源。

Abstract: Most legal NLP resources draw from federal case law and focus on coarse classification, leaving administrative adjudication, where the vast majority of government decisions occur, essentially unaddressed. We introduce ImmigrationReason, a large-scale structured dataset derived from 12,375 non-precedent decisions of the U.S. Citizenship and Immigration Services (USCIS) Administrative Appeals Office (AAO) spanning 2005 to 2026. Each record captures the applicable legal framework, per-criterion evidence-sufficiency findings under a five-category label, verbatim adjudicator-criticism quotes, all citations, and final dispositions, alongside high-quality Claude-transcribed source text. Extraction quality is validated through a three-pass pipeline combining two independent modalities with comparison-prompt adjudication by Opus 4.7, and verified by domain experts on a 500-record sample. The dataset documents nearly 9,000 verbatim instances of AAO-identified legal errors, spans a natural legal-regime transition (the 2016 Dhanasar rule change), and covers 21 years of adjudication. We analyze the dataset in detail and outline research directions it enables, from outcome prediction and adjudicator-error analysis to agent design for high-stakes regulatory domains.

[16] Evaluation-as-Search: Adaptive Discovery of Grounding Failures in Meeting Assistants cs.CL | cs.AIPDF

Sami Khairy, Yasaman Hosseinkashi, Vishak Gopal, Ross Cutler

TL;DR: 本文提出了一种名为‘评估即搜索’的自适应方法，用于系统性地评估LLM驱动的会议助手的‘接地’保真度。该方法将质量评估构建为对自然问题空间的适应性搜索，通过迭代学习评估者反馈来集中探测最可能失败的认知需求，从而构建了一个包含3000多个标注问答对的基准MeetingProbe。实验表明，自适应搜索发现的失败率是随机探测的2.5倍，并识别出主要由语用-话语挑战主导的八类常见失败模式。

Details

Motivation: 当前LLM会议助手已大规模部署，但对其‘接地’保真度的系统评估仍局限于静态基准，这些基准无法捕捉与特定话语结构或推理需求相关的失败模式。

Result: 在消融实验中，自适应搜索的失败发现率（7.1%）是随机探测（2.9%）的2.5倍，其中策略规划器贡献最大。在三个模型上观察到了清晰的能力梯度，并识别出八类主要由语用-话语挑战而非事实回忆错误主导的重复失败类别。MeetingProbe基准在多个模型系列和提供商上得到验证，发现了一个清晰的能力梯度和一个所有模型都无法处理的通用失败子集。

Insight: 创新点在于将质量评估重新定义为一种基于反馈的自适应搜索过程，利用UCB评分覆盖图和盲多维质量评估来引导探测。这提供了一种更高效、更系统地发现模型在复杂、动态对话场景中‘接地’失败的方法，超越了静态基准的局限性。

Abstract: LLM-powered meeting assistants are deployed at scale, yet systematic evaluation of their grounding fidelity remains limited to static benchmarks that miss failure modes tied to specific discourse structures or reasoning demands. We propose Evaluation-as-Search (EaS), a feedback-driven methodology that frames quality evaluation as an adaptive search over the space of natural questions a meeting participant might ask. Rather than sampling uniformly, EaS learns from evaluator feedback across iterations to concentrate probing effort on cognitive demands where failures are most likely, guided by a UCB-scored coverage map and blind multi-dimensional quality evaluation. Using EaS, we construct MeetingProbe, a benchmark of over $3{,}000$ annotated question–answer pairs spanning 20 transcripts from three meeting genres and three LLM assistants. In ablations, adaptive search surfaces $2.5\times$ more failures than random probing ($7.1%$ vs. $2.9%$ finding rate), with the strategic planner contributing the largest individual effect. Across three models, we observe a clear capability gradient and identify eight recurring failure categories dominated by discourse-pragmatic challenges rather than factual recall errors. We further validate MeetingProbe across multiple model families and providers, finding a clean capability gradient and a curated subset of universal failures that no model handles. MeetingProbe is released publicly to support reproducible evaluation of meeting assistant grounding fidelity.

[17] LingShu: A Large-Scale Symptom-Centric Contextualized Knowledge Graph Bridging Traditional Chinese Medicine and Modern Biomedicine cs.CL | cs.AIPDF

Rui Hua, Zixin Shu, Kai Chang, Dengying Yan, Jianan Xia

TL;DR: 本文介绍了LingShu，一个大规模、以症状为中心的情境化知识图谱，旨在连接传统中医与现代生物医学。它整合了多源数据，包含超过1700万个原子级实体和3900多万条关系记录，并采用混合数据模型（三元组与情境化四元组）来编码条件性医学知识。

Details

Motivation: 传统二元关系知识图谱难以表示生物医学知识的条件性本质，而症状为连接依赖症状模式进行辨证论治的传统中医，以及将临床表现与疾病和分子机制关联的现代生物医学，提供了一个共享的表型层。

Result: 导出的LingShu版本包含1733万个原子级实体记录和3947万条关系记录（包括1719万条语义三元组和2229万条情境化四元组），并开发了集成了图谱可视化、基于图谱的推理和基于证据的知识问答代理的Web平台。

Insight: 核心创新在于其混合数据模型：既维护了64种类型化的三元组关系模式以确保广泛的连通性，又引入了35种情境化四元组关系模式来显式编码条件性医学关联（如证候依赖的草药功效、疾病情境化的药物效果），从而提供了对医学关系相关情境的细粒度表示。

Abstract: Biomedical knowledge graphs (KGs) are pivotal for knowledge organization, yet traditional binary relations often struggle to represent the conditional nature of biomedical knowledge. Symptoms provide a shared phenotypic layer for linking Traditional Chinese Medicine (TCM), which relies on symptom patterns for syndrome differentiation and treatment selection, with modern biomedicine, which connects clinical manifestations to diseases and molecular mechanisms. We present LingShu, a large-scale symptom-centric contextualized knowledge graph designed to bridge TCM and modern biomedicine. The exported version of LingShu analyzed in this study comprises 17.33 million atom-level entity records and 39.47 million relation records, including 17.19 million semantic triples and 22.29 million contextualized quadruples. LingShu integrates multi-source data, including clinical electronic medical records, authoritative TCM texts, biomedical ontologies, and curated knowledge bases, through a pipeline combining natural language processing, terminology normalization, and human-in-the-loop verification. A key innovation of LingShu is its hybrid data model: it maintains 64 typed triple relation patterns to ensure broad connectivity, while incorporating 35 contextual quadruple relation patterns to capture conditional medical associations. This dual-structure approach explicitly encodes conditional knowledge, providing a granular representation of the contexts associated with medical relations. These contextualized relations cover syndrome-dependent herb efficacy, disease-contextualized drug effects, population-specific clinical associations, and mechanism-related therapeutic responses. Furthermore, we developed a web platform (http://www.tcmkg.com/) that integrates graph visualization, graph-based reasoning, and an evidence-grounded knowledge question-answering agent.

[18] When Failures Propagate: Causal Failure Attribution in Agentic Retrieval-Augmented Generation cs.CL | cs.AIPDF

Lauren Pothuru

TL;DR: 本文提出了AgenticRAG-FP，一个用于因果故障归因的干预性基准测试，旨在解决智能体化检索增强生成（RAG）中因早期检索错误传播导致最终答案错误的问题。该基准通过在指定步骤注入认证故障并重新执行下游轨迹，评估事后诊断方法能否准确识别故障源头。

Details

Motivation: 在智能体化RAG中，检索、推理和生成在多步骤中交错进行，早期步骤的检索错误可能仅在后续步骤才表现为错误答案，而后续检索也可能修复轨迹，这使得故障归因变得困难。

Result: 在80个三跳MuSiQue问题上的严格密集Claude Haiku 4.5扫描中，基于覆盖率的诊断在第一步的准确率为0.91，但在第二、三步为0.00。在深度为2的内容破坏研究中，基于覆盖率的诊断准确率为0.00，而冻结步骤反事实探针的准确率为0.67。

Insight: 论文的创新点在于提出了一个因果干预基准来显式评估故障传播深度对诊断的影响，并揭示了事后信号丢失的普遍证据，强调了在智能体化RAG中区分故障源头与后续传播的重要性。

Abstract: Agentic retrieval-augmented generation (RAG) interleaves retrieval, reasoning, and answer generation across multiple hops. A retrieval error at hop 1 can surface only as a wrong answer at hop 3, while later retrieval can also repair the trajectory. This paper introduces AgenticRAG-FP, an interventional benchmark for causal failure attribution in agentic RAG. The benchmark injects a certified fault at a specified hop, re-executes the downstream trajectory, and evaluates diagnosers against the known intervention. Its central question is whether a post-hoc trace still identifies the injected hop after the suffix changes. In the completed strict dense Claude Haiku 4.5 sweep on 80 three-hop MuSiQue questions, coverage-based diagnosis is 0.91 at hop 1 and 0.00 at hops 2 and 3 (n=43,36,21 failed trajectories). A smaller content-corruption study changes an answer-bearing or bridge fact in topically intact evidence. At depth 2, where 18 failed cases remain after filtering, coverage-based diagnosis is 0.00 and a frozen-hop counterfactual probe is 0.67 in an exploratory pooled comparison. Depth-3 content estimates are descriptive only because they contain three failed cases. These results make propagation depth an explicit evaluation axis for diagnosing agentic RAG failures while distinguishing broad evidence of post-hoc signal loss from small-sample method comparisons.

[19] AgentMercury: Your Agent Can Synthesize Verifiable Environments for Business Scenarios at scale cs.CL | cs.AIPDF

Minbyul Jeong, Chanwoong Yoon

TL;DR: 本文提出了AgentMercury，一个可扩展的框架，用于从高层次业务场景中合成可执行的环境，以支持智能体学习。该框架首先实例化一个包含实体、服务、工具、状态和可执行跨服务不变量的持久世界，从中可以自然涌现出多样化的任务和交互轨迹。作者构建了涵盖14个行业和50个国家的4,783个可执行环境，并将其用作强化学习的训练基底，实验表明，在这些面向业务的环境中训练的策略在多个领域基准测试中取得了显著提升。

Details

Motivation: 当前智能体通过与环境的交互进行学习，但训练环境通常是围绕预定义任务和基准手动构建或合成的，这种以任务为中心的模式难以扩展，无法反映现实且不断演进的、能自然涌现多样化任务的工作流。

Result: 在AgentMercury生成的环境中训练的策略，在EnterpriseOps-GYM基准上，Qwen3.5-4B模型的得分从12.3提升至15.7；在AIME26基准上，得分从45.9提升至56.0。此外，通过微调Qwen3.5-35B-A3B模型学习环境构建过程，其在保留业务场景上的可执行世界创作成功率从3.3%大幅提升至83.3%。

Insight: 创新点在于从高层次业务场景出发，先构建一个持久、可执行的底层世界，而非针对特定任务构建环境，从而支持多样化任务的涌现。这为智能体提供了超越特定基准、更具通用性的学习信号，并且环境构建过程本身也可以通过学习来掌握，实现了从场景到可训练环境的自动化、规模化生成。

Abstract: Agents learn to act through interaction with environments, yet the environments used for training are often manually constructed or synthesized around predefined tasks and benchmarks. This task-centric paradigm makes it difficult to scale environments that reflect realistic and evolving workflows where diverse tasks can naturally emerge from the underlying world. We introduce AgentMercury, a scalable framework for synthesizing executable environments from high-level business scenarios. Rather than constructing an environment for a specific task, AgentMercury first instantiates a persistent world with entities, services, tools, state, and executable cross-service invariants, from which diverse tasks and interaction trajectories can subsequently emerge. We construct 4,783 executable environments spanning 14 industries and 50 countries, and use them as training substrates for reinforcement learning. Despite being generated without targeting the evaluation benchmarks, policies trained on these business-oriented environments improve substantially on both enterprise workflows and out-of-domain benchmarks spanning reasoning, coding, scientific computing, and tool use. In our experiments, Qwen3.5-4B improves from 12.3 to 15.7 on EnterpriseOps-GYM and from 45.9 to 56.0 on AIME26 after training on AgentMercury environments. We further show that the construction process itself can be learned: fine-tuning Qwen3.5-35B-A3B on construction traces increases executable-world authoring success from 3.3% to 83.3% on held-out business scenarios. These results show that scenario-grounded environments can provide useful and generalizable learning signals beyond benchmark-specific training, while their construction can itself become a learnable capability.

[20] STAR-OPD: Structured Aspect-Cascade-Aware On-Policy Reward Distillation for ABSA Quadruple Extraction cs.CL | cs.AIPDF

Tong Sun, Mingyang Ma, Jiayang Yu

TL;DR: 本文提出了一种名为STAR-OPD的结构化方面级联感知在线策略奖励蒸馏方法，用于解决基于方面的情感分析四元组提取任务。该方法针对传统离线策略蒸馏在处理目标-方面接口错误时导致结构无效状态的问题，通过在学生模型生成的轨迹上进行训练，并应用针对绑定一致性、目标基础和细粒度方面消歧的集合结构奖励，显著提升了蒸馏后小模型的性能。

Details

Motivation: 动机在于解决将大型思维链模型在ABSA四元组提取任务上的能力蒸馏到更小、可部署模型时的困难。作者发现，传统离线策略蒸馏存在特定于任务的失败模式：学生在目标-方面接口处的错误会产生结构无效状态，进而破坏下游预测。

Result: 在E-ABSA20K和SemEval-2014基准上的实验表明，STAR-OPD持续优于离线策略和通用在线策略基线，减少了目标幻觉，并在结构困难的案例上显著提升了性能。使用Qwen3-4B模型时，该方法大幅缩小了师生模型之间的性能差距，同时提高了推理效率。

Insight: 论文的核心创新点在于将通用的在线策略蒸馏方法实例化用于ABSA四元组提取，并设计了级联感知的集合结构奖励，直接针对绑定一致性、目标基础和细粒度方面消歧进行优化。这强调了在线策略结构校正对于蒸馏ABSA提取任务的重要性，为解决结构化预测任务中的蒸馏问题提供了新思路。

Abstract: Aspect-based sentiment analysis (ABSA) quadruple extraction requires jointly predicting target, aspect, opinion, and sentiment over reviews that often contain multiple fine-grained sentiment tuples. While large chain-of-thought (CoT) models perform well on this task, distilling them into smaller deployable models remains difficult. We identify a task-specific failure mode in distilled ABSA extraction: student errors at the target-aspect interface create structurally invalid states, such as broken target-aspect bindings and hallucinated targets, which then corrupt downstream predictions. Conventional off-policy distillation is poorly suited to this setting because it trains only on teacher-generated trajectories and provides little supervision on the student-induced structural states that dominate inference. To address this mismatch, we propose STAR-OPD (STructured Aspect-cascade-aware On-Policy Reward Distillation), which builds on generic on-policy distillation and instantiates it for ABSA quadruple extraction with cascade-aware, set-structured rewards. STAR-OPD trains on student rollouts and applies set-structured rewards that directly target binding consistency, target grounding, and fine-grained aspect disambiguation. Experiments on E-ABSA20K and SemEval-2014 show that STAR-OPD consistently outperforms off-policy and general on-policy baselines, reduces target hallucination, and substantially improves performance on structurally hard cases. With Qwen3-4B, STAR-OPD substantially narrows the student-teacher gap while improving inference efficiency, highlighting the importance of on-policy structural correction for distilled ABSA extraction.

[21] Ontology-Driven Structural Regularization for Document-Level Relation Extraction cs.CLPDF

Laura Menotti, Stefano Marchesin, Gianmaria Silvello

TL;DR: 本文提出了一种本体驱动的结构正则化框架，用于解决文档级关系抽取（DocRE）中远监督数据存在的结构不一致噪声问题。该方法通过量化并强制实施关系三元组的结构一致性（如本体约束和逻辑矛盾），有效提升了模型在噪声数据上的泛化性能。

Details

Motivation: DocRE严重依赖昂贵的人工标注数据，而远监督数据（如DocRED distant）因噪声未被充分利用；关键噪声源在于关系三元组的结构不一致性，包括违反本体约束和逻辑矛盾。

Result: 分析显示DocRED distant存在大量结构噪声，且不一致性会传播到模型预测；在训练中强制结构良好性显著减少了逻辑矛盾，并一致提高了泛化性能。

Insight: 将结构一致性确立为DocRE中缺失的监督维度，提出结构正则化作为大规模利用远监督数据的有效策略；本体驱动的框架为噪声数据下的关系抽取提供了可解释的约束方法。

Abstract: Document-Level Relation Extraction (DocRE) relies heavily on costly manually annotated datasets, while large distant supervision resources such as DocRED distant remain underexploited due to noise. We show that a critical yet overlooked source of noise lies in structural inconsistencies within relational triples, including violations of ontology constraints and logical contradictions. We introduce an ontology-driven framework to quantify and enforce structural consistency in DocRE datasets. Our analysis reveals substantial structural noise in DocRED distant and demonstrates that such inconsistencies propagate to model predictions. Enforcing structural well-formedness during training significantly reduces logical contradictions and consistently improves generalization performance. These findings establish structural consistency as a missing axis of supervision in DocRE and highlight structural regularization as an effective strategy for leveraging distant data at scale.

[22] KREL: Automatic Medical Coding via Knowledge-Guided Reasoning over Clinical Evidence with LLMs cs.CL | cs.AIPDF

Xubin Chen, Yipeng Zhou, Wen Sun, Chengkai Huang, Xiaoming Fu

TL;DR: 本文提出KREL框架，通过结合外部ICD编码指南作为结构化知识，利用大语言模型进行临床文本理解和推理，以解决自动医疗编码任务。该方法在基准数据集上表现优于现有基于预训练语言模型和基于大语言模型的方法。

Details

Motivation: 自动医疗编码面临临床文本过长、ICD标签空间巨大以及复杂编码规则难以被大语言模型显式捕获等挑战，现有方法未能有效整合领域知识。

Result: 在基准数据集上的实验表明，KREL一致性地超越了基于预训练语言模型的强基线方法和当前最先进的基于大语言模型的基线方法。

Insight: 创新点在于将外部ICD编码指南作为结构化知识与大语言模型推理紧密耦合，这减少了幻觉并提高了对编码标准的遵从性，实现了领域知识与模型推理的有效整合。

Abstract: Automatic Medical Coding (AMC), which assigns standardized International Classification of Diseases (ICD) codes to clinical notes, is essential for medical reimbursement, quality reporting, and clinical research. Existing pre-trained language model (PLM)-based methods typically formulate AMC as an extreme multi-label classification problem over a predefined code set, while recent large language model (LLM)-based approaches instead frame it as generation or multi-step reasoning. However, key challenges remain, including the extreme length of clinical notes that hinders effective interpretation, the vast ICD label space, and complex coding rules that are not explicitly captured by LLMs. In this work, we propose Knowledge-Guided Reasoning over Clinical Evidence with LLMs (KREL), a framework that leverages LLMs for clinical text understanding and reasoning while integrating external ICD coding guidelines as structured knowledge. This design enables tight coupling between domain knowledge and LLM reasoning, reducing hallucinations and improving compliance with coding standards. Experiments on benchmark datasets show that KREL consistently outperforms strong PLM-based and state-of-the-art LLM-based baselines.

[23] Quantization-Aware Healing: A Practical Recipe for Recovering Compressed, 4-Bit LLMs cs.CL | cs.AI | cs.LG | cs.PFPDF

Bakbergen Ryskulov, Iker García-Ferrero, David Montero, David Jansen, Ali Hashemi

TL;DR: 本文提出了一种称为量化感知修复（QAH）的实用方法，用于恢复经过结构压缩和4位量化的LLMs的性能。该方法通过直接从未压缩的原始模型蒸馏到4位学生模型，避免了传统量化感知训练（QAT）收敛慢和性能崩溃的问题。在GPT-OSS 120B到60B再到MXFP4的流程中，QAH方法在减少约4倍权重内存和一半参数的情况下，在9个基准测试中的7个上匹配或超越了其bfloat16源模型的性能。

Details

Motivation: 当前为了低成本部署LLMs，通常需要将模型进行结构压缩和4位量化，但这会严重损害模型在推理、数学、编码和长上下文等方面的能力，因此需要在部署前进行性能恢复。传统的量化感知训练（QAT）方法收敛缓慢且容易在峰值后崩溃，因此需要一种更稳定、高效的恢复方法。

Result: 在GPT-OSS 120B压缩至60B并量化为MXFP4的流程中，QAH方法在9个基准测试中的7个上匹配或超越了其bfloat16源模型的性能，同时权重内存减少约4倍，参数数量减半。与匹配的QAT基线相比，QAH达到可比峰值速度快约7倍，并且在持续训练下保持稳定，无需手动调整早停。

Insight: 核心创新点在于提出QAH方法，它绕过了传统QAT中需要从压缩后的全精度检查点（本身已是蒸馏恢复的近似）进行训练的限制，而是直接从原始未压缩模型蒸馏到4位量化模型。这显著提升了训练效率和稳定性，并实现了在显著压缩和量化下的高性能恢复，为实际部署提供了无需长时间超参数搜索的实用方案。

Abstract: Serving large language models cheaply increasingly means shipping models that are both structurally compressed to a fraction of their parameters and quantized to 4 bits. Together these steps degrade reasoning, mathematics, coding, and long-context behavior enough to require a recovery, or healing, stage before deployment. The default recipe, quantization-aware training (QAT), re-fits the compressed, quantized model to hard labels; in our pipeline it converged slowly and collapsed past its peak. We adopted Quantization-Aware Healing (QAH) instead. Because a structurally compressed model is never independently trained at full precision, its bfloat16 checkpoint is a distillation-recovered approximation of the original; QAH distills the 4-bit student directly from the original, uncompressed model. On a GPT-OSS 120B to 60B to MXFP4 pipeline, the QAH student matches or beats its bfloat16 source on 7 of 9 benchmarks at roughly 4 times less weight memory and half the teacher’s parameter count, and is released open-weight as Hypernova-60B. Against a matched QAT baseline it reaches a comparable peak about 7 times faster and stays stable under continued training, without hand-tuned early stopping. We also report deployment lessons, including a large, reproducible quality gap between distributed-training backends. Our aim is a recipe deployable without a multi-week hyper-parameter search.

[24] Evidence-Consistent Generative Detection under Scenario-Level Distribution Shift cs.CLPDF

San Kim, JinYeong Bak

TL;DR: 该论文针对社会工程欺诈检测中的场景级分布偏移问题，提出了一种证据一致生成框架ECoG，通过结合证据跨度监督和理由标签一致性目标，提升模型在未见攻击场景下的泛化能力。

Details

Motivation: 传统同分布评估可能因训练和测试数据共享特定任务模式或表面线索而高估模型鲁棒性，这在攻击者可能改变场景、冒充实体或措辞的社会工程欺诈检测中尤为突出。

Result: 在基于0.5B解码器的实验中，ECoG相比未使用一致性正则化的相同骨干网络，将OOD挑战实例的Macro-F1提高了3.22点，减少了4.22点的预测理由与标签不一致情况，并在四个解码器骨干上均表现出预测理由不一致性的降低。

Insight: 创新点在于引入场景级分布外检测评估框架，并提出证据监督与理由标签一致性目标，以缓解模型对场景特定线索的记忆依赖，促进基于决策相关证据的泛化。

Abstract: Conventional in-distribution evaluation can overestimate robustness when training and test data share recurring task-specific patterns or surface cues. This risk is especially relevant in social-engineering fraud detection, where attackers can preserve malicious intent while changing the scenario, impersonated entity, or wording. We study this problem as scenario-level out-of-distribution (SL-OOD) detection for SMS and voice phishing, where entire attack scenarios are held out from training while the label space remains fixed. This setting tests whether models can generalize to unseen attack scenarios using decision-relevant evidence rather than familiar scenario-specific cues. Using this SL-OOD evaluation, we find that high in-distribution performance does not reliably predict held-out robustness across feature-, encoder-, and decoder-based baselines. We interpret this gap as scenario memorization: reliance on recurring scenario-specific lexical or entity cues rather than decision-relevant evidence. We propose ECoG, an evidence-consistent generative framework that combines evidence-span supervision with a rationale-label consistency objective during training. On the 0.5B decoder, relative to the same backbone trained without consistency regularization, ECoG raises Macro-F1 on OOD challenging instances by 3.22 points, reduces the share of predictions whose generated rationale supports the opposite label by 4.22 points, and increases token-level overlap with reference evidence spans by 8.38 points; the reduction in prediction-rationale inconsistency is consistent across four decoder backbones. These results suggest that compact generative detectors can benefit from evidence supervision and rationale-label consistency under social-engineering shift.

[25] Free-Text Evaluation of LLMs for 5G Domain Knowledge and Fault Analysis using LLM-as-Judge cs.CL | cs.AI | cs.NIPDF

Rishiraj Sengupta, Sotiris Chatzimiltis, Mohammad Shojafar, Xiatian Zhu

TL;DR: 本文评估了三种轻量级大语言模型（Claude-Haiku-4.5、GPT-5.4-Mini和Gemini-3.1-Flash-Lite）在5G领域自由文本故障诊断任务上的表现。研究采用LLM-as-Judge方法，在TeleQNA ORAN FT、5G-Faults FT和TeleInter FT三个基准上对模型的领域知识和故障分析能力进行了开放式生成评估。

Details

Motivation: 解决在现实5G/6G网络故障分析中，需要领域专家解读自由文本诊断报告（如根因解释和行动建议）的自动化需求，并探究轻量级、可边缘部署的模型是否具备深度自由文本诊断能力。现有基于固定答案多选题的基准无法满足此评估需求。

Result: 所有模型在故障诊断任务上准确率均达到90%以上，但在零样本召回3GPP和O-RAN规范方面表现不佳（均低于60%）。评估者间一致性均值至少为0.90，表明LLM-as-Judge方法能产生一致且可复现的评分。Gemini-3.1-Flash-Lite在准确率、推理成本和延迟方面取得了最佳效率权衡，最适合生产部署。

Insight: 创新点在于将5G领域评估从限制性多选题转向自由文本生成范式，并系统验证了LLM-as-Judge方法在开放式电信响应评估中的一致性和可重复性。客观来看，研究为轻量级模型在专业领域的边缘部署提供了实证评估框架和效率权衡分析。

Abstract: Real-world fault analysis in 5G and emerging 6G networks demands domain expertise to analyze free-text diagnostics, including root-cause explanations and recommended actions. LLMs have emerged as a promising approach to automating this, yet whether lightweight, edge-deployable models are capable of performing in-depth free-text diagnostics remains an open question. While existing benchmarks rely on restrictive MCQs with fixed answer keys, this paper evaluates 5G domain understanding and fault analysis in a free-text generation format. Transitioning to this paradigm requires evaluating lightweight, edge-deployable AI models on open-ended diagnostic reasoning, alongside a dependable framework to validate these text outputs at scale. To address this we evaluate three lightweight LLMs, Claude-Haiku-4.5, GPT-5.4-Mini, and Gemini-3.1-Flash-Lite, on free-text 5G domain knowledge and fault-analysis tasks across three benchmarks, TeleQNA ORAN FT, 5G-Faults FT, and TeleInter FT. Three independent frontier judges score outputs, and pairwise inter-judge agreement is measured as an empirical test of the LLM-as-Judge methodology. All three models reach at least 90% accuracy on fault diagnosis, while zero-shot recall of 3GPP and O-RAN specifications remains the critical gap, with all models scoring below 60%. Mean inter-judge agreement is at least 0.90 across all runs, indicating that multi-judge LLM scoring produces consistent, reproducible grades for open-ended telecom responses. Operationally, Gemini-3.1-Flash-Lite offers the best efficiency trade-off, combining competitive accuracy with the lowest inference cost and latency, making it the most suitable candidate for production telecom deployments.

[26] EnSI-RAG: Entity-Structure-Indexed Retrieval-Augmented Generation for Long-Document Question Answering cs.CL | cs.AI | cs.DB | cs.IRPDF

Xuanyu Meng, Jiashuo Sun, Jash Rajesh Parekh, Jiawei Han

TL;DR: 本文提出了EnSI-RAG框架，用于解决长文档问答中证据分散和多跳推理的挑战。该方法构建了一个与查询无关、以实体为中心的结构化索引，将文档信息组织为（实体、类型、语义类别、值）的记录。在查询时，利用这些记录作为检索依据，并由大语言模型合成最终答案。

Details

Motivation: 现有基于检索增强生成的方法通常将文档索引为原始文本块并通过嵌入相似性检索，当证据跨越多个实体或需要多跳推理时性能会下降。本文旨在解决长文档问答中证据定位和关系推理的难题。

Result: 在Loong和Oolong基准测试上，EnSI-RAG的平均准确率达到78.24%，比已发布的基线分数高出6.62个百分点，表明其有效性。

Insight: 创新点在于构建了查询无关的实体-结构索引，将证据定位与答案生成解耦，同时保留了可追溯的原始证据链接。这种结构化索引方法能更好地捕捉实体间关系，支持多跳推理，为长文档理解提供了新思路。

Abstract: Question answering (QA) over long, connected documents remains challenging because relevant evidence may span multiple entities and their relationships. Existing retrieval-augmented generation (RAG) methods typically index documents as raw chunks and retrieve them through embedding similarity. Their performance degrades when chunk boundaries separate entities from supporting evidence or when a question requires multi-hop reasoning across the corpus. We propose EnSI-RAG (Entity-Structure-Indexed Retrieval-Augmented Generation), a framework that constructs a query-independent, entity-centered index. Each record (e, t, k, v) represents an entity e, its type t, a semantic category k in {property, relation, aspect}, and a value v, while retaining links to the original source passages. At query time, these records serve as retrieval handles, and an LLM synthesizes the retrieved passages into the final answer. This design separates evidence localization from answer synthesis while preserving traceable source evidence. Across Loong and Oolong, EnSI-RAG achieves an average accuracy of 78.24. Relative to the published baseline scores used as references, this is 6.62 points higher, suggesting its effectiveness across these settings. The code is available at https://github.com/RamonMeng/EnSI-RAG.

[27] Memory Augmentation Unlocks Efficient Chain-of-Thought Reasoning cs.CLPDF

Simeng Zhang, Yilong Chen, Wenyuan Zhang, Zhenyu Zhang, Yao Chen

TL;DR: 本文提出了一种名为“记忆增强压缩”的训练无关框架，旨在解决大语言模型进行思维链推理时推理轨迹冗长导致的推理开销问题。该方法通过从历史推理轨迹中构建可重用的推理记忆，并将其作为预填充侧的脚手架来补偿压缩过程中丢失的信息，从而在保持逻辑连贯性的同时提升推理效率。

Details

Motivation: 动机在于思维链推理虽然能提升大语言模型处理复杂任务的能力，但其冗长的推理轨迹会带来显著的推理开销。现有的思维链压缩方法虽然缩短了生成过程，但激进的压缩可能会破坏逻辑连贯性并导致性能下降。

Result: 实验表明，该方法在数学推理、复杂推理和科学问答任务上持续提升了基于提示的思维链草图压缩方法的性能，在GSM8K、MATH、BBH和MMLU-Sci基准上分别比CoD方法获得了21.4、28.0、29.5和6.61个百分点的准确率提升，同时实现了1.14-1.49倍的延迟加速，优于标准思维链推理。

Insight: 核心创新点在于形式化了“上下文-生成替代定律”，并据此提出了一个无需训练的框架，通过构建和检索可重用的推理记忆（总结推理模式、关键约束和操作）来补偿压缩损失，而非使用原始演示。客观来看，其将记忆机制与推理压缩相结合，提供了一种高效且通用的性能提升途径。

Abstract: Large language models often rely on Chain-of-Thought (CoT) reasoning to solve complex tasks, but verbose reasoning traces introduce substantial inference overhead. CoT compression shortens generation, yet aggressive compression may disrupt logical coherence and degrade performance. We formalize this trade-off as the \textit{Context-Generation Substitution Law}, where explicit reasoning context substitutes for part of decode-time generation. Based on this principle, we propose \textit{Memory-Augmented Compression}, a training-free framework that constructs reusable reasoning memories from historical traces and retrieves them as prefill-side scaffolds. Rather than using raw demonstrations, these memories summarize reusable reasoning patterns, key constraints, and critical operations to compensate for information lost during compression. Experiments show that Memory consistently improves prompt-based Chain-of-Draft (CoD) compression across mathematical reasoning, complex reasoning, and science question answering tasks, yielding accuracy gains of 21.4, 28.0, 29.5, and 6.61 points over CoD on GSM8K, MATH, BBH, and MMLU-Sci, while achieving a 1.14–1.49$\times$ latency speedup over standard CoT. Memory is also compatible with token-level, reasoning-trace-level, and inference-state compression mechanisms. Further analyzes show that the gains come from relevant reasoning memories rather than simply increasing context length.

cs.CV [Back]

[28] Aggregating Visual Information with Optimal Transport for VideoLM Token Compression cs.CVPDF

Wenti Yin, Xiaotian Han, Junyuan Shang, Yuchen Ding, Shuohuan Wang

TL;DR: 本文提出了AVIOT（基于最优传输的视觉信息聚合）方法，用于压缩视频语言模型中的视觉令牌序列。该方法将视频令牌压缩建模为将密集的帧观测经验测度传输到紧凑目标测度的最优传输问题，并通过任务和空间轴进行自适应调整，以在压缩过程中保留分散在帧间的视觉信息。

Details

Motivation: 视频语言模型将视频处理为密集的视觉令牌序列，存在大量表征冗余，增加了语言模型解码的负担。核心挑战在于在压缩这些序列的同时，保留分散在不同帧中的视觉信息。

Result: 在不同压缩比下的评估表明，AVIOT在多个视频理解基准测试中匹配或超越了未压缩的基线模型，并在更高的压缩比下保持了强劲的性能。

Insight: 创新点在于将视频令牌压缩形式化为最优传输问题，通过源-目标耦合直接指定压缩表示的构建方式。进一步通过问题条件调节传输成本以聚焦任务相关内容，并通过多空间粒度计算区域特定的传输计划，使同一紧凑表示的不同区域能从不同时刻汲取信息。

Abstract: Video language models process videos as dense visual-token sequences with substantial representational redundancy. Compressing these sequences is therefore essential for reducing the visual-token burden on language-model decoding. The central challenge is to preserve visual information dispersed across frames under such compression. To this end, we introduce Aggregating Visual Information with Optimal Transport (AVIOT), which casts video token compression as transporting a dense empirical measure of frame observations onto a compact target measure. The resulting source-to-target coupling induces a distribution over source observations for each target support, directly specifying how the compressed video representation is constructed. We further adapt this construction along task and spatial axes. Question conditioning modulates the transport cost between source frames and target supports, while influencing how many supports are allocated to each temporal segment, thereby directing representation capacity toward question-relevant content. At multiple spatial granularities, AVIOT computes region-specific temporal transport plans and adaptively fuses the representations they yield, allowing different regions within the same compact representation to draw from different moments. Evaluations across varying compression ratios show that AVIOT matches or outperforms the uncompressed baseline on multiple video-understanding benchmarks while retaining strong performance at higher compression ratios.

[29] RISE: Adaptive Imagination for World Action Models cs.CVPDF

Hongbo Lu, Liang Yao, Chenghao He, Hao Han, Fan Liu

TL;DR: 本文提出了RISE（通过选择性推演优化想象），一种系统级自适应想象框架，用于改进世界行动模型（WAMs）。RISE通过一个潜在评估器估计当前前缀揭示的风险以及继续想象可能带来的规划收益，并结合一个推演门权衡预期收益与额外计算成本，从而动态决定是否继续推演。此外，作者构建了CounterDrive反事实数据集，以丰富未来动态并提供局部风险监督。在NAVSIM和nuScenes上的实验表明，RISE在减少不必要推演的同时，实现了最佳的整体规划性能。

Details

Motivation: 现有世界行动模型（WAMs）为每个场景分配固定的想象预算，但这种方法效率低下，因为并非所有场景都需要同等深度的未来推演。RISE旨在通过自适应地分配想象资源，根据场景的预期规划收益动态决定推演深度，以优化计算效率并提升规划性能。

Result: 在NAVSIM和nuScenes基准测试中，RISE实现了最佳的总体规划性能，同时显著减少了不必要的推演计算。额外的迁移实验结果支持了其作为插件模块在不同WAM架构中的通用性。

Insight: 创新点在于引入了系统级自适应想象框架，通过序列化的Roll/Stop决策机制动态调整想象预算；同时，构建了CounterDrive反事实数据集，为安全关键的世界建模研究提供了包含多样化结果和风险级别的可重用资源，这有助于更准确地监督和评估风险动态。

Abstract: World Action Models (WAMs) improve planning by incorporating future world evolution into action generation, yet existing methods allocate a fixed imagination budget to every scene. We propose RISE (\textbf{R}efining \textbf{I}magination through \textbf{SE}lective Rollout), a system-level adaptive imagination framework that makes sequential \textsc{Roll}/\textsc{Stop} decisions according to the expected planning benefit of continued rollout. At each step, a Latent Evaluator estimates the risk revealed by the current prefix and how much planning could improve if imagination continues, while a Rollout Gate weighs this expected benefit against additional computation cost. Since factual driving logs expose only one realized future, we further construct \textbf{CounterDrive}, a counterfactual dataset with diverse outcomes and risk levels, to enrich future dynamics and provide localized risk supervision. Each retained sample undergoes expert verification and annotation of trajectory validity, incident onset, and causal category, providing a reusable resource for safety-critical world-modeling research. Experiments on NAVSIM and nuScenes show that RISE achieves the best overall planning performance while reducing unnecessary rollout, with additional transfer results supporting its plug-in generality across WAM architectures.

[30] DiffVC-ONE: Diffusion-based Generative Video Compression with One-Step Video Diffusion Transformer cs.CVPDF

Wenzhuo Ma, Zhenzhong Chen

TL;DR: DiffVC-ONE是一个基于扩散模型的生成式视频压缩框架，其核心是使用一步式视频扩散变换器。该框架通过统一的单向潜在压缩器高效压缩视频，并利用混合条件生成器提取多种条件，在单步扩散增强中提升感知质量与时间一致性，同时保持低推理成本。

Details

Motivation: 解决生成式视频压缩在低码率下同时实现高时间一致性和低推理成本的挑战。

Result: 在多个标准基准测试上的广泛实验表明，DiffVC-ONE在感知质量和时间一致性方面达到了最先进的水平，且推理成本低。

Insight: 创新点包括统一的单向潜在压缩器、基于视频DiT的一步扩散增强器以及混合条件生成器，实现了高效的单步时空感知增强与条件控制。

Abstract: Generative video compression can recover rich visual details at low bitrates, but simultaneously achieving high temporal consistency and low inference cost remains challenging. To address this issue, we propose DiffVC-ONE, a diffusion-based generative video compression framework built on a one-step Video Diffusion Transformer. First, we introduce a Unified Unidirectional Latent Compressor that uses a shared model to efficiently and uniformly compress compact latent slices. We then develop a Video DiT-based One-Step Diffusion Enhancer that uses the reconstructed latent slices as content anchors and performs single-step spatio-temporal perceptual enhancement over an entire group of pictures. Finally, a Hybrid Condition Generator extracts structural, strength, and semantic conditions from the reconstructed content and quantization information. These conditions preserve faithful regions, control the degree of generative enhancement, and supplement content-aware perceptual details during one-step diffusion enhancement. Extensive experiments on multiple standard benchmarks demonstrate that DiffVC-ONE achieves state-of-the-art perceptual quality and temporal consistency with low inference cost.

[31] Annotations as Rollouts: Efficient and Scalable Reinforcement Learning for Video MLLMs cs.CVPDF

Yunheng Li, Guohong Mu, Hao Li, Shengsheng Qian, Dingwen Zhang

TL;DR: 本文提出OraRL，一种用于视频多模态大语言模型（MLLMs）的高效可扩展强化学习后训练方法。该方法创新性地将标注数据作为“预言机轨迹”直接整合到策略优化中，并通过解耦优势估计和符号平衡剪枝来克服优势反转问题，显著提升了样本效率和训练速度。

Details

Motivation: 现有视频MLLMs在大型多任务数据集上进行强化学习后训练时面临挑战，因为传统方法依赖少量高质量轨迹（如思维链生成），导致样本效率低且成本高昂。

Result: OraRL在多个基准测试上取得SOTA性能：Video-ORA-9B在时间mIoU（66.0 vs 62.5）、跟踪AO（78.2 vs 73.0）、分割（70.4 vs 64.3）以及空间智能宏观平均（56.1 vs 51.0）上均超越先前最佳模型；在VSI-Bench上得分73.1，显著优于GPT-5（55.0）和Gemini-3-Pro（55.1）。训练效率方面，其步长时间仅为SFT的2.2倍，远低于GRPO with CoT的4.9倍，且解码速度大幅提升（130 ms vs 4,780 ms）。

Insight: 核心创新在于将标注数据重新定义为可直接优化的“预言机轨迹”，并设计了解耦优势估计器来避免优势反转问题；同时，符号平衡剪枝策略有效提升了训练效率。该方法为MLLMs的高效RL后训练提供了新范式，实现了性能与效率的同步提升。

Abstract: Multimodal large language models (MLLMs) have become a prevailing paradigm for unified video perception. However, post-training on large multi-task datasets remains challenging, as existing reinforcement learning methods sample on-policy groups with few high-quality rollouts even with costly chain-of-thought (CoT) generation. In this paper, we study the sample efficiency and scalability of RL post-training for video MLLMs and introduce OraRL. We identify an overlooked role for annotations: Beyond scoring rollouts, each can enter its on-policy group as an oracle rollout, a direct positive optimization target. Direct oracle integration, however, is nontrivial: a high-reward oracle raises the group baseline and inverts otherwise positive policy advantages, a failure we term advantage inversion. At the core of OraRL is a decoupled advantage estimator: policy rollouts determine an oracle-free baseline, while the oracle-policy gap modulates both a directional gain and a separate detached oracle advantage. Sign-balanced pruning improves efficiency: by retaining only the oracle and the strongest rollouts of each sign, OraRL requires just 2.2x the step time of SFT, less than half the 4.9x required by GRPO with CoT. OraRL scales with model size and data, surpassing its backbone from 0.8B to 9B and GRPO up to 100k prompts. Without chain-of-thought, Video-ORA-9B decodes in 130 ms instead of 4,780 ms. Compared with the respective prior best models, it raises temporal mIoU from 62.5 to 66.0, tracking AO from 73.0 to 78.2, segmentation from 64.3 to 70.4, and the three-benchmark spatial-intelligence macro average from 51.0 to 56.1; on VSI-Bench, it scores 73.1 against 55.0 for GPT-5 and 55.1 for Gemini-3-Pro.

[32] Grounded-Exo2Ego: Structured Semantic Grounding for Robust Exocentric-to-Egocentric Video Generation cs.CVPDF

Shengze Wang, Michael Stengel, Tianye Li, Seonwook Park, Amrita Mazumdar

TL;DR: 本文提出Grounded-Exo2Ego框架，用于解决从单视角外中心视频生成高质量自我中心视频的难题。该框架采用双分支视频扩散模型，结合几何锚定分支与新颖的语义接地分支，并引入相机重定位算法解决相机-重建错位问题，同时在合成数据引擎支持下，在EgoExo4D数据集上大幅超越现有方法。

Details

Motivation: 解决在极端视角变化和大范围不可观测区域下，传统基于几何条件的新视角合成方法不可靠的问题，以提升外中心到自我中心视频生成的质量和鲁棒性。

Result: 在具有挑战性的EgoExo4D数据集上，该方法在所有指标上均大幅超越近期最先进（SOTA）方法，并通过详尽的消融实验验证了数据和架构层面各贡献的有效性。

Insight: 创新性地将语义接地分支与几何锚定分支结合，突破了主流纯几何方法的局限；提出的相机重定位算法有效解决了相机与3D重建错位这一被忽视的关键问题；全自动合成数据引擎为模型训练提供了高质量、可扩展的数据支持。

Abstract: Generating egocentric video from a single exocentric video is an emerging and important topic for AR/VR and physical AI. Compared with conventional novel view synthesis, exo-to-ego generation is a significantly harder task because the standard geometric conditioning becomes highly unreliable under extreme view changes and large unobservable regions. We present Grounded-Exo2Ego, a principled framework that addresses these challenges at both the architectural and data levels. Architecturally, Grounded-Exo2Ego is a dual-branch video diffusion model that couples a geometric anchoring branch, which conditions the generation on the rendering of a 3D reconstruction, with a novel semantic grounding branch, which goes beyond the prevailing geometry-based approach and improves quality by synthesizing challenging regions based on object-level context. Additionally, we found that the overlooked issue of camera-reconstruction misalignment severely undermines exo-to-ego learning. We thus introduce a camera re-localization algorithm that resolves this issue and substantially improves quality across all metrics. We further develop a fully automated synthetic data engine that generates and renders rigged 3D characters in procedurally generated environments. Evaluation on the challenging EgoExo4D dataset shows that our method outperforms recent state-of-the-art approaches by large margins across all metrics. Detailed ablations validate improvements from each of our contributions at both the data and architectural level.

[33] Keep Your Friends Close, and the Right Neighbours Closer: Disaster-Conditioned Kernel-Regularized Graph Attention for Building Damage Classification cs.CV | cs.LGPDF

Fuad Hasan, Chul Min Yeum

TL;DR: 本文提出了一种用于建筑物损毁分类的灾难条件化核正则化图注意力方法，通过结合可学习的多尺度空间核先验，使模型能够根据灾难类型自适应调整有效邻域尺度，并引入残差去相关损失来避免过度平滑。

Details

Motivation: 现有方法主要依赖建筑物自身外观线索，忽略了灾难损毁的空间结构化特性，且不同灾难类型（如洪水、飓风、野火）的空间聚类行为差异显著，导致空间上下文容易被误用。

Result: 在xBD数据集上采用留一事件出（LOEO）协议评估，模型在零样本事件偏移下提升了宏观F1分数，并显著降低了预测残差的空间自相关性；在从xBD到Ida-BD的跨数据集迁移中也表现出更可靠的泛化能力。

Insight: 创新点在于将灾难类型作为条件注入图注意力机制，通过可学习核先验实现邻域尺度的自适应调整，并结合残差去相关损失来区分真实空间模式与过度平滑，从而更有效地利用空间上下文信息。

Abstract: Disaster damage is spatial: buildings rarely fail in isolation. Yet using spatial context for damage classification remains surprisingly underexplored, and many pipelines still rely primarily on per-building appearance cues even when the dominant uncertainty is spatially structured. Complicating matters, the right neighbourhood is not the same across events. Floods, hurricanes, and wildfires can exhibit very different clustering behaviour, making spatial reasoning valuable but easy to misuse - naive context aggregation can improve visual coherence while oversmoothing boundaries or propagating structured errors. We study this tension on xBD (the dataset used in the xView2 challenge) in a controlled post-localization, classification-only setup: each building is represented by a pre/post combined (PPC) patch cropped from the provided polygons, and spatial context is modelled with GPS-derived building graphs. Our approach keeps local evidence “close” by preserving strong spatial relationships in disaster damage patterns, while bringing only the right neighbours “closer” through a disaster-type-conditioned graph model that injects a learnable multi-scale spatial kernel prior into attention, allowing the effective neighbourhood scale to adapt across disaster types rather than being learned as a single global smoothing rule. To discourage coherence-by-smoothing, we add a residual de-correlation loss that penalizes positive Moran’s~I in prediction residuals. We evaluate the method under event and dataset shift with a leave-one-event-out (LOEO) protocol on xBD and cross-dataset transfer from xBD to Ida-BD. The model improves macro-F1 and substantially reduces residual spatial autocorrelation under zero-shot event shift, indicating better use of spatial context rather than naive smoothing and enabling more reliable transfer to unseen events within known disaster types.

[34] Learning Prostate Anatomy at Test Time for Cancer Detection in Micro-Ultrasound cs.CV | cs.LGPDF

Obed Korshie Dzikunu, Mohammad Mahdi Abootorabi, Mohamed Harmanani, Paul F. R. Wilson, Emma Willis

TL;DR: 本文提出了一种名为ANT的分割引导测试时适应框架，用于解决前列腺癌检测中因不同成像硬件或采集协议导致的领域偏移问题。该方法通过在测试时利用冻结的预训练分割网络生成的伪掩码，监督一个辅助的前列腺分割任务，从而将预训练的癌症检测编码器适应到目标领域，校正领域特定的特征漂移。

Details

Motivation: 解决深度学习模型在前列腺癌检测中，因不同临床中心使用不同成像硬件或协议而产生的领域偏移问题，现有测试时适应方法主要纠正图像外观的统计差异，但忽略了目标领域的解剖结构。

Result: 在采用留一中心出协议的统一评估条件下，与不进行适应的方法相比，ANT在活检核心水平和患者水平上分别将平均AUC提高了2.9%和3.6%，优于其他测试时适应基线方法。

Insight: 创新点在于将解剖结构信息（通过分割任务）引入测试时适应过程，以校正特征漂移并保留癌症判别性结构，这为医学影像领域适应提供了一种结合解剖先验知识的新思路。

Abstract: Domain shift across clinical centers using different imaging hardware or acquisition protocols remains a fundamental barrier to deploying deep learning models for prostate cancer (PCa) detection. Existing test-time adaptation (TTA) methods address distribution shift through entropy minimization or augmentation-based self-supervision, correcting for statistical differences in image appearance but ignoring the anatomical structure of the target domain. We propose ANT, a segmentation-guided TTA framework that adapts a pretrained cancer detection encoder to the target domain by solving an auxiliary prostate segmentation task at test time, supervised by pseudo-masks from a frozen pretrained segmentation network. By aligning encoder representations to prostate anatomy in the target domain, ANT corrects domain-specific feature drift while preserving cancer-discriminative structure. The model was trained on 693 patients imaged with an earlier-generation micro-ultrasound scanner in a multi-center clinical trial, and evaluated on 118 patients acquired with a newer-generation system across two centers in another clinical trial. Under a leave-one-center-out protocol with identical evaluation conditions across all methods, ANT improves mean AUC by 2.9% and 3.6% at the biopsy-core and patient levels, respectively, over no adaptation, outperforming TTA baselines. Code is available at: https://github.com/ObedDzik/ant.git.

[35] Bridging Language and Spherical Space: Object-Centric Control for Text-to-Panorama Generation cs.CVPDF

Derui Li, Qian Qiao, Yuhao Sun, Wenhao Guo, Peng Lu

TL;DR: 本文提出了PanoCtrl，一个以对象为中心的可控文本到全景图生成框架。该方法通过将文本描述转换为结构化的对象级球面条件，并将其集成到扩散过程中，显式地桥接了自然语言与球面全景空间。

Details

Motivation: 现有文本到全景图方法主要依赖隐式空间推理，难以忠实地将对象级方向描述在球面全景场景中落地。引入显式布局会降低基于语言交互的灵活性，且无法直接解决以自我为中心的方向性语言与全景图像空间之间的错位问题。

Result: 大量实验表明，PanoCtrl在空间对齐和图像质量方面均达到了最先进的性能。

Insight: 核心创新在于提出了PanoParse文本解析器来预测对象语义和球面边界视场参数，以及PanoControl模块通过对象感知注意力和空间残差增强将对象级语义和空间指导注入扩散Transformer。此外，构建了带有对象级球面标注和多样化方向描述的PanoGround数据集来支持该任务。

Abstract: Panoramic image generation is increasingly important for immersive applications such as virtual reality, augmented reality, and 3D content creation. Unlike perspective images, panoramic images represent a viewer-centered $360^\circ$ surrounding space, where directional expressions such as left, right, front, and behind play a central role in spatial understanding. However, existing text-to-panorama methods largely rely on implicit spatial reasoning and often fail to faithfully ground object-level directional descriptions in spherical panoramic scenes. A straightforward alternative is to introduce explicit layouts, but requiring manually specified spatial conditions reduces the flexibility of language-based interaction and does not directly resolve the misalignment between egocentric directional language and panoramic image space. To address this issue, we propose PanoCtrl, an object-centric framework for controllable text-to-panorama generation. Our method explicitly bridges natural language and spherical panoramic space by converting textual descriptions into structured object-level spherical conditions and integrating them into the diffusion process. Specifically, we introduce PanoParse, a text-conditioned parser that predicts object semantics and spherical bounding field-of-view (BFoV) parameters, and \textbf{PanoControl}, which injects object-level semantic and spatial guidance into the diffusion transformer through object-aware attention and spatial residual enhancement. To support this task, we construct PanoGround, a dataset with object-level spherical annotations and diverse directional descriptions for controllable panoramic generation. Extensive experiments demonstrate that PanoCtrl achieves state-of-the-art performance in both spatial alignment and image quality.

[36] Identity-Aware Human-Object Interaction Motion Captioning cs.CV | cs.AIPDF

Yiming Wang, Yonghao Dang, Huilai Li, Jiawei Tu, Jianqin Yin

TL;DR: 本文提出了身份感知的人-物交互（HOI）运动描述任务，要求生成的描述不仅说明交互动作，还需指明主体身份（如‘Sub_ID举起椅子’而非‘一个人举起椅子’）。为此，作者基于BEHAVE和InterCap数据集构建了身份感知的HOI运动描述数据，并提出了ID-HOINet模型，该模型通过多视角视频学习，支持单视角下的身份感知HOI运动描述生成。

Details

Motivation: 现有HOI运动描述方法通常使用‘一个人’等通用术语指代主体，缺乏对主体身份的识别和描述，限制了描述的准确性和实用性。本文旨在解决这一局限性，使模型能够生成包含具体身份信息的交互描述。

Result: 实验表明，ID-HOINet在身份感知HOI运动描述任务上取得了最先进的性能（state-of-the-art），具体基准未在摘要中明确提及，但基于BEHAVE和InterCap数据集进行评估。

Insight: 创新点包括：1）首次提出身份感知HOI运动描述任务，扩展了传统HOI描述的范围；2）设计了ID-HOINet模型，其核心包含多视角身份-运动学习模块（MVIML），通过建模时间阶段和相机视角间的依赖关系来捕获身份和交互特征；3）采用两阶段描述重写策略（TSCR），先检索身份并生成身份无关描述，再重写为身份感知描述，提高了生成准确性和效率。

Abstract: Existing human-object interaction (HOI) motion captioning methods typically describe what happens while referring to the subject using generic terms such as “a person” or “someone”, without grounding the caption in subject identity. To address this limitation, we introduce Identity-Aware Human-Object Interaction Motion Captioning task. This task requires each generated caption to specify both the subject identity and the corresponding HOI motion. For example, the model generates “Sub_ID lifts the chair” rather than “A person lifts the chair”. For this task, we design identity-aware HOI motion captions based on the BEHAVE and InterCap datasets. We further propose ID-HOINet, which learns from multi-view videos while supporting single-view identity-aware HOI motion caption generation. ID-HOINet contains two core components: Multi-View Identity-Motion Learning Module (MVIML) and Two-Stage Caption Rewriting Strategy (TSCR). MVIML learns from multi-view videos by modeling dependencies across temporal stages and camera viewpoints, capturing identity and interaction motion features. At inference, the TSCR first retrieves the subject identity and generates identity-agnostic HOI motion captions. TSCR then rewrites these captions with the predicted identity to produce the final identity-aware HOI motion captions. Experiments demonstrate that ID-HOINet achieves state-of-the-art performance. Code will be released upon acceptance.

[37] ArtiMo: Agent-Driven Articulated Mesh Animation cs.CVPDF

Chunyu Zou, Peng Dai, Yi-Hua Huang, Ze Yuan, Jingwei Huang

TL;DR: ArtiMo是一个基于智能体驱动的框架，用于文本引导的关节化3D网格动画生成。该框架利用大型语言模型和视觉语言模型（LLMs/VLMs）构建智能体流程，通过结合URDF的显式运动学约束与智能体的推理规划能力，以零样本方式生成因果连贯的部件运动和交互。此外，它引入了视觉自改进机制，通过渲染关键帧和运动线索让VLM迭代诊断和修正错误，并贡献了一个包含21个关节物体类别的新基准数据集。

Details

Motivation: 现有数据驱动的网格动画方法缺乏针对文本引导关节动画的任务特定训练数据和显式关节监督，难以满足严格的运动学约束、部件间因果交互建模以及指令保真度的要求。

Result: 在涵盖21个关节物体类别的新基准数据集上进行的大量实验表明，ArtiMo显著优于基线方法，特别是在复杂、因果驱动的运动上表现出色。

Insight: 创新点在于将智能体推理与显式运动学约束（URDF）相结合，以零样本方式实现因果连贯的动画生成，并引入视觉自改进机制进行迭代错误修正；同时构建了带有因果关系注释的高质量基准数据集，推动了该领域评估的发展。

Abstract: Animating articulated 3D meshes via text requires satisfying strict kinematic constraints, modeling causal interactions between parts, and achieving instruction fidelity. Due to the absence of task-specific training data and explicit articulation supervision, existing data-driven mesh animation methods are largely inapplicable to this setting. To address this, we propose ArtiMo, a novel agent-driven framework for text-guided articulated mesh animation. Operating in a zero-shot manner, ArtiMo develops an agentic pipeline powered by Large Language and Vision-Language Models (LLMs/VLMs) to orchestrate motion generation. By synergizing the explicit kinematic constraints of URDF with the agent’s reasoning and planning capabilities, it effectively produces causally coherent part motions and interactions without requiring model fine-tuning. To ensure motion correctness, the agent additionally utilizes a visual self-improvement mechanism: generated animations are rendered into compact keyframes and motion cues, enabling the VLM to iteratively diagnose and correct errors. Furthermore, we contribute a new benchmark dataset spanning 21 articulated object categories, featuring high-quality motion annotations enriched with causal relationships. Extensive experiments demonstrate that ArtiMo significantly outperforms baselines, particularly on complex, causally driven motions. The project page is available at https://zou-2004.github.io/ArtiMo/.

[38] AGIDefect-4K: A Richly Annotated Dataset for AI-Generated Image Defect Detection, Localization and Explanation cs.CVPDF

Xiangfei Sheng, Weidong Zou, Tianjiao Gu, Zhichao Yang, Pengfei Chen

TL;DR: 本文介绍了AGIDefect-4K数据集，这是一个包含4000张来自15种最先进生成模型（包括开源和闭源系统）图像的、具有丰富注释的数据集，用于AI生成图像（AGI）的缺陷检测、定位和解释。数据集提供了分层的缺陷注释：检测标签、像素级分割掩码和详细的文本解释，以及整体质量评分。基于此，作者提出了AGIDA基线框架，利用多模态大语言模型（MLLMs）进行联合缺陷检测、定位、解释和质量预测。

Details

Motivation: 尽管生成式AI能产生高度逼真的图像，但当前模型仍存在微妙但关键的缺陷，影响其可靠性。现有的AI生成图像评估基准虽有进展，但全面的AGI缺陷诊断仍未得到充分探索。

Result: 在AGIDefect-4K上的综合基准测试表明，AGI缺陷理解仍然具有挑战性，凸显了该数据集的价值。AGIDA框架作为基线模型在该数据集上进行了评估。

Insight: 主要创新点是构建了一个大规模、多层次注释（检测、分割、文本解释、质量评分）的AGI缺陷诊断数据集AGIDefect-4K，填补了该领域的空白。同时，提出了一个基于MLLMs的联合任务基线框架AGIDA，为后续研究提供了基准和方向。

Abstract: Generative AI can now produce highly realistic images, yet current models still exhibit subtle but critical defects that undermine their reliability. While existing AI-generated image (AGI) evaluation benchmarks have made notable progress, comprehensive AGI defect diagnosis remains underexplored. To bridge this gap, we introduce AGIDefect-4K, a richly annotated dataset of 4,000 images from 15 state-of-the-art generative models spanning both open-source and closed-source systems. AGIDefect-4K features hierarchical defect annotations: (1) detection labels identifying whether defects exist, (2) pixel-level segmentation masks localizing defective regions, and (3) detailed textual explanations characterizing defect types and their perceptual impact. Each image is further annotated with an overall quality score. Building on this, we present AGIDA (AGI Defect Assistant), a baseline framework leveraging Multimodal Large Language Models (MLLMs) for joint defect detection, localization, explanation, and quality prediction. Comprehensive benchmarking on AGIDefect-4K reveals that AGI defect understanding remains challenging, underscoring the value of this dataset. The dataset is publicly available at https://github.com/sxfly99/AGIDefect-4K.

[39] AffordAny: Open-World 3D Affordance Grounding from Monocular RGB Images via Vision-Language-Guided Geometric Reasoning cs.CVPDF

Junqi Wu, Kaihua Tang, Xuanwen Chen, Hongzhi Li, Jianqiang Huang

TL;DR: 本文提出了AffordAny，一个端到端的框架，用于从单目RGB图像进行开放世界的3D功能可及性（affordance）定位。该方法通过视觉-语言引导的几何推理，利用冻结的视觉语言模型（VLM）特征与3D几何信息融合，并采用伪标签自训练提升泛化能力。

Details

Motivation: 现有方法通常依赖于预构建的以对象为中心的3D几何和封闭的可及性本体，限制了从原始RGB观测中的部署。本文旨在解决开放世界3D可及性定位问题，即给定自由形式的语言查询，在3D中定位功能对象部件。

Result: 在系统泛化协议下评估未见对象、未见类别和未见指令释义，自训练后在未见对象上达到0.428 IoU，在未见类别上达到0.315 IoU，未见类别的mIoU相对提升6.3%（p<0.01），指令敏感性差距仅为0.105，证明了方法的有效性和鲁棒性。

Insight: 创新点包括：通过自动化流程构建大规模文本条件3D部件监督数据集；设计解码器将冻结的Cosmos-2B特征与3D几何通过空间投影、指令条件语义压缩和双向几何-语义交互进行渐进融合；采用最小扰动伪标签自训练扩展新对象而无需人工标注。

Abstract: Open-world 3D affordance grounding requires localizing functional object parts in 3D given free-form language queries. Existing methods typically assume pre-built object-centric 3D geometry and closed affordance ontologies, limiting deployment from raw RGB observations. We present AffordAny, an end-to-end framework that uses one monocular RGB image to construct large-scale text-conditioned 3D part supervision, ground affordances with a frozen vision-language model (VLM) guided decoder, and improve open-world generalization through pseudo-label self-training. Our automated pipeline produces a benchmark of 5,334 objects and 10,633 part-level samples spanning 473 categories, an order-of-magnitude increase in categorical diversity over prior work. The decoder progressively fuses frozen Cosmos-2B features with 3D geometry through spatial projection, instruction-conditioned semantic compression, and bidirectional geometry-semantics interaction. Minimal-perturbation pseudo-label self-training further adds new objects without human annotation. Under a systematic generalization protocol evaluating unseen objects, unseen categories, and unseen instruction paraphrases, our approach achieves 0.428 IoU on unseen objects and 0.315 IoU on unseen categories after self-training, with unseen-category mIoU improving by 6.3% relative (p<0.01) and an instruction sensitivity gap of only 0.105, demonstrating effectiveness and robustness of our method.

[40] Generating Multi-view Adversarial Examples for Visual Geometry Grounded Transformer cs.CVPDF

Qi Song, Ziyuan Luo, Haoliang Han, Renjie Wan

TL;DR: 本文提出了一种名为MVAP-G的多视角对抗扰动生成器，用于攻击视觉几何基础Transformer（VGGT）模型。该方法通过前馈方式生成跨多视角图像的一致且难以察觉的对抗扰动，无需昂贵的逐场景优化，显著降低了VGGT在3D重建任务上的性能。

Details

Motivation: 动机在于高性能的VGGT模型在部署时可能存在严重的安全漏洞，而传统的对抗扰动方法（如逐场景优化）成本高昂，且通用对抗扰动（UAP）依赖单一静态模式，无法有效攻击VGGT。

Result: 实验表明，MVAP-G在推理时无需迭代优化，就能显著降低VGGT的性能，揭示了3D基础模型存在严重漏洞。

Insight: 创新点在于提出了首个针对3D基础模型的多视角对抗攻击方法，设计了跨视角对抗对齐机制以确保扰动在不同场景和视角间的一致性，这为评估和提升3D视觉系统的鲁棒性提供了新思路。

Abstract: The Visual Geometry Grounded Transformer (VGGT) enables unified feed-forward 3D reconstruction from multi-view images. However, deploying such a high-performance model may expose critical security vulnerabilities. Traditional adversarial perturbations require costly per-scene optimization, while Universal Adversarial Perturbations (UAPs) rely on a single static pattern and fail to effectively attack VGGT. To address these limitations, we propose \textbf{MVAP-G}, a multi-view adversarial perturbation generator that produces imperceptible consistent perturbations across multiple views in a single feed-forward pass. To ensure perturbation consistency across diverse scenes, we design a cross-view adversarial alignment mechanism to process multi-view images. Experiments demonstrate that MVAP-G significantly degrades VGGT performance without iterative optimization during inference. This work pioneers multi-view adversarial attacks on 3D foundation models, uncovering severe vulnerabilities and underscoring the urgent need for robust 3D vision systems. The code is available at https://github.com/qsong2001/mvap-g.

[41] VisTa3D: A Dataset and Benchmark for Thin Object Reconstruction from Vision, Tactile, and 3D Point Clouds cs.CVPDF

Shania Guo, Yeongsik Seo, Andrew Fu, Mei Hao, Iris Xia

TL;DR: 该论文提出了VisTa3D数据集，这是首个包含同步RGB图像、深度图、触觉响应图以及激光扫描地面真值数据的薄物体数据集，用于评估和提升薄物体的3D重建性能。论文发现现有最先进的3D重建模型在薄物体上表现不佳，并假设触觉数据可以提供局部形状和变形信息以辅助重建。为此，论文还引入了一个结合视觉、深度和触觉数据的基线3D重建模型进行验证。

Details

Motivation: 现有基于视觉或深度数据的3D重建模型在薄物体上性能不足，因为薄物体在RGB图像和3D点云中占据空间较小，导致重建精度低。论文旨在通过引入触觉数据来解决这一问题，并为此创建了一个专门的数据集来测试触觉数据是否有助于改善薄物体的重建。

Result: 论文在VisTa3D数据集上对当前3D重建模型进行了基准测试，证实了它们在薄物体上的重建保真度较低。作为基线，论文提出的首个视觉-深度-触觉融合3D重建模型展示了触觉数据可能带来的改进潜力，但具体定量结果未在摘要中详细说明。

Insight: 论文的主要创新点在于创建了首个多模态（视觉、深度、触觉）薄物体数据集VisTa3D，并提出了融合触觉数据的3D重建基线模型。从客观角度看，这为薄物体重建领域提供了新的数据资源和研究方向，强调了触觉信息在补充视觉和深度数据不足方面的潜在价值。

Abstract: State-of-the-art 3D reconstruction models, whether from visual, range, or both, tend to underperform on thin objects. This is partially due to the small amount of space such objects occupy in RGB images and in 3D point clouds. To test the extent of their errors, we collected the first thin object dataset comprising of synchronized RGB images, depth maps, and tactile response maps, where each frame is associated with inertial measurements, camera pose and calibration, and groundtruth depth and segmentation maps obtained from laser scanning of thin objects. We hypothesize that tactile data can aid in the reconstruction of thin objects as their response maps provide local shape and deformation information. Our dataset, termed VisTa3D, comprises of 387 scenes covering 70 thin objects over 17 environments. We benchmarked current 3D reconstruction models on VisTa3D and found that, indeed, they exhibit low fidelity on thin objects. To test if tactile data can help, we introduce the first visual-range-tactile 3D reconstruction model as a baseline. Code and data: https://huggingface.co/datasets/shaniaguo/VisTa3D.

[42] Identity-Preserving Text-to-Video Generation via Agentic Enhancement and Semantic Repair cs.CVPDF

Jiayi Gao, Changcheng Hua, Jiaqi Tang, Yuxin Peng, Yang Liu

TL;DR: 本文提出了一种名为AESR的轻量级增强框架，旨在解决身份保持文本到视频生成中的身份漂移、指令跟随不完整和视觉细节缺失等问题。该框架包含全局代理提示增强模块和样本级视觉语义修复模块，通过优化提示构造和修复生成视频中的错误来提升生成质量。在ACM MM 2026挑战赛中，该方法在Track 1中排名第一，验证了其有效性。

Details

Motivation: 现有商业视频生成模型虽然视觉质量和运动真实感强，但在复杂提示下容易出现身份漂移、指令跟随不完整和细节缺失等问题，且由于模型通常是闭源黑盒，无法直接通过参数优化进行改进。

Result: 在ACM MM 2026身份保持视频生成挑战赛的官方评估协议下，所提出的系统MIPL_Video在Track 1中排名第一，证明了AESR框架在实际身份保持视频生成任务中的有效性。

Insight: 创新点在于提出了一个无需修改底层生成模型参数的轻量级增强框架，通过代理循环学习模型特定的提示格式、人类中心先验和领域经验来优化提示，并利用VLM定位错误、设计修复指令，结合视频编辑模型进行局部语义或身份错误修复，同时采用专家混合选择策略确保输出可靠性。

Abstract: Identity-preserving video generation aims to synthesize videos that follow natural-language instructions while maintaining the visual identity of a given subject. Recent commercial video generation models have achieved strong visual quality and motion realism, but they still suffer from identity drift, incomplete instruction following, and missing visual details under complex prompts. Since these models are usually closed-source black boxes, directly improving them through parameter optimization is often infeasible. We therefore propose Agentic Enhancement and Semantic Repair (AESR), a lightweight enhancement framework for identity-preserving video generation. To improve prompt construction before generation and mitigate the above failures, AESR introduces a global agentic prompt enhancement module. This module learns model-specific prompting formats from official documentation, acquires human-centered video generation priors from human-interaction data, and accumulates test-domain identity-preserving generation experience into a reusable playbook through an agentic loop. To further repair errors in videos generated with enhanced prompts, AESR introduces a sample-level visual semantic repair module, which uses a VLM to locate erroneous video segments and design repair instructions, edits selected frames into explicit visual references, and guides a video editing model to fix local semantic or identity-related errors. We also adopt a lightweight Mixture-of-Experts selection strategy to choose reliable outputs from different generation and refinement paths. Under the official evaluation protocol of the ACM MM 2026 Identity-Preserving Video Generation Challenge, our system MIPL_Video ranked first in Track 1, demonstrating the effectiveness of AESR for practical identity-preserving video generation. The code is available at https://github.com/oceanflowlab/AESR.

[43] SPARK-SAM: Self-Prompt Adaptation with Response Knowledge for SAM in Infrared Small Target Segmentation cs.CVPDF

Aji Mao, Zhenming Peng, Bailin Mu, Tian Pu

TL;DR: 本文提出SPARK-SAM方法，旨在解决SAM模型在红外小目标分割任务中直接迁移时，空间提示与目标域掩码响应不匹配的问题。该方法通过学习目标域响应知识，并利用图像条件化的联合自提示状态来调节解码器，从而显著提升了分割性能。

Details

Motivation: 动机在于发现可提示分割模型直接迁移到自动红外小目标分割时，空间提示与目标域掩码响应之间存在严重不匹配，导致性能极低（例如在NUAA-SIRST数据集上IoU仅4.69%），因此需要一种自适应方法来弥合这一差距。

Result: 在NUAA-SIRST、NUDT-SIRST和IRSTD-1K三个基准测试上，SPARK-SAM分别达到了75.78%、86.49%和68.34%的IoU，仅增加0.726M参数，在14个重新训练的SAM变体和自适应方法中，在两个基准上排名第一，实现了SOTA性能。

Insight: 创新点在于引入了目标域响应知识学习和图像条件化的联合自提示状态，结合基准掩码监督和可靠性感知的响应指导进行训练；客观分析表明，响应适应在预测点获得可靠目标定位前就能贡献大部分最终IoU提升，且提示监督和高分辨率提示细化在所有数据集上带来一致的精度增益。

Abstract: Promptable segmentation models provide a reusable interface, but direct transfer to automatic infrared small-target segmentation (IRSTD) exposes a mismatch between spatial prompts and target-domain mask responses. In a diagnostic using target-covering loose-box prompts deterministically derived from test reference masks, the best official SAM2.1 results are only 4.69%, 1.64%, and 2.28% IoU on NUAA-SIRST, NUDT-SIRST, and IRSTD-1K. We introduce SPARK-SAM (Self-Prompt Adaptation with Response Knowledge for SAM), which learns target-domain response knowledge and conditions the decoder through an image-conditioned joint self-prompt state. Training combines benchmark-mask supervision with reliability-aware response guidance. SPARK-SAM achieves 75.78%, 86.49%, and 68.34% IoU with 0.726M additional parameters, ranking first on two benchmarks among 14 retrained SAM variants and adaptations evaluated as automatic image-to-mask methods. The staged IRSTD-1K diagnostic shows that response adaptation reaches most of the final IoU before the predicted points acquire reliable target grounding. Prompt supervision aligns the predicted prompt candidates with target locations, and frozen-weight interventions measure output sensitivity to the joint self-prompt state. Matched ablations show consistent accuracy gains from response guidance and high-resolution prompt refinement across all three datasets. Code is available at https://github.com/Sakauma/SPARK-SAM.

[44] Vis-Poison: Poisoning Visual Knowledge in Multimodal Retrieval-Augmented Generation cs.CV | cs.AIPDF

Rujin Liang, Zhongpu Chen, Yuhao Lei, Xin Miao

TL;DR: 本文提出了Vis-Poison，一种针对多模态检索增强生成（RAG）系统的新型视觉知识投毒攻击方法。该方法通过自动化多智能体方法生成视觉上合理的投毒图像，直接污染作为外部知识源的图像本身，而无需篡改文本元数据。实验表明，该方法在黑盒设置下对多种多模态RAG流程、嵌入模型和生成模型均能有效诱导错误生成。

Details

Motivation: 随着多模态RAG系统越来越多地依赖图像作为外部知识源，被污染的视觉证据会严重危害多模态大语言模型的生成。现有攻击多依赖于篡改文本元数据，本文旨在探索一种仅通过图像本身作为攻击载荷的新型视觉知识投毒攻击。

Result: Vis-Poison在两个代表性的多模态RAG流程、四种嵌入模型和六种生成模型上进行了评估。在黑盒设置下，针对包含3万条目的多模态知识库，其端到端攻击成功率在40.16%到65.40%之间。即使对于仅凭参数化知识就能正确回答的各种MLLMs，其平均成功率仍保持在60%以上。

Insight: 创新点在于首次提出并实例化了不依赖文本元数据、仅通过视觉内容本身进行投毒的攻击范式。其自动化多智能体方法用于构造视觉上合理的投毒图像，这为评估多模态RAG系统的安全性提供了一个新的、更隐蔽的攻击视角和基准。

Abstract: While multimodal retrieval-augmented generation (RAG) systems increasingly rely on images as external knowledge sources, the introduction of poisoned visual evidence can severely compromise multimodal large language model (MLLM) generation. Unlike prior attacks that rely on altering textual metadata, we introduce Vis-Poison, a novel visual knowledge poisoning attack where the poisoned image itself is the attacker-controlled payload, without manipulating captions, summaries, metadata, or other associated text. Specifically, this attack is instantiated through an automated multi-agent method that constructs visually plausible poisoned images. To assess its impact, we evaluate Vis-Poison across two representative multimodal RAG pipelines, four embedding models, and six generation models. Empirically, Vis-Poison achieves an end-to-end attack success rate of 40.16% to 65.40% against 30k-entry multimodal knowledge bases in \emph{black-box} settings. Moreover, Vis-Poison remains effective against various MLLMs that can answer correctly from parametric knowledge alone, with an average success rate above 60%. Code and data are available at https://github.com/SWUFE-DB-Group/Vis-Poison.

[45] CARD: Diagnosing Belief to Action Routing Failures in Vision Language Models cs.CV | cs.AIPDF

Souptik Kumar Majumdar, Fabian Kögel, Andreas Bulling

TL;DR: 该论文提出了跨轴路由诊断（CARD）方法，用于诊断视觉语言模型（VLMs）中信念表征到行动预测的路由失败问题。研究通过在一个名为Relay Chain的新合作网格世界基准上对开源VLM进行实验，发现模型未能将内部表征的智能体信念信息有效整合到其下一步行动预测中。

Details

Motivation: 动机在于探究视觉语言模型内部表征的智能体心理状态（如信念、知识、意图）是否以及如何被下游预测所利用，以揭示模型在利用这些丰富内部信息进行决策时可能存在的缺陷。

Result: 在提出的新基准Relay Chain上应用CARD方法进行诊断，结果表明模型存在关键的路由失败：它们未能将信念表征整合到下一步行动预测中，导致关于合作伙伴的有价值信息未被利用。

Insight: 创新点在于提出了CARD这一诊断框架，能够通过沿一个轴激活表征并测量另一个轴预测的响应，来系统性地揭示VLM内部表征与下游任务之间的脱节问题，为理解模型决策机制提供了新工具。

Abstract: Linear probes and activation steering have uncovered that vision-language models (VLMs) internally represent mental states such as agents’ beliefs, knowledge, and intentions. However, it is unclear whether and how these representations are used by downstream predictions along these axes. To close this gap, we introduce Cross-Axis Routing Diagnostic (CARD), which steers activations along one axis while measuring the response of a different axis’s prediction. Applied to open-weight VLMs on Relay Chain – a new cooperative grid-world benchmark we propose – we diagnose a critical routing failure: models fail to incorporate belief representations into their next action prediction, effectively leaving valuable information about their partners unused.

[46] MotionPhys: Detecting AI-Generated Videos via Physical Consistency of Optical-Flow Trajectories cs.CVPDF

Haojin He, Hao Tan, Zichang Tan, Ajian Liu, Jun Wan

TL;DR: MotionPhys提出了一种通过分析光流轨迹的物理一致性来检测AI生成视频的轻量级框架。该方法将稀疏运动轨迹视为物理证据，通过建模多时间尺度下的轨迹几何演化来揭示生成视频中难以察觉的运动不一致性，并将其转化为紧凑表示用于高效检测。

Details

Motivation: 现有AI视频生成模型主要优化像素或潜在空间的分布匹配，未显式强制执行惯性、连续力和轨迹几何等现实世界约束，导致生成的视频在视觉上看似合理但缺乏物理运动一致性。

Result: 在多个数据集上的实验表明，MotionPhys能有效检测生成视频中的物理不一致性，并且在不同视频生成器之间具有良好的泛化能力。

Insight: 创新点在于将稀疏运动轨迹作为物理证据而非依赖外观伪影或生成器特定痕迹，通过多时间尺度几何演化建模来捕捉传统视觉线索难以捕获的微妙运动不一致性，实现了轻量且可解释的检测。

Abstract: Modern AI video generation models can produce videos with high visual fidelity and seemingly smooth temporal transitions. However, visual realism does not necessarily imply physical motion consistency. Existing generative models mainly optimize distribution matching in pixel or latent spaces, without explicitly enforcing real-world constraints such as inertia, continuous forces, and trajectory geometry. Our experiments show that AI-generated videos remain visually plausible over short sequences of consecutive frames, yet fail to preserve physical motion consistency throughout a complete object action, resulting in systematic statistical discrepancies in their motion trajectories. Based on this observation, we introduce MotionPhys, a lightweight and interpretable framework that treats sparse motion trajectories as physical evidence rather than relying on appearance artifacts or generator-specific traces. By modeling the geometric evolution of trajectories across multiple temporal scales, MotionPhys reveals subtle motion inconsistencies that are difficult to capture with conventional visual cues and transforms them into a compact representation for efficient detection. Experiments on multiple datasets show that MotionPhys can effectively detect physical inconsistencies in generated videos and generalizes well across different video generators.

[47] CertVLA: Certified Defense against Physical Visual Attacks for Vision-Language-Action Models cs.CV | cs.AIPDF

Hui Lu, Zhijie Peng, Yuqi Lin, Zaijia Yang, Jiaming He

TL;DR: 本文提出CertVLA，一种针对视觉-语言-动作（VLA）策略在闭环控制中抵御有界补丁和纹理攻击的认证防御方法。该方法通过校准行为一致动作区域，并结合确定性覆盖掩码，确保至少有一个被检查的预测是无攻击的，从而将动作级认证扩展到完整的闭环执行过程。

Details

Motivation: 现有认证补丁防御方法主要针对离散标签，无法直接认证连续且时间相关的动作，而VLA策略容易受到局部物理扰动攻击，因此需要一种能够为闭环VLA控制提供认证防御的解决方案。

Result: 在仿真和真实世界的实验中，CertVLA展示了对抗补丁攻击的经验性和认证有效性，并在仿真中进一步验证了对纹理攻击的防御效果。

Insight: 核心创新在于提出了一个校准的行为一致动作区域概念，并通过归一化动作分歧及确定性覆盖掩码机制，实现了对连续动作序列的认证防御，其认证结果独立于补丁内容、生成方法和物理变换。

Abstract: Vision-Language-Action (VLA) policies are vulnerable to localized physical perturbations, yet existing certified patch defenses target discrete labels and cannot directly certify continuous, temporally correlated actions. We introduce CertVLA, a certified defense for closed-loop VLA control under bounded patch and texture attacks. CertVLA proposes a calibrated region of behaviorally consistent actions, while deterministic covering masks ensure that at least one checked prediction is attack-free. Specifically, CertVLA normalizes action disagreement by the benign variation of each mask pair and accepts a single-mask anchor only when it remains consistent under every second mask. It then calibrates the resulting max-min-max episode score to provide finite-sample clean coverage. Conjoining query-level decisions extends the action certificate to the complete closed-loop rollout. Furthermore, we prove that against any adaptive attacker satisfying the bounded-support threat model, every rollout certified by CertVLA executes only action chunks consistent with attack-erased clean predictions. Under dual-mask rollout correctness, this consistency certificate further guarantees task success. The certificate is independent of patch content, generation method, and physical transformation. Experiments in simulation and the real world demonstrate the empirical and certified effectiveness of CertVLA against patch attacks, with additional simulation validation on texture attacks.

[48] TRACE: Training-time Report-guided and Clinically Ordered Concept Editing cs.CV | cs.AIPDF

Wentao Yue, Tianyou Lai, Jiayu Luo, Qingyu Mao, Ziying Wang

TL;DR: 本文提出了一种名为TRACE的训练时报告引导和临床有序概念编辑框架，用于乳腺超声诊断。该框架利用结构化放射学报告作为特权概念监督，在测试时实现仅图像诊断，通过教师引导的编辑机制在恶性感知有序概念空间中细化图像衍生概念，并引入战略概念缺失训练和编辑蒸馏来应对不完整标注问题。

Details

Motivation: 解决乳腺超声诊断中现有端到端深度学习方法缺乏可解释性和鲁棒性，以及基于概念的方法通常需要完整标注或多模态输入从而限制实际应用的问题。

Result: 在多个数据集上的实验表明，与现有方法相比，TRACE实现了更优的性能和更强的跨域鲁棒性。

Insight: 创新点在于提出了一个训练时报告引导的框架，通过教师引导的编辑机制和战略概念缺失训练，在仅使用图像输入的情况下实现概念细化，并引入了链接图像、标签和结构化属性的概念增强基准BUSC。

Abstract: Breast ultrasound diagnosis relies on clinically meaningful semantic concepts, yet most deep learning methods adopt end-to-end image-to-label paradigms that lack interpretability and robustness. While concept-based approaches offer a promising alternative, they often assume complete annotations or require multimodal inputs at inference, which significantly limits their real-world applicability. To tackle these issues, we propose Training-time Report-guided and Clinically Ordered Concept Editing (TRACE), a training-time report-guided framework that leverages structured radiology reports as privileged concept supervision while enabling image-only diagnosis at test time. TRACE refines image-derived concepts through a teacher-guided editing mechanism within a malignancy-aware ordered concept space. To address incomplete annotations, we introduce Strategic Concept Missing Training (SCMT) and train an image-only self-editor via edit distillation for autonomous concept refinement. Besides, we introduce BUSC, a concept-enriched benchmark linking images, labels, and structured attributes. Experiments across multiple datasets demonstrate that TRACE achieves superior performance and improved cross-domain robustness compared to existing methods.

[49] Routing Before Looking: Query-Adaptive Evidence Acquisition for Long-form Video Understanding cs.CVPDF

Tianyue Wang, Xuying Wu, Yuxiang Ma, Ruiming Liang, Jiaxuan Kang

TL;DR: 本文提出Route2Look框架，用于长视频理解中的查询自适应证据获取。该框架采用Route-Look-Memorize循环，包含全局浏览、时序定位和语义检索三种工具，并通过路由策略动态选择工具。通过两阶段设计（从生成与检索轨迹的对比分析中蒸馏路由技能，并在推理时应用硬路由规则和继续/停止准则），在多个长视频基准测试中取得了最先进的性能，同时保持了较高的帧效率。

Details

Motivation: 解决长视频理解中查询需求与证据获取策略不匹配的问题。现有规划-感知方法通常依赖单一的生成或检索策略，难以处理多样化的查询需求。

Result: 在具有挑战性的长视频基准测试上，Route2Look实现了最先进的性能，并在不同数据集和查询类型上保持了强大的帧效率。Oracle路由分析进一步揭示了查询自适应证据获取的潜力。

Insight: 核心创新在于提出了一个轻量级、模型无关的查询自适应证据获取框架，其动态路由策略通过对比分析蒸馏获得，能灵活组合不同工具以适应多样查询，为长视频理解提供了新的高效范式。

Abstract: Long-form video understanding remains challenging for video agents due to the mismatch between query demands and evidence acquisition strategies. Although recent planning-before-perception methods outperform query-agnostic pipelines, they often rely on a single dominant strategy, either generation-based strategy or retrieval-based strategy, limiting their ability to handle diverse query demands. We propose Route2Look, a lightweight and model-agnostic framework for query-adaptive evidence acquisition in long-form video understanding. Route2Look operates in a Route-Look-Memorize loop with three tools: Global Browse for holistic context, Temporal Ground for explicit temporal cues, and Semantic Retrieve for semantic search. The core component is a routing policy that dynamically selects evidence acquisition tools based on the query. To build this policy, Route2Look adopts a two-stage design: first distilling the routing skill from differential contrastive analysis between generation-based and retrieval-based trajectories, and then applying the distilled skill with hard routing rules and continue-or-stop criteria during inference. Experiments on challenging long-video benchmarks show that Route2Look achieves state-of-the-art performance while maintaining strong frame efficiency across datasets and query types. Oracle routing analysis further reveals the potential of query-adaptive evidence acquisition for future long-form video understanding.

[50] Enhancing Localized Reasoning for Long Video Understanding via Efficient Segment-to-Video Supervision cs.CVPDF

Beibei Zhang, Chao Xu, Jun Lan, Zongyi Li, Lai Wei

TL;DR: 本文提出了一种名为Segment-to-Video Supervision (S2V)的新方法，旨在高效提升多模态大语言模型在长视频理解任务中的细粒度推理能力。该方法通过基于局部视频片段生成问答对，并将其迁移至整个视频上进行训练，从而引导模型关注易被忽略的细节，同时避免长视频中的干扰噪声。

Details

Motivation: 当前多模态大语言模型在长视频理解中面临挑战，复杂冗长的上下文包含干扰噪声，容易掩盖局部细节，导致模型产生错误答案。现有方法如强化微调框架存在训练开销大、标注成本高、奖励设计复杂等问题，而自反思和迭代感知机制则导致推理延迟高。

Result: 实验结果表明，S2V方法在多个长视频理解基准测试中持续提升了性能，不仅在长视频理解准确率上超越了通用多模态大语言模型和基于推理的方法，而且在训练和推理效率方面也表现更优。

Insight: 创新点在于提出了一种高效的片段到视频监督训练范式，仅需少量样本和简单的准确率奖励进行强化学习，即可使模型通过单次前向传播和有限输出令牌进行预测，实现了细粒度推理增强与效率的平衡。

Abstract: Though Multimodal Large Language Models (MLLMs) have shown impressive potential in video understanding, long video understanding (LVU) remains challenging since distracting noise in complex and lengthy contexts can obscure localized details, misleading MLLMs to produce incorrect answers. Recent works mitigate these issues by incentivizing deep reasoning to include relevant evidence. However, these methods have two main problems: First, the reinforcement fine-tuning framework (RFT) they leveraged incurs substantial training overheads, including high annotation costs and complicated reward designs. Second, the self-reflective and iterative-perception mechanism in some methods causes lengthy outputs and high inference latency. To alleviate these problems, we propose a novel Segment-to-Video Supervision} method (S2V) to efficiently enhance fine-grained reasoning in LVU. Specifically, we generate question answer pairs (VQA) based on localized segments, and then transfer these segment-based VQA back to the whole video for training. Due to focusing on short segments, segment-based VQA can naturally notice details which tend to be overlooked from a whole-video perspective. Training on such data can enforce MLLMs to correctly associate fine-grained details with QA while avoiding distracting noise in the whole video. The S2V training involves just reinforcement learning (RL) with a simple accuracy reward based on only 10K VQA samples and the resulting S2V model predicts answer using a single forward pass with limited output tokens. Experimental results demonstrate that S2V can consistently improve LVU performance across multiple LVU benchmarks, outperforming both general MLLMs and reasoning-based methods not only in LVU accuracy but also in training and inference efficiency.

[51] Identify, Locate, Link: End-to-End Key-Value Extraction from Document Images cs.CV | cs.CLPDF

A. Said Gurbuz, Ahmed Nassar, Christoph Auer, Maksym Lysak, Lucas Morin

TL;DR: 本文提出了一种端到端的键值对提取方法，通过微调紧凑型视觉语言模型SmolDocling，直接从文档图像中联合完成键值对的识别、定位和关联，无需OCR预处理。该方法扩展了DocTags标签体系以支持多对多关系，并设计了数据增强和布局感知评估框架。在多个数据集上，该模型在保持较小参数量和较快推理速度的同时，优于更大的零样本VLM基线。

Details

Motivation: 传统文档处理流程将OCR与下游信息提取模型级联，导致多阶段错误传播。本文旨在通过端到端模型直接从图像中提取结构化键值对，以解决这一问题。

Result: 在FUNSD、XFUND和一个大规模私有数据集上，该模型在布局感知评估下优于更大的零样本VLM基线，其参数量仅为Qwen2.5-VL（7B）的1/27，推理速度快5倍以上。

Insight: 创新点包括：1）端到端联合建模识别、定位和关联任务；2）扩展DocTags以统一序列表示多对多关系；3）结合合成填充和图裁剪的数据增强策略；4）引入包含空间框验证的布局感知评估框架。模型紧凑高效，为文档理解提供了新范式。

Abstract: Document processing pipelines traditionally cascade optical character recognition (OCR) engines with downstream models for structured information extraction, leading to multi-stage error propagation. We fine-tune SmolDocling, a compact 256M-parameter vision-language model (VLM), to perform end-to-end key-value extraction directly from document images, jointly solving identification, localization, and association in a single pass without OCR preprocessing. We extend DocTags with specialized key, value, region, and link tags, enabling many-to-many relationships in a unified output sequence. To address data limitations, we design an augmentation pipeline combining synthetic form filling and graph-based crops that preserve complete key-value subgraphs. We further introduce a layout-aware evaluation framework extending text matching with spatial bounding box verification. On FUNSD, XFUND, and a large-scale private dataset, our model outperforms larger zero-shot VLM baselines under layout-aware evaluation, while being 27 times smaller than Qwen2.5-VL (7B) and over 5 times faster at inference. The model weights will be released publicly after publication.

Meda Lazar, Sourab Sridhar, Shashwata Gupta, Alexandra Tripcea, Varun Ravi

TL;DR: 本文提出了一种用于自动驾驶的鲁棒、长距离、跨区域交通标志检测框架，通过融合相机和激光雷达数据，利用强度感知可变形融合模块对齐两种模态的特征，并引入双运动模型跟踪器处理车辆接近时的非线性透视畸变，同时开发了语义属性分类管道以提供下游规划所需的情境信息。

Details

Motivation: 解决现有基于视觉的交通标志检测方法在跨区域泛化性差、远距离小目标检测性能下降以及车辆接近标志时非线性透视畸变导致跟踪脆弱性这三个根本性限制。

Result: 在涵盖60多个国家、超过2500小时驾驶数据的自建数据集上进行广泛评估，所提方法在221,068个评估序列中实现了0.49%的目标漏检率，证明了其在商用级自动驾驶系统中具有全球可泛化的交通标志感知能力。

Insight: 创新点包括：1) 强度感知可变形融合模块，利用激光雷达的几何不变性而非区域特定的视觉外观来锚定检测；2) 双运动模型跟踪器，显式建模非线性透视变换以提升时间一致性；3) 语义属性分类管道，为下游任务提供可操作的上下文信息。从客观角度看，多模态融合策略有效应对了单一视觉模态的固有局限。

Abstract: Reliable traffic sign detection is a prerequisite for the global deployment of autonomous driving systems, where regulatory compliance and road safety depend on perceiving signs correctly across regions, ranges, and weather conditions. Despite recent progress, vision-based methods continue to face three fundamental limitations: poor cross-regional generalization due to high diversity across countries, degraded performance on small-object detection at long ranges (traffic signs occupy as little as $10{\times}10$ pixels at 200m), and fragile temporal tracking under the strongly non-linear perspective distortion that occurs as a vehicle approaches a sign. In this paper, we address the problem of robust, long-range, region-agnostic traffic sign perception by combining camera and Light Detection and Ranging (LiDAR) sensing. We present a multi-modal detection framework whose Intensity-Aware Deformable Fusion module aligns retro-reflective LiDAR cues with camera features, anchoring detection on geometric invariants rather than region-specific visual appearance. We further introduce a dual motion-model tracker that explicitly accounts for non-linear perspective transformations during vehicle approach, substantially improving temporal consistency over linear motion assumptions. Additionally, we develop a semantic attribute classification pipeline that estimates occlusion level, readability, sign embeddedness, and road relevance, providing actionable context to downstream planning. Extensive evaluation on our dataset, spanning 60+ countries and 2,500+ hours of driving data, shows that the proposed pipeline achieves an Object Miss Ratio (OMR) of 0.49% across 221,068 evaluation sequences, demonstrating globally generalizable traffic sign perception in commercial-grade autonomous driving systems.

[53] EviRank: Structured Relevance Evidence for Multimodal Image Re-ranking cs.CV | cs.LGPDF

Enjun Du, Siyi Liu, Zirong Chen, Xinyu Zuo, Jinwen Luo

TL;DR: 本文提出了EviRank，一种用于多模态图像重排序的结构化相关性证据框架。它将查询解析为包含六个语义槽（如实体、属性、关系）的统一证据包，并将重排序任务转化为证据条件验证问题，结合了确定性规则评分和基于证据的列表比较。该方法在多个基准测试上取得了最先进的性能，并可蒸馏出轻量级学生模型。

Details

Motivation: 解决现实世界中多模态、组合式图像搜索查询（如“找这件粉色的衬衫”）的挑战。现有重排序方法要么将复杂的相关性压缩为不透明的嵌入，要么依赖自由形式的思维链，容易遗漏或产生细粒度约束的幻觉。

Result: 在涵盖文本到图像、图像到图像和组合图像检索的五个基准测试中，EviRank实现了最先进的性能。其蒸馏出的学生模型以显著更低的成本保留了教师模型90%以上的能力。

Insight: 创新点在于将多模态图像重排序重新定义为语义约束满足问题，并引入结构化证据包（包含类型化标准和需求标签）作为统一表示。该方法提供了一种无需训练、可解释的验证流程，并可利用显式证据作为结构化监督来蒸馏轻量模型。

Abstract: Real-world image search queries are multimodal and compositional: ``find this shirt in pink’’ specifies an entity to retain, an attribute to modify, and context to ignore. Yet existing re-rankers either compress such multifaceted relevance into an opaque embedding or rely on free-form chain-of-thought that easily omits or hallucinates fine-grained constraints. Drawing on rubric- and checklist-based evaluation from NLP, we recast multimodal image re-ranking as a semantic constraint satisfaction problem and propose EviRank, which parses any query - text-only, image-only, or composed - into a unified evidence package: typed criteria across six semantic slots (e.g., entities, attributes, relations), each labelled required, forbidden, or ignorable. Re-ranking then reduces to evidence-conditioned verification, combining deterministic rubric scoring and evidence-grounded listwise comparison in a single training-free procedure. The explicit evidence can further serve as structured supervision for optionally distilling a lightweight student. Across five benchmarks spanning text-to-image, image-to-image, and composed image retrieval, EviRank achieves state-of-the-art performance, and the distilled student preserves over 90% of the teacher’s capability at substantially lower cost.

[54] A Collaborative Multi-Modality Interaction for VLA-based End-to-End Autonomous Driving cs.CV | cs.ROPDF

Jingtao Sun, Xiaohai He, Yike Zhang, Dong Huang, Yaonan Wang

TL;DR: 本文提出了一种基于视觉-语言-动作（VLA）模型的鲁棒端到端自动驾驶系统，通过结合多模态交互与多轨迹规划优化，旨在实现更可靠、可解释且安全的驾驶决策。核心创新包括亲和力引导的最优传输用于主-辅模态双向交互、分布一致模态转移用于异构模态分布对齐与跨模态交互，以及面向感知的轨迹优化多模态多轨迹规划。

Details

Motivation: 现有VLA模型大多将端到端自动驾驶视为视觉问答任务，导致决策推理不可靠且可解释性差；同时，它们未能有效建立异构传感器间的多模态交互，限制了在长尾驾驶场景下的鲁棒感知和可靠推理。

Result: 在开环和闭环数据集上的实验结果表明，该系统在安全长时程驾驶推理和道路场景感知方面优于现有驾驶系统，验证了其多模态交互与多轨迹规划优化方法对可扩展VLA系统的有效性。

Insight: 创新点在于将多模态交互与多轨迹规划优化深度耦合，通过亲和力引导最优传输和分布一致模态转移实现异构模态的有效对齐与交互，并利用面向感知的轨迹细化提升对长尾场景的决策鲁棒性，为VLA模型提供了更可靠的感知-决策统一框架。

Abstract: Vision-Language-Action (VLA) models have emerged as a powerful paradigm for end-to-end autonomous driving by jointly integrating perception, reasoning, and decision making within a unified multimodal framework. However, most existing VLA models formulate end-to-end autonomous driving as a visual question answering task, leading to unreliable and less interpretable decision reasoning. In addition, they fail to establish effective multi-modal interaction across heterogeneous sensors, thereby limiting robust scene perception and reliable driving reasoning in long-tail driving scenarios. To this end, we propose a robust VLA-based end-to-end autonomous driving system that combines multi-modality interaction with multi-trajectory planning and optimization, enabling more reliable, interpretable, and safer driving decisions. Our method comprises three core components: (1) Affinity-Guided Optimal Transport for main-auxiliary modality two-way interaction; (2) Distribution-Consistent Modality Transfer for heterogeneous modality distribution transfer and cross-modal interaction; (3) Multi-modal Multi-Trajectory Planning along with Perception-Oriented Trajectory Refinement for better driving decisions to long-tail driving scenarios. Experimental results in open-loop and closed-loop datasets demonstrate improvements in safety long-horizon driving reasoning and road scene perception over existing driving systems, highlighting the ability of our mutli-modality interaction and multi-trajectory planning and optimization for scalable VLA-based systems.

[55] EmotionDialogCN: A Spontaneous Multimodal Dataset for Mandarin Emotional Dialogue cs.CVPDF

Yi Zheng, Yifan Xu, Yan Zhou, Hejia Chen, Chunyu Qiang

TL;DR: 本文介绍了EmotionDialogCN，一个大规模的中文多模态情感对话数据集，包含21,880个由专业演员在20个日常场景中进行的对话会话，覆盖18种情感类别，总时长超过400小时。该数据集通过新颖的采集框架最小化设备干扰，以捕捉真实的面部表情和情感表达，其情感分布与真实人类情感统计的偏差仅为0.64，远优于先前数据集的5.65。

Details

Motivation: 现有情感对话数据集存在情感标注不足、情感多样性差和规模小的问题，无法充分捕捉真实面对面交流中的丰富情感和社会线索，因此需要构建一个大规模、高质量的多模态情感对话数据集来推动相关研究。

Result: EmotionDialogCN在情感分布上与真实人类情感统计的偏差为0.64（先前数据集为5.65），并保持了一致的拍摄主体构图（52-59%的帧占比）。在声学、词汇和视觉模态上均表现出稳定的单模态和多模态性能，融合结果进一步强调了强大的多模态对齐和跨模态互补性。

Insight: 创新点在于提出了一个最小化设备干扰的数据采集框架，以促进自然和细腻的情感表达，从而构建了目前最大、最全面的中文多模态情感对话数据集。从客观角度看，该数据集在情感真实性和多模态对齐方面为情感计算和对话系统研究提供了高质量基准。

Abstract: Face-to-face audiovisual interaction is central to human communication, conveying rich emotional and social cues. However, existing multimodal dialogue datasets remain limited by inadequate emotion annotations, poor emotional diversity, and small scale. We introduce EmotionDialogCN, a large-scale audiovisual-emotional dataset designed to capture authentic face-to-face communication. It contains 21,880 dialogue sessions performed by 119 professional actors across 20 everyday scenarios, covering 18 emotion categories with over 400 hours of recordings, the largest and most comprehensive dataset of its kind. A novel data collection framework minimizes equipment interference, enabling natural and nuanced emotional expressions. EmotionDialogCN achieves an emotion distribution deviation of 0.64 from real human emotion statistics (versus 5.65 for prior datasets) and consistent subject framing (52-59% frame occupancy). Together, these properties translate into stable unimodal and multimodal performance across acoustic, lexical, and visual modalities, with fusion results further underscoring strong multimodal alignment and cross-modal complementarity.

[56] InfinityEdit: Infinite Video Editing with a Lightweight Edit-Ignition Adapter cs.CVPDF

Yunze Tong, Mushui Liu, Canyu Zhao, Shiyi Zhang, Didi Zhu

TL;DR: 本文提出了InfinityEdit，一种轻量级编辑适配器，用于实现无限视频编辑任务。该方法通过三个注意力模块（历史交叉注意力、时序因果自注意力和编辑交叉注意力）将编辑指令注入到流式视频生成器中，使其能够在持续的视频流中应用编辑请求，同时保持生成质量的稳定性。

Details

Motivation: 现有基于指令的视频编辑方法大多依赖于固定时间跨度的逐帧对齐编辑假设，无法处理开放式的视频流（如实时游戏风格化或持续镜头的摄像机移动）。本文旨在解决无限视频编辑问题，即模型需要根据前序片段和编辑请求生成下一个片段，并在编辑指令不断到达时持续应用编辑。

Result: 大量实验表明，InfinityEdit能够在每个编辑指令下忠实延续视频流，并在无限编辑序列中保持稳定的生成质量。

Insight: 创新点包括：1) 设计了针对无限视频编辑任务的数据收集流程；2) 提出了轻量级编辑适配器架构，通过三个注意力模块实现编辑指令的注入和时序一致性的保持；3) 采用推理时仅在编辑请求到达的块激活适配器、后续块使用原始模型并重置锚定帧的方案，既应用了编辑又保留了原始模型的无限生成能力。

Abstract: With large pretrained models, existing methods have effectively improved instruction-based video editing. However, most of them rely on an in-place editing assumption. They align the edited video with the given source clip frame by frame over a fixed time span. This pattern fails for open-ended streams, e.g., restyling a live game or applying a camera move to an ongoing shot. In such cases, edits must extend to future frames as they arrive, rather than be applied to a static input clip. In this paper, we study this setting and name it infinite video editing: given a preceding segment and an edit request, a model must generate the next segment that continues the stream while applying the requested edit. This process repeats as an unbounded sequence of edit instructions arrives. This task brings two challenges: the edit must be a faithful continuation rather than a frame-wise rewrite, and generation quality must remain stable as edits accumulate. To address them, we first design a data-collection pipeline for infinite video editing. Based on the collected data, we propose InfinityEdit, a lightweight edit adapter that equips a streaming video generator with unbounded editing ability. The adapter contains three attention modules. History cross-attention guides the denoising frames using the input frames. Temporal causal self-attention keeps temporal cues flowing only from earlier frames to later ones. Edit cross-attention injects the edit request into generation. During inference, the adapter is activated only in the chunk where an edit request arrives. Subsequent chunks are generated by the original model with a reset anchor frame. This scheme applies the edit while preserving the original model’s infinite generation ability. Extensive experiments show that InfinityEdit faithfully continues the stream under each edit, and stays stable over unbounded edit sequences.

[57] Semantically Compatible Knowledge Distillation for Cross-Domain Object Detection with Vision Foundation Models cs.CVPDF

Qifeng Zhang, Ting Xiang, Zeyuan Bai, Changjian Chen

TL;DR: 本文提出了一种名为语义定位增强教师（SLE-T）的框架，用于解决基于视觉基础模型（VFMs）的跨域目标检测（DAOD）中存在的语义不兼容问题。该方法通过一个轻量级的SLE适配器，将预训练的局部纹理先验知识注入DINOv2模型，生成与检测器学生网络在空间和语义上兼容的密集特征表示，并通过伪标签学习或特征对齐进行知识蒸馏。

Details

Motivation: 现有基于视觉基础模型（VFMs）的跨域目标检测方法，忽视了教师与学生特征图之间的空间尺度差异，导致语义不兼容，削弱了特征对齐和伪标签学习的效果。同时，域偏移会导致源域训练的VFM教师模型漏检目标域对象，限制了伪标签的质量。

Result: 在三个DAOD基准测试上的广泛实验表明，该方法取得了最先进的性能。消融研究证实了师生语义兼容性的重要性。值得注意的是，使用DINOv2-B的SLE-T模型，在仅使用DINOv2-G约四分之一训练时间和更少GPU内存的情况下，生成了具有竞争力或更优的伪标签。

Insight: 核心创新点是提出了一个轻量级的SLE适配器，通过注入局部纹理先验和特征重表述，解决了VFM教师与学生检测器之间的语义和空间不兼容问题，实现了高效、高性能的跨域知识蒸馏，为在有限计算资源下利用大型VFM提供了可行方案。

Abstract: Vision foundation models (VFMs) offer strong generalization capabilities for domain-adaptive object detection (DAOD). However, existing VFM-based methods overlook the spatial-scale discrepancy between teacher and student feature maps, resulting in semantic incompatibility that weakens both feature alignment and pseudo-label learning. Moreover, domain shift can cause source-trained VFM teachers to miss target-domain objects, limiting the quality of their pseudo-labels. To address these issues, we propose the Semantic Localization-Enhanced Teacher (SLE-T), a semantically compatible knowledge-distillation framework built around a lightweight SLE Adapter for DINOv2. SLE Adapter injects pretrained local-texture priors into DINOv2 to improve cross-domain recognition and reformulates its features into dense representations that are spatially and semantically compatible with the student detector. SLE-T transfers the resulting teacher knowledge through either pseudo-label learning or feature alignment. We instantiate SLE-T with DINOv2-B and DINOv2-L (the ViT-B and ViT-L variants) and compare them with the larger DINOv2-G teacher. Extensive experiments on three DAOD benchmarks demonstrate that our method achieves state-of-the-art performance, and ablation studies confirm the importance of teacher-student semantic compatibility. Notably, SLE-T with DINOv2-B produces competitive or superior pseudo-labels using approximately one-quarter of the training time of DINOv2-G and substantially less GPU memory, demonstrating efficient VFM knowledge transfer under limited computational resources.

[58] OccluRank: Controllable Occlusion-Aware Layout-to-Image Generation by Adding Just an Ordinal Rank cs.CVPDF

Wenyang Hong, Yuan Wang, Yanbin Hao, Lanqing Xue, Ke Wang

TL;DR: 本文提出OccluRank，一种简单可控的遮挡感知布局到图像生成框架。该方法通过为每个边界框添加一个序数等级来编码用户指定的遮挡顺序，并引入顺序感知实例交互模块在聚合前联合更新实例表示，从而无需额外几何输入或复杂推理即可实现遮挡关系的控制。

Details

Motivation: 现有布局到图像生成方法仅通过边界框控制实例位置，无法表示遮挡顺序，且可能依赖额外几何条件、复杂推理过程或未显式建模遮挡依赖的交互。

Result: 在构建的OccluLayout-Bench评估基准上，实验表明OccluRank在保持目标实例、遵循指定布局和实现期望遮挡关系方面更可靠，同时保持了可比的属性一致性和整体图像质量（FID）。

Insight: 创新点在于仅使用轻量级的序数等级条件来编码遮挡顺序，并通过顺序感知实例交互模块显式建模遮挡依赖的交互；同时构建了直接从已知场景几何生成遮挡顺序和amodal标注的合成数据集，避免了依赖辅助预测模型从部分遮挡图像进行估计。

Abstract: Layout-to-image generation enables explicit spatial control through bounding-box layouts, yet bounding boxes specify only instance locations and cannot represent their occlusion order. Existing methods may rely on additional geometric conditions, employ complex inference procedures, or aggregate independently constructed instance representations without explicitly modeling their occlusion-dependent interactions. We propose OccluRank, a simple and controllable occlusion-aware layout-to-image framework that augments each bounding box with only one ordinal rank. OccluRank encodes the user-specified occlusion order through lightweight rank-based conditioning and introduces an Order-aware Instance Interaction (OII) module to jointly update rank-conditioned instance representations before aggregation. This allows the specified order to guide information exchange among occluding instances without additional geometric inputs or specialized inference-time optimization. We further construct OccluLayout, a synthetic training dataset whose occlusion order and amodal annotations are derived directly from known scene geometry rather than estimated from partially occluded images using auxiliary prediction models. For comprehensive evaluation, we introduce OccluLayout-Bench, which uses multiple multimodal large language model evaluators to assess instance presence, spatial layout, attributes, and occlusion order, together with FID for overall image quality. Experiments show that OccluRank more reliably preserves target instances, follows specified layouts, and realizes desired occlusion relationships while maintaining comparable attribute consistency and overall image quality.

[59] GAP-SAM: A Global Artifact Prior for Generalizable AI-Generated Image Manipulation Localization cs.CVPDF

Haozhen Yan, Siyuan Shan, Zijian Yu, Youqi Wang, Yan Hong

TL;DR: 该论文提出了GAP-SAM方法，旨在解决AI生成图像篡改定位任务中，模型因像素级监督与数据集特定掩码几何及语义边界纠缠而导致的跨域（OOD）性能不佳问题。通过构建COCO-ControlNet数据集来对齐语义和几何信息，并设计了一个全局伪影标记（global artifact token）注入SAM3特征金字塔，以抑制语义边界捷径并提升定位能力。

Details

Motivation: 现有AI生成图像篡改定位方法在跨域（OOD）场景下性能不足，主要原因是像素级监督会混淆取证证据与数据集特定的掩码几何形状及语义边界。此外，还观察到微调的分割模型存在‘边界粘连’现象，即预测会贴合语义对象轮廓而非真实的篡改边界。

Result: 在六个数据集上的实验表明，GAP-SAM平均Pixel-F1达到79.8，比之前最强的方法高出12.6个百分点。在JPEG压缩、高斯模糊和尺寸调整等各种退化严重程度下，GAP-SAM均取得了最佳性能。

Insight: 核心创新点在于提出了一个全局伪影先验（Global Artifact Prior），通过将图像及其冻结的VAE重建编码为一个全局伪影标记，并利用零门控FiLM将其注入SAM3的特征金字塔，从而在不指定空间区域的情况下调制密集解码，有效抑制了语义边界捷径，提升了定位的泛化能力。该方法揭示了VAE重建伪影与扩散修复伪影之间的迁移性差异，并针对性地设计了解决方案。

Abstract: AI-generated image manipulation localization identifies edited pixels, but its OOD performance lags behind image-level detection partly because pixel supervision entangles forensic evidence with dataset-specific mask geometry and semantic boundaries. Extending image-level distribution alignment to localization, we construct COCO-ControlNet with source-image Canny edges and depth maps to align semantics and geometry, improving OOD performance across multiple localizers. Yet tighter Mask-VAE Reconstruction Alignment (Mask-VAE) underperforms COCO-ControlNet, showing that VAE reconstruction artifacts transfer poorly to local diffusion-inpainting artifacts. We also identify \emph{boundary adhesion}, where fine-tuned segmentation models snap predictions to semantic object contours rather than true manipulation boundaries. These findings motivate GAP-SAM, which encodes an image and its frozen VAE reconstruction into a global artifact token and injects it into SAM3’s feature pyramid via zero-gated FiLM before pixel decoding. Without prescribing a spatial region, this token modulates dense decoding to preserve localization while suppressing semantic-boundary shortcuts. Across six datasets, GAP-SAM averages 79.8 Pixel-F1, outperforming the strongest prior method by 12.6 points. It also performs best at every tested severity of JPEG compression, Gaussian blur, and resizing.

[60] SuppreSensing: Expert-Guided Feature Recalibration and Discrepancy Augmentation for Multimodal Object Detection cs.CVPDF

Xin Wu, Zhenyu Gao, Qiankun Zhang, Shaoyong Guo

TL;DR: 本文提出SuppreSensing方法，用于解决遥感多模态目标检测中的语义异构和模态特定噪声干扰问题。该方法将多模态融合重新表述为一个选择性协作过程，通过专家驱动的特征重校准、模态特定属性增强和定制化特征纯化模块，共同建模共享信息和模态特定线索。在DroneVehicle和VEDAI数据集上取得了最先进的检测性能，并在自然场景数据集上验证了其鲁棒性和泛化能力。

Details

Motivation: 解决遥感多模态目标检测中因语义异构和模态特定噪声干扰导致的融合挑战，旨在更有效地联合建模共享与特定信息。

Result: 在DroneVehicle和VEDAI数据集上实现了最先进的检测性能；在FLIR和LLVIP等自然场景数据集上的跨域评估进一步验证了其卓越的鲁棒性和泛化能力。

Insight: 创新点在于将多模态融合重新定义为选择性协作过程，并引入了基于多专家选择的特征重校准、双向差异模式建模的属性增强以及仿体检范式的迭代特征纯化机制，以缓解对称性陷阱和跨模态异构问题。

Abstract: Multimodal object detection in remote sensing faces challenges due to semantic heterogeneity and modality-specific noise interference. To this end, we propose SuppreSensing, which reformulates multimodal fusion as a selective collaboration process that jointly models shared information and modality-specific cues. SuppreSensing first designs an Expert-driven Multimodal Feature Recalibration (EMFR) module, which reformulates shared-consensus extraction as an input-adaptive multi-expert selection process to alleviate the symmetry trap in multimodal fusion. Complementing this, a modality-specific attribute augmentation strategy is employed to enhance specific modality features by modeling bidirectional discrepancy patterns, mitigating cross-modal heterogeneity. Furthermore, we propose an Expert-driven Customized Feature Purification (ECFP) module based on a “specialized inspection-comprehensive analysis-diagnostic update” physical examination paradigm to iteratively filter redundancies and reinforce task-relevant semantics. Extensive experiments on the DroneVehicle and VEDAI datasets demonstrate that SuppreSensing achieves state-of-the-art detection performance. Cross-domain evaluations on natural scene datasets (FLIR and LLVIP) further validate its superior robustness and generalization capability across diverse environmental conditions.

[61] Kinematic Knowledge Maps for Pattern Alignment: Structured Latent Representational Learning in Multimodal Gait Analysis cs.CVPDF

Chen Dong, He Zonglin, Cheung Kenneth M. C

TL;DR: 本文提出了一种名为ScoliDetect的可解释框架，用于从单目步态视频中筛查青少年特发性脊柱侧凸。该框架的核心是运动学知识图谱（KKM），这是一种固定索引的结构化表示，编码了绝对运动、自骨架配置和关节间信号相关性等步态特征，并辅以基于模板的运动学文本。通过双向交叉注意力与潜在瓶颈聚合整合视频、KKM和文本，在外部筛查队列中显示出优于单模态模型和后期拼接的性能。

Details

Motivation: 多模态临床AI面临输入弱对齐和缺乏领域特定可解释表示的挑战，尤其是在从密集视频流、结构化时间序列和基于模板的运动学文本中学习时。本文旨在解决这些问题，为脊柱侧凸筛查提供一个可解释的框架。

Result: 在一个多中心队列（排除后n=1,858）的外部筛查队列中，预设的监督消融实验表明，KKM介导的多模态融合优于单模态模型和后期拼接。在分阶段训练协议下，架构选择后应用三模态对比预训练作为表示初始化，将外部ROC-AUC从0.961提升至0.972。

Insight: 主要创新点在于提出了运动学知识图谱（KKM）作为一种固定索引的结构化潜在表示，它实现了锚定参考的多模态融合和因子级解释。客观来看，将明确的结构拓扑嵌入潜在空间，显著增强了多模态模式分析系统的泛化能力和可解释性，为临床AI提供了新的结构化表征学习思路。

Abstract: Multimodal clinical AI is limited by weakly aligned inputs and the absence of domain-specific interpretable representations, particularly when learning from dense video stream, structured time-series, and template-based kinematic text. Here we present ScoliDetect, an explainable framework for adolescent idiopathic scoliosis screening from monocular gait video, built around a kinematic knowledge map (KKM) and complementary template-based kinematic text derived from per-sequence pose statics. KKM is a fixed-index structured representation that encodes gait features across absolute motion, self-skeleton configuration and joint-joint signal correlation, providing anchor-referenced multimodal fusion and factor-level interpretation. We integrate video, KKM, and template-based kinematic text through bidirectional cross-attention with latent-bottleneck aggregation. In a multicenter cohort (n = 1,858 after exclusions), prespecified supervised ablations on an external screening cohort show that KKM-mediated multimodal fusion outperforms unimodal models and late concatenation. Under a staged training protocol, trimodal contrastive pretraining is applied after architecture selection as representation initialization, improving external ROC-AUC from 0.961 to 0.972. Furthermore, the structured nature of the KKM provides inherent, factor-level attributions mapped directly to specific kinematic phases and skeletal indices, offering verifiable interpretability. The results demonstrate that embedding explicit structural topologies into latent spaces significantly enhances both the generalization and explainability of multimodal pattern analysis systems.

[62] MigrationNarrate: A Dataset for Detection of Migration Narratives in YouTube Videos cs.CV | cs.CL | cs.CYPDF

Fatima Haouari, Carolina Scarton, Kalina Bontcheva

TL;DR: 本文介绍了MigrationNarrate，这是首个用于检测英国移民叙事的多模态数据集，包含1,115个YouTube视频转录本，并采用两级分类法标注了12个超级叙事和53个叙事标签。论文详细描述了数据集的设计、收集和标注过程，并提供了基于预训练编码器和大型语言模型的基准测试结果。

Details

Motivation: 现有研究缺乏针对移民叙事的专门标注数据集，且公共传播已转向以视频为中心的平台，但视频中的叙事检测仍未被充分探索，因此需要构建一个多模态数据集来填补这一空白。

Result: 论文提供了使用预训练编码器模型和开源/闭源大型语言模型组合的基准测试结果，但摘要中未具体说明定量性能指标或是否达到SOTA水平。

Insight: 创新点在于构建了首个专注于移民主题的多模态视频叙事检测数据集，并引入了两级叙事分类法，为理解视频平台上的公共话语提供了新的数据资源和分析框架。

Abstract: Narratives are central to how social communication is framed, making their detection critical for understanding and analysing public discourse. Prior work has explored narrative detection and extraction across diverse domains; however, migration narratives remain significantly understudied, primarily due to the absence of dedicated annotated datasets. Furthermore, public communication has recently shifted towards video-centric platforms, where narratives are conveyed through multimodal signals and consumed at scale. Despite this shift, narratives in videos remain largely unexplored. To bridge these gaps, we introduce MigrationNarrate, the first multimodal dataset for detection of migration narratives in the UK, consisting of 1,115 YouTube video transcripts annotated using a two-level taxonomy of 12 migration super-narratives and 53 narrative labels. This paper details the dataset design, collection, and annotations; together with benchmark results using a combination of pre-trained encoder models and both open- and closed-source Large Language Models. Finally, a thorough error analysis offers insights for future work.

[63] WA-JEPA: Rethinking the Video JEPA Paradigm for World-Action Modeling in Autonomous Driving cs.CV | cs.AIPDF

Xinlin Wang, Yujiao Xiang, Yuheng Zhou, Jingqi Wang, Minqing Huang

TL;DR: 本文提出WA-JEPA，一种基于视频联合嵌入预测架构（V-JEPA）的世界-动作模型，专为自动驾驶规划任务设计。该方法通过混合未来掩码预训练和条件流匹配技术，改进了传统V-JEPA的随机掩码补全和确定性回归范式，实现了对未来场景和自车轨迹的联合预测。在nuPlan数据集上预训练并在NAVSIM上微调后，模型在多个自动驾驶基准测试中取得了最先进的性能。

Details

Motivation: 传统V-JEPA采用随机掩码补全和确定性回归，不适合需要与动作紧密耦合的未来导向预测的自动驾驶规划任务。本文旨在重新思考V-JEPA范式，构建一个原生支持世界-动作建模的架构，以更好地服务于自动驾驶规划。

Result: 在NAVSIM-v2基准测试上达到91.7 EPDMS，分别超过最强的端到端基线和世界-动作基线1.6和1.3 EPDMS。在闭环HUGSIM基准测试上，在相同评估协议下，无需HUGSIM特定微调即获得了最佳的HD-Score（0.4462）。

Insight: 创新点包括：1）用混合未来掩码预训练取代随机时空掩码，使模型能从观测上下文推断未来潜在特征；2）用潜在未来的条件流匹配取代确定性回归，显著提升生成合理未来潜在特征的能力；3）提出联合未来-动作预测器，在统一的时空潜在空间中同时对未来场景令牌和自车轨迹去噪，使动作监督能直接塑造与规划相关的世界表示。这为自动驾驶规划提供了一个强大且可扩展的V-JEPA原生世界-动作建模范式。

Abstract: Video Joint Embedding Predictive Architecture (V-JEPA) learns powerful spatiotemporal representations from video through self-supervised latent feature prediction. However, V-JEPA is built around random-mask completion and deterministic regression, making it fundamentally ill-suited for autonomous driving planning that demands future-directed prediction tightly coupled with action. To address this, we rethink the V-JEPA paradigm and present WA-JEPA, a V-JEPA-native world-action model designed for autonomous driving planning. Instead of random spatiotemporal masking, WA-JEPA employs hybrid future-masked pre-training, where the model infers future latents from observed context. Departing from deterministic regression, we recast future prediction as conditional flow matching over latent futures, which substantially improves the model’s ability to generate plausible future latents for downstream planning. Finally, a joint future-action predictor is proposed to denoise future scene tokens and ego trajectories together in a unified spatiotemporal latent space, allowing action supervision to directly shape planning-relevant world representations. Pre-trained on nuPlan videos and fine-tuned on NAVSIM, WA-JEPA reaches 91.7 EPDMS on NAVSIM-v2, surpassing the strongest end-to-end and world-action baselines by 1.6 and 1.3 EPDMS, and, without HUGSIM-specific fine-tuning, attains the best HD-Score of 0.4462 on the closed-loop HUGSIM benchmark under the same evaluation protocol. These results validate V-JEPA-native world-action modeling as a powerful and scalable paradigm for autonomous driving planning. Code is available at https://github.com/AFARI-Research/WA-JEPA.

[64] Latent Ordinal Evidence, Misaligned Outputs: Inference-Time Ordinal Lens Alignment for Multimodal LLMs cs.CVPDF

Haiming Li, Yingsheng Liu, Jingmin Zhu, Siyuan Yan, Xieji Li

TL;DR: 本文研究了多模态大语言模型（MLLMs）在处理年龄估计、图像质量评估等序数回归任务时，其内部表征与最终数字标记输出之间的对齐问题。研究发现，虽然模型的隐藏状态蕴含了清晰的序数信息，但通过词嵌入矩阵解码后的原生数字标记输出却严重丢失了这种结构。为此，作者提出了一种名为序数透镜对齐（OLA）的推理时方法，该方法在冻结主干模型参数的情况下，训练轻量级透镜来校正数字标记的logits，从而显著提升序数任务的性能。

Details

Motivation: 解决MLLMs在处理序数回归任务时，其内部蕴含的序数证据与最终输出的数字标记之间存在的严重不对齐问题，即模型未能可靠地将内部的序数信息转化为有序的数字输出。

Result: 在四个序数基准测试和四个MLLM骨干网络上，OLA方法在大多数设置下超越了当前最先进的LoRA微调基线OrderChain，同时保持了主干模型冻结；在大多数情况下也超过了判别式序数基线，并且在所有设置中都优于离线透镜方法。

Insight: 创新点在于揭示了MLLMs内部表征与输出层之间存在“序数信息过滤”现象，并提出了一种轻量、推理时的透镜对齐方法（OLA），通过校正输出层的logits来高效利用模型已有的序数知识，无需微调整个模型，实现了性能的显著提升。

Abstract: Multimodal LLMs apply the language model interface to visual inputs, where ordinal regression tasks such as age estimation, image quality assessment, and disease grading require autoregressive decisions over ordered class labels. We ask whether MLLMs reliably convert internal ordinal evidence into ordered digit-token outputs. Across four ordinal benchmarks and four MLLM backbones, ordinal labels are linearly recoverable from hidden states with Spearman correlation up to 0.938, and a task-designed prompt further sharpens this structure. Yet native digit-token outputs weakly expose it: the unembedding matrix filters the ordinal direction, and the digit-token row space retains below 1.15% across all 16 model-dataset combinations, with a 16 to 77 absolute-point accuracy gap between linear-probe and native outputs. We introduce Ordinal Lens Alignment (OLA), a frozen-backbone inference-time method that trains lightweight W_S-anchored lenses on mid-to-deep decoder layers, fuses them into an ordinal distribution, and corrects only digit-token logits at generation. OLA outperforms the SOTA LoRA-tuned OrderChain baseline in most settings while keeping the MLLM frozen, surpasses discriminative ordinal baselines in most cells, and improves over an offline lens in every setting.

[65] Recognition-Conditioned Reasoning: A Training-Free Multimodal-LLM Pipeline for Fine-Grained Micro-Action Understanding cs.CV | cs.MMPDF

Fengshun Wang, Jin’ang Han, Zhigang Tu

TL;DR: 本文提出了一种名为‘识别条件推理’的训练免费多模态大语言模型流水线，专门用于细粒度微动作理解。该系统在禁止微调和真实监督的MAC 2026微动作挑战赛细粒度理解赛道中获得第一名。其核心是动态地将八个子任务路由到最适合的冻结MLLM：判别式MLLM处理封闭式识别任务，生成式MLLM处理开放式描述和推理任务。

Details

Motivation: 微动作是细微、短暂、低幅度的身体动作，能可靠地泄露情感和心理状态。理解它们不仅需要分类标签，还需要描述具体移动的身体部位并推理其与细粒度类别的关系。现有方法难以同时满足这些需求，尤其是在禁止微调和监督的竞赛环境下。

Result: 在MAC 2026微动作挑战赛的MA-Bench细粒度理解赛道上，该系统取得了第一名。在开放式任务上获得了统计显著的性能优势，平均得分达到2.68（五分制），而第二名方法的平均得分仅为1.44。

Insight: 主要创新点在于提出了一种训练免费的、基于提示的流水线架构，通过动态任务路由，将不同的子任务分配给专门化的冻结MLLM（判别式与生成式），从而在无需额外训练的情况下，协同完成复杂的细粒度视频理解任务。这为利用现成MLLM构建高效专用系统提供了新思路。

Abstract: Micro-actions are subtle, short, low-amplitude body movements, such as a fidgeting hand or a slight head tilt, that humans perform with little conscious intent yet that reliably leak emotional and psychological state. Understanding them goes beyond assigning a label: a model must also describe which body parts move and reason, faithfully, about why a clip warrants a particular fine-grained category. We present the training-free, prompt-only system that won first place in the fine-grained understanding track (MA-Bench) of the MAC~2026 Micro-Action Challenge, where both fine-tuning and ground-truth supervision are disallowed. Built entirely upon frozen multimodal large language models (MLLMs), the system dynamically routes each of the eight sub-tasks to the MLLM empirically best suited for that task: a discriminative MLLM for closed-ended recognition tasks and a generative MLLM for open-ended description and reasoning tasks. This architecture achieves a statistically significant performance advantage on open-ended tasks, attaining an average score of 2.68 (on a five-point scale) compared to 1.44 for the second-best approach.

[66] CoAnchor: Robust Collaborative Perception under Spatio-Temporal Misalignment via Object-Level Anchors cs.CV | cs.AIPDF

Chi Li, Rui Lin, Aobo Ji, Dongzhu Xu

TL;DR: CoAnchor提出了一种以目标为中心的时空对齐框架，用于解决异步协同感知中由通信延迟和相对位姿噪声共同导致的观测过时、空间错位和特征融合不稳定问题。该方法通过构建稀疏的目标级时空锚点作为共享接口，在一个统一循环中紧密连接空间细化、时间传播和当前时间验证，实现了轻量化的校正过程。

Details

Motivation: 现实部署中，协同感知接收的消息常受通信延迟和相对位姿噪声共同影响，导致观测过时、空间错位和特征融合不稳定。现有方法通常单独处理时空问题，难以高效统一地应对这些联合扰动。

Result: 在模拟和真实世界数据集上的大量实验表明，CoAnchor在干净设置下保持竞争力，并在联合延迟和位姿扰动下显著提升了鲁棒性，取得了良好的精度与效率权衡。

Insight: 创新点在于将密集BEV特征推理转为稀疏目标级时空锚点作为共享接口，将空间细化、时间传播和当前验证统一在一个循环中，实现了轻量化且鲁棒的联合时空校正。从客观角度看，这种锚点中心的统一循环设计为处理异步协同感知的复杂扰动提供了一种新颖且高效的范式。

Abstract: Collaborative perception extends the sensing range of a single vehicle by fusing observations from nearby agents, which improves the robustness of autonomous driving. In realistic deployments, however, the received collaborator messages are often affected by both communication delay and relative-pose noise, which jointly cause stale observations, spatial misalignment, and unstable feature fusion. Existing methods usually address these issues from either the spatial or temporal side, but handling them jointly in a unified and efficient manner remains challenging. In this paper, we propose CoAnchor, an anchor-centric spatio-temporal alignment framework for asynchronous collaborative perception. Instead of directly reasoning on dense BEV features, CoAnchor builds sparse object-level spatio-temporal anchors as a shared interface for pose correction and tightly connects spatial refinement, temporal propagation, and current-time verification within one unified loop, while keeping the overall correction process lightweight. Extensive experiments on both simulated and real-world datasets illustrate that CoAnchor remains competitive under clean settings and improves the robustness under joint delay and pose perturbations with a favorable practical accuracy-efficiency trade-off.

[67] COMET: Contrastive Motion-Enhanced Temporal Reasoning for Video Multimodal Large Language Models cs.CV | cs.CL | cs.LGPDF

Chenghua Zhu, Zhaolu Kang, Qifan Shi, Siyan Wu, Kehan Jiang

TL;DR: 本文提出了COMET框架，旨在增强视频多模态大语言模型在细粒度运动-时序理解方面的能力。该框架通过显式的时序表示、外观-运动融合和方向感知优化，系统地强化了模型对帧间变化、外观-运动交互以及时序方向敏感性的建模。

Details

Motivation: 现有视频MLLMs在细粒度的运动与时序理解方面仍然薄弱，其核心瓶颈不仅在于稀疏的帧采样，更在于缺乏一个完整的时序建模流程来显式表征帧间变化、促进外观与运动的交互、并优化模型对时序方向的敏感性。

Result: 在Qwen3-VL-8B模型上，COMET方法在动作中心任务（STAR, SSv2）上平均提升4.9%，在时序推理任务（NExT-QA, CLEVRER, LLaVA-178K）上相比BL-GRPO提升2.1%，而在静态感知任务（PerceptionTest）上保持相当水平。该增益模式在InternVL2.5-8B模型上也得到验证，表明COMET具有良好的跨模型泛化能力。

Insight: 创新点在于构建了一个系统性的时序增强框架，其架构上通过基于泰勒帧差构建时序运动分支，并利用时序注意力偏置增强的交叉注意力将运动证据注入外观流；优化上结合了时序先验蒸馏和前后向TC-GRPO阶段，将时序顺序转化为直接的学习信号，强化了模型对方向性运动模式的使用。这为视频理解提供了一个显式、完整且可泛化的时序建模解决方案。

Abstract: Video multimodal large language models have advanced significantly, yet fine-grained motion-temporal understanding remains fragile. The core bottleneck is not only sparse frame sampling, but also the lack of a complete temporal modeling pipeline for explicitly representing frame-to-frame change, enabling appearance-motion interaction, and optimizing temporal direction sensitivity. We propose COMET, a temporally grounded framework that systematically strengthens video MLLMs through explicit temporal representation, appearance-motion fusion, and direction-aware optimization. Architecturally, COMET introduces a temporal motion branch built on Taylor frame differences and injects its motion evidence into the appearance stream via temporal attention bias-enhanced cross-attention. For optimization, COMET combines temporal prior distillation with a forward-reverse TC-GRPO stage that turns temporal order into a direct learning signal and strengthens the model’s use of directional motion patterns encoded by the temporal motion branch. The method achieves consistent overall improvements with a pronounced motion-temporal bias: on Qwen3-VL-8B, action-centric tasks (STAR, SSv2) improve by 4.9% on average, temporal reasoning tasks (NExT-QA, CLEVRER, LLaVA-178K) by 2.1% over BL-GRPO, while static perception tasks (PerceptionTest) remain on par. The same gain pattern also transfers to InternVL2.5-8B, indicating that COMET generalizes across model families.

[68] AT-ViT: Area-Targeted Multi-View Vision Transformer with Cross-Attention and Multi-Scale Patching for Plant Trait Recognition in Herbarium Images cs.CV | cs.AIPDF

Amani Sedrat, Takieddine Chehhat, Youcef Sklab, Hanane Ariouat, Abderrazak Sebaa

TL;DR: 本文提出了AT-ViT，一种用于植物标本图像性状识别的双分支视觉Transformer。它通过多尺度、多视角交叉注意力融合方案，联合编码原始扫描图像和分割掩码导出的图像，并引入掩码引导的补丁加权机制，以增强植物相关区域并抑制背景干扰。

Details

Motivation: 植物标本图像的自动性状识别对植物科学至关重要，但背景元素（如文本标签、装裱痕迹和色卡）会引入捷径学习，导致模型依赖虚假的非植物线索而非植物形态，从而损害模型的泛化能力和可解释性。

Result: 在多个性状分类任务（如叶基形状、刺）上，AT-ViT取得了持续的准确率提升，改善了注意力在植物区域的定位，并在合成背景扰动下表现出更强的鲁棒性。具体而言，相对于CrossViT，AT-ViT显著提升了植物区域对齐（Avg IoU_p: +15.66 至 +18.03 个百分点），并减少了背景重叠（Avg IoU_b: -27.92 至 -31.02 个百分点）；在背景噪声条件下，其准确率最高超过ResNet101达32.32个百分点，超过CrossViT达5.07个百分点。

Insight: 论文的创新点在于提出了一个结合多视图（原始图像与分割图像）和多尺度补丁的双分支Transformer架构，并设计了掩码引导的补丁加权机制来引导模型关注植物器官。从客观角度看，这种利用分割先验信息来抑制背景、增强目标区域特征的方法，为解决图像识别中背景干扰导致的捷径学习问题提供了有效的技术思路。

Abstract: Automated plant traits recognition from herbarium images is essential for plant sciences, yet remains challenging because background elements (e.g., textual labels, mounting artifacts, and color charts) can introduce shortcut learning, leading models to rely on spurious non-plant cues rather than plant morphology. This bias degrades both generalization and interpretability. In this paper, we introduce AT-ViT, a dual-branch Vision Transformer that jointly encodes raw herbarium scans and their segmented-derived counterparts via a multi-scale, multi-view cross-attention fusion scheme. AT-ViT further incorporates a mask-guided patch weighting mechanism that amplifies plant-relevant regions and attenuates background-driven features. By learning from the original scans while being guided by segmentation masks through the mask-guided patch reweighting mechanism, the model is encouraged to focus on plant organs and learn plant-centric representations more effectively. Across multiple trait classification tasks (e.g., leaf base shape, thorns), AT-ViT delivers consistent accuracy gains, improves attention localization on plant regions, and exhibits increased robustness under synthetic background perturbations. Specifically, AT-ViT substantially improves spatial attention grounding, boosting plant-region alignment (Avg IoU_p: +15.66 to +18.03 pp) while reducing background overlap (Avg IoU_b: -27.92 to -31.02 pp) relative to CrossViT, and remains markedly more robust to background perturbations, outperforming ResNet101 by up to +32.32 accuracy points and CrossViT by up to +5.07 points under background-noise conditions.

[69] Robust Validation to Geometric Perturbations for Autonomous Pose Estimation cs.CVPDF

Gregoire Theau, Melanie Ducoffe

TL;DR: 本文针对自主姿态估计系统在安全关键领域部署时面临的几何扰动鲁棒性验证问题，提出了一种基于全局Lipschitz优化（GLO）的鲁棒验证框架。研究表明，传统的基于梯度的启发式方法（如APGD）在姿态估计任务上效果不佳，甚至不如随机采样基线，而GLO方法能有效定位导致姿态偏差超出安全限的临界故障模式，并大幅缩减搜索空间。

Details

Motivation: 在安全关键领域部署自主系统时，需要保证系统对物理上合理的几何扰动（如相机旋转、光照变化）具有鲁棒性，而非仅针对抽象的像素级噪声。现有基于梯度的启发式方法在姿态估计的鲁棒性验证中存在优化瓶颈，无法有效发现故障模式。

Result: 在针对YOLOv8-Pose关键点检测器和PnP求解器，对抗旋转和对比度扰动的评估中，GLO框架成功定位了位置偏差超出安全操作限制的临界故障模式，同时将搜索空间快速剪枝超过80%。

Insight: 创新点在于将几何鲁棒性验证扩展到连续关键点回归和深度目标检测任务，并首次引入全局Lipschitz优化（GLO）这一具有理论收敛保证的原则性方法来解决姿态估计的鲁棒验证优化瓶颈，为认证鲁棒的自主感知提供了实用步骤。

Abstract: Deploying autonomous systems in safety-critical domains demands guaranteed robustness against physically plausible geometric perturbations rather than abstract pixel-wise noise. In vision-based navigation and autonomous landing, machine learning components require rigorous validation under dynamic operational conditions such as camera rotations and lighting shifts. Extending findings on the failure of first-order spatial attacks in classification, we show that standard gradient-based heuristics (e.g. APGD) similarly fail on for pose estimation, often performing worse than a simple random sampling baseline. To overcome these optimization bottlenecks, we reformulate pose estimation robustness within the framework of Global Lipschitzian Optimization (GLO). We argue that GLO offers a principled approach to robust validation, effectively localizing global optima with strong theoretical convergence guarantees. We evaluate this framework on a YOLOv8-Pose keypoint detector with a Perspective-n-Point (PnP) solver against rotation and contrast. In our evaluations, GLO successfully isolates critical failure modes where position deviations exceed safe operational limits, while rapidly pruning the search space by over 80%. To the best of our knowledge, this is the first study to extend geometric robustness validation to continuous keypoint regression and deep object detection, establishing a practical step toward certifying robust autonomous perception.

Jiekang Feng, Zhihe Fan, Yunqi Zhu, Xinjie Yao, Yueying Zhang

TL;DR: 本文提出A2DINOv3，一个基于社会化协作协议的多模态目标检测框架，旨在解决现有密集跨模态融合方法可能破坏预训练表示的问题。该框架将RGB和红外分支建模为异构专家，通过选择性交互保留各自专业知识并交换互补信息，同时采用零初始化策略平滑激活跨模态协作。

Details

Motivation: 现有方法在将视觉基础模型（如DINOv3）适配到多模态场景时，常采用密集的跨模态融合策略，这可能导致冗余信息引入并破坏有价值的预训练表示。

Result: 在四个多模态基准测试（GAIIC、FLIR、LLVIP、M3FD）上的广泛实验表明，A2DINOv3在多模态目标检测中持续取得了最先进的性能。

Insight: 创新点在于从社会化学习视角重新思考多模态融合，提出异构专家选择性协作的协议，以及零初始化策略以实现从模态特定学习到协作表示学习的平滑过渡，有效减轻了有害的跨模态干扰。

Abstract: Multi-modal object detection is essential for robust scene understanding in challenging conditions, including low-light and adverse environments. Recent vision foundation models (e.g., DINOv3) have exhibited strong representation capabilities, yet adapting them to multi-modal scenarios remains challenging. Existing dense cross-modal fusion strategies often force heterogeneous modalities to interact indiscriminately, which may introduce redundant information and disrupt the valuable pre-trained representations. To address this issue, we revisit multi-modal fusion from the perspective of socialized learning and propose adapter to DINOv3 (A2DINOv3), a multi-expert collaboration framework with a Socialized Collaboration Protocol (SCP). Specifically, RGB and infrared branches are modeled as heterogeneous experts that independently preserve their specialized knowledge while exchanging complementary information through selective and constrained interactions. This design mitigates harmful cross-modal interference and prevents degradation of pre-trained priors during adaptation. Furthermore, a zero-initialization strategy is introduced to gradually activate cross-modal collaboration, enabling a smooth transition from modality-specific learning to cooperative representation learning. Extensive experiments on four multi-modal benchmarks, including aerial detection (GAIIC), autonomous driving (FLIR), low-light surveillance (LLVIP), and diverse real-world scenarios (M3FD), demonstrate that A2DINOv3 consistently achieves state-of-the-art performance in multi-modal object detection.

[71] Masking Is Not Enough: Generative Restoration for Multimodal De-Identification in Medical AI cs.CV | cs.CRPDF

Shiva Shrestha, Zongxing Xie, Chen Zhao, Liran Ma, Zhipeng Cai

TL;DR: 本文提出了ClinX，一个用于医疗图像-文本数据多模态去识别的端到端框架。该框架通过光学字符识别检测可见标识符，构建PHI掩码，并应用生成式恢复模块进行隐私处理；同时对文本侧PHI进行多级去识别。研究评估了其在医学视觉问答任务中的表现，发现仅靠OCR掩码不足，而基于恢复的去识别方法能更好地保护临床相关视觉上下文并显著减少可恢复的PHI。

Details

Motivation: 医疗图像-文本数据可能通过可见图像内容和伴随文本暴露受保护的健康信息，这阻碍了隐私保护医疗AI系统的发展，现有方法往往将去识别与下游推理分开评估。

Result: 在医学视觉问答任务中评估，结果显示仅使用OCR掩码作为独立解决方案不足，而基于恢复的去识别方法在显著减少可恢复PHI的同时更好地保留了临床相关视觉上下文。

Insight: 创新点在于提出端到端多模态PHI净化框架，结合生成式恢复模块与多级文本去识别，并首次在MedVQA任务中联合评估PHI泄漏和下游效用，强调了恢复方法在隐私保护中的重要性。

Abstract: Medical image-text data can expose protected health information (PHI) through both visible image content as well as accompanying text, creating a barrier to privacy-preserving medical AI systems. This risk is especially prominent in multimodal systems, where images, questions, reports, and clinical context may enter training, evaluation, or inference pipelines. Existing medical vision-language benchmarks primarily emphasize task utility, while de-identification methods are often evaluated separately from downstream reasoning. We introduce ClinX, an end-to-end multimodal PHI sanitization framework for medical image-text data. ClinX detects visible identifiers with optical character recognition (OCR), constructs binary PHI masks, and applies ClinX-PRISM, a no-skip generative restoration module with privacy-oriented post-processing for burned-in identifier suppression. In parallel, text-side PHI is reduced through progressive de-identification levels: regex masking, context-aware masking, and rewrite-based sanitization. We evaluate ClinX in medical visual question answering (MedVQA), jointly measuring PHI leakage and downstream utility across image-side, text-side, and combined de-identification settings. Results show that OCR-only masking is not sufficient as a standalone solution, and restoration-based sanitization better preserves clinically relevant visual context while sharply reducing recoverable PHI.

[72] Llama-Mobile: Efficient 2.7-Bit Quantization of VLMs cs.CV | cs.LGPDF

Luka Ribar, Jeevan Bhoot, Douglas Orr

TL;DR: 本文提出了一种名为Llama-Mobile的框架，用于高效量化视觉语言模型（VLMs），以在资源受限的移动设备上部署。该方法结合了利用模型自身生成训练数据的量化流程（无需访问原始训练设置）和一种新颖的2.7比特/参数格式，支持在Arm CPU上高效执行。作者通过将Llama 3.2 11B Vision Instruct模型压缩至3.7 GB（使用8位激活），并在标准视觉问答任务上保持强劲性能，验证了其有效性。

Details

Motivation: 解决在内存和计算资源有限的移动设备上部署视觉语言模型（VLMs）的挑战，因为VLMs通常有巨大的资源需求。

Result: 将Llama 3.2 11B Vision Instruct模型压缩到3.7 GB（激活为8位），在一系列标准视觉问答任务上保持了强大的性能，验证了方法的有效性。

Insight: 创新点在于结合了无需原始训练数据的自生成训练数据量化流程和一种新颖的2.7比特/参数格式，专门针对Arm CPU优化，实现了高压缩比下的性能保持，为移动端VLM部署提供了高效解决方案。

Abstract: Deploying vision-language models (VLMs) on mobile devices is challenging due to their significant memory and compute requirements. We present a framework for quantizing VLMs for efficient inference on resource-constrained hardware. Our approach combines a quantization pipeline that uses the model itself to generate training data and does not require access to the training setup, with a novel 2.7-bit-per-parameter format supporting efficient execution on Arm CPUs. We validate our approach by compressing the Llama 3.2 11B Vision Instruct model to 3.7 GB with 8-bit activations, preserving strong performance on a set of standard visual question answering tasks.

[73] Stream3Dv2: Geometric-Semantic Fusion Enhanced Streaming Zero-Shot 3D Scene Understanding cs.CVPDF

Jie Xu, Na Zhao

TL;DR: Stream3Dv2是一个无需训练、用于鲁棒流式3D场景理解的框架。它通过一种新颖的嵌套局部-历史架构处理序列RGB-D数据，并引入几何-语义融合机制以及基于流形距离的点云优化策略，以解决现有方法在流式输入处理和噪声2D分割掩码方面的不足。

Details

Motivation: 当前基于视觉基础模型的开放词汇零样本3D场景理解方法，在真实世界部署中存在两大关键限制：无法高效处理流式RGB-D输入，以及对噪声2D分割掩码的固有脆弱性。

Result: 在公开数据集上的大量实验表明，Stream3Dv2在基础的开放词汇流式3D分割和检测任务上，持续优于现有基线方法。

Insight: 创新点在于：1）嵌套局部-历史架构，在保证多视图一致性的同时支持实时响应；2）将3D分割明确建模为点与集合的合并与划分问题，利用语义指导进行几何-语义融合；3）提出基于流形距离的点云优化策略，利用局部流形图进行点-流形优化，并结合几何边界框实现快速的流形-流形细化，以克服欧氏距离度量导致的边界划分失败问题。

Abstract: Recently, open-vocabulary zero-shot 3D scene understanding using vision foundation models has emerged as a promising alternative to data-intensive supervised methods. However, deploying these models in real-world scenarios is severely hindered by their inability to efficiently handle streaming RGB-D inputs and their inherent vulnerability to noise 2D segmentation masks. To address these critical limitations, we propose Stream3Dv2, a novel training-free framework designed for robust streaming 3D perception. Stream3Dv2 processes sequential data through an original nested local-to-historical architecture, capturing multi-view consistency while circumventing the high computational overhead so as to support timely responses. At its core, we introduce a comprehensive geometric-semantic fusion mechanism that resolves geometric noise and semantic ambiguity by explicitly utilizing semantic guidance and formulating 3D segmentation as solving point-and-set merging and partitioning problems. Furthermore, we present an innovative manifold-distance-based point cloud refinement strategy. This approach leverages local manifold graphs for point-to-manifold optimization that mitigates the boundary delineation failures caused by Euclidean-distance metrics, and employs geometric bounding boxes to dynamically activate and update historical instances for achieving rapid manifold-to-manifold refinement. Extensive experiments on public datasets demonstrate that Stream3Dv2 consistently outperforms existing baselines in foundational open-vocabulary streaming 3D segmentation and detection. Finally, we show that integrating our framework with an LLM-based agent enables advanced language-driven 3D scene understanding, underscoring its potential for open-world embodied intelligence. Code will be updated at https://github.com/SubmissionsIn/Stream3D.

[74] Human-JEPA: A Human-Centric Vision Model that Perceives and Anticipates cs.CV | cs.LGPDF

Hui Wei, Licai Sun, Guoying Zhao

TL;DR: 本文提出了Human-JEPA，一种以人为中心的视觉模型，通过锚定预测在视频上进行训练，旨在同时感知当前状态和预测未来。该方法通过将密集预测目标锚定到初始化的冻结副本上来防止感知崩溃，并采用纯过去到未来的分割来避免性能下降。

Details

Motivation: 现有以人为中心的视觉模型主要在静态图像上预训练，擅长密集感知但无法处理运动和预测。本文旨在开发一个能同时理解人类当前状态并预测其未来行为的模型。

Result: 在冻结探针评估下，Human-JEPA在姿态估计和行人重识别任务上，以2.7倍更少的参数领先于像素锚定的专家模型，同时其发布的预测器头是首个不损害预测性能的。

Insight: 创新点在于锚定预测训练策略，通过冻结初始化副本作为锚点来稳定密集感知，以及采用纯时序分割而非块掩码，有效避免了模型在动作识别和重识别任务上的性能崩溃，实现了感知与预测的统一。

Abstract: Machines that understand humans should perceive the present and anticipate the future. Existing human-centric vision model are pretrained on human images, set the state of the art in static dense perception, so motion and anticipation are out of reach. Here we present Human-JEPA, a human-centric vision model trained on video by anchored forecasting: dense targets are pinned to a frozen copy of the initialization, preventing a silent collapse of dense perception, and block masks are replaced by a pure past-to-future split, avoiding a five-point action tax and a seventeen-point re-identification collapse. Under frozen probes, Human-JEPA leads the pixel-anchored specialists on pose and person re-identification at 2.7 times fewer parameters, conceding high-resolution dense parsing, and its released predictor head is the first that does not degrade anticipation. A single safely adapted model thus serves both halves of understanding humans.

[75] Is Visual Prompting All You Need? Studying VLM Spatial Reasoning under Progressive Visual Scaffolds cs.CV | cs.AIPDF

Lars Benedikt Kaesberg, Tianyu Yang, Florian Valentin Wunderlich, Terry Ruas, Jan Philip Wahle

TL;DR: 本文研究了视觉语言模型在空间推理任务中的表现，通过引入渐进式视觉支架来探究视觉呈现方式对模型性能的影响。研究发现，在保持相同底层推理问题的情况下，轻量级的输入侧视觉支架能显著提升模型在SPaRC基准测试中的准确率，并减少与视觉定位相关的错误。

Details

Motivation: 动机在于探究视觉语言模型在多模态推理中失败的原因是否与视觉呈现方式有关，特别是在空间推理任务中，视觉结构如何影响模型的性能与失败模式。

Result: 在SPaRC基准测试中，视觉支架使多个VLM的任务准确率最高提升34.0个百分点，结合GRPO训练后还能额外提升4.6个百分点，而原始视觉输入上的增益接近零。分析表明，这些提升主要源于减少视觉定位错误，但规则推理仍具挑战性。

Insight: 创新点在于通过渐进式视觉支架分离视觉感知与下游推理，揭示视觉呈现是决定VLM基准测试衡量内容的关键因素；从客观角度看，该方法为评估和提升VLM的空间推理能力提供了可借鉴的轻量化干预策略。

Abstract: Vision-language models (VLMs) have advanced rapidly in multimodal reasoning, yet recent work shows that their failures often reflect an interaction between visual grounding and downstream reasoning. What remains less clear is how the visual presentation of a task shapes model performance and failure modes when the underlying reasoning problem is unchanged. We study this question in SPaRC, a benchmark for grid-based visual spatial planning, by introducing lightweight input-side scaffolds that preserve the visual modality while making spatial structure more accessible. Across multiple VLMs, these scaffolds improve task accuracy over the original visual setting by up to 34.0 percentage points and further complement GRPO-based training, yielding up to 4.6 additional accuracy points compared with near-zero gains on the original visual input. Analyses on both end-to-end task solving and object detection show that these gains are closely tied to reductions in grounding-related errors, while rule reasoning remains comparatively challenging. We find that visual presentation is a central factor that determines whether VLM benchmarks measure grounded perception, downstream reasoning, or a mixture of both.

[76] A Modular Agent for Reliable and Auditable Spatial Relation Verification in CT Scans cs.CV | cs.AIPDF

Simon Vincent Abel, Heiko Hillenhagen, Michael Götz, Timo Ropinski, Ayhan Can Erdur

TL;DR: 本文提出了一种模块化的医学影像智能体，用于在轴向CT切片中进行可靠且可审计的空间关系验证。该系统将空间关系验证任务分解为语言解析、解剖定位和确定性几何验证三个阶段，而非端到端预测答案。在MIRP空间问答基准测试中，该方法达到了94.1%的准确率和94.2%的F1分数，显著优于端到端视觉语言模型基线，并保持了可解释的中间表示和可审计的推理过程。

Details

Motivation: 当前医学视觉语言模型在受控空间推理方面存在不足，难以可靠地将空间关系基于图像证据进行定位，这给依赖解剖结构相对位置理解的放射学诊断准确性带来了风险。

Result: 在MIRP空间QA基准测试上，最佳混合配置达到了94.1%的准确率和94.2%的F1分数，比直接使用Qwen2-VL提示的准确率高出42.5个百分点，实现了SOTA性能。

Insight: 创新点在于采用模块化、分阶段的显式推理架构（语言解析→目标检测→几何规则计算）替代端到端黑箱模型，从而在提升空间关系验证可靠性的同时，保证了推理过程的可解释性和可审计性，为面向报告的医学影像系统提供了可靠的构建模块。

Abstract: Reliable spatial understanding is an important prerequisite for future medical vision-language systems that aim to support radiological report generation and structured image understanding. While modern vision-language models (VLMs) show promising performance on many medical imaging tasks, recent evidence suggests they remain weak in controlled spatial reasoning and often fail to reliably ground spatial relations in image evidence. Given that radiological reasoning hinges on understanding the relative positions of anatomical structures and findings, this spatial weakness poses risks to diagnostic accuracy. We present a modular medical imaging agent for binary spatial relation verification in axial CT slices. Instead of directly predicting spatial answers end-to-end, the system decomposes the task into explicit stages: language parsing, anatomical localization, and deterministic geometric verification. Natural-language queries are converted into structured relation tuples, queried organs are localized with a YOLO-based detector, and the final spatial decision is computed from object centers using deterministic geometric rules. We evaluate the approach on the held-out MIRP spatial QA benchmark and compare it against representative end-to-end VLM baselines. The best-performing hybrid configuration reaches 94.1% accuracy and 94.2% F1, outperforming direct Qwen2-VL prompting by 42.5 percentage points in accuracy, while preserving interpretable intermediate representations and auditable reasoning stages. The results suggest that explicit modular spatial verification can serve as a promising building block for future report-oriented medical imaging agents.

[77] Towards Investigating Residual Hearing Loss: Quantification of Fibrosis in a Novel Cochlear OCT Dataset cs.CV | cs.AIPDF

Julia Dietlmeier, Benjamin Greenberg, Wenxuan He, Teresa Wilson, Rubing Xing

TL;DR: 本研究旨在通过计算机视觉方法量化耳蜗植入后纤维化程度，以评估其对残余听力的影响。作者构建了一个新型的光学相干断层扫描（OCT）数据集，并应用语义分割模型（特别是改进的UNET架构）自动识别纤维化区域，为研究耳蜗纤维化提供了高分辨率、可重复的量化工具。

Details

Motivation: 耳蜗植入物（CIs）尤其是混合型CIs结合电声刺激（EAS）时，植入引发的耳蜗内纤维化可能损害残余听力，降低EAS疗效。因此，需要一种客观、高分辨率的方法来量化纤维化负担，以推动减少纤维化、改善患者预后的转化研究。

Result: 在手动标注的OCT图像数据集上，作者测试了多种先进的语义分割模型，发现改进的UNET架构（2D-OCT-UNET）在放大输入分辨率下性能最佳，能够可靠地计算耳蜗纤维化负担，验证了计算机视觉技术在此类OCT数据中的首次成功应用。

Insight: 创新点在于首次将计算机视觉技术应用于植入耳蜗纤维化的OCT数据集，并开发了针对高分辨率OCT图像优化的2D-OCT-UNET模型，实现了纤维化的自动量化；这为耳科医学研究提供了可重复的客观评估工具，数据集和代码开源促进了相关领域的发展。

Abstract: Objective: Cochlear implants (CIs) are bionic prostheses that restores hearing via electrical stimulation of the auditory nerve. Hybrid CIs, which use electroacoustic stimulation (EAS), combine residual low-frequency acoustic hearing with CI electrical stimulation. Intracochlear fibrosis, which forms in response to the presence of the implant, may impede residual hearing function and gradually reduce the efficacy of EAS. It is therefore a translational objective to study the formation of cochlear fibrosis in rodents, with the goal of reducing fibrotic burden and improving outcomes for CI patients. Methods: We generate and annotate a novel dataset of optical coherence tomography (OCT) images from chronically implanted guinea pigs as part of an ongoing study focused on implant induced fibrosis. Objectively assessing fibrotic burden in this model, with high resolution and repeatability, presents an obvious use case for computer vision methods. Results: We present the results of several state-of-the-art semantic segmentation models and compare their efficacy for identifying cochlear fibrosis and other relevant annotations, using a new library of manually segmented OCT images. Conclusions: We find that the best performance is achieved by using a modified version of the well-known UNET architecture (which we term 2D-OCT-UNET) that operates on the upscaled OCT input resolution. Significance: For the first time, we have successfully applied computer vision techniques to an OCT dataset of implanted cochleae with fibrosis. Using this deep learning model, the cochlear fibrotic burden calculation can be reliably carried out as we verify in our experimental section. The dataset and the project code are available at: https://github.com/juliadietlmeier/CF-OCT-segmentation

[78] A VLM Answer Is Not an Anomaly Score: Rank Compression in Training-Free Video Anomaly Detection cs.CVPDF

Inpyo Song, Jangwon Lee

TL;DR: 本文探讨了基于视觉语言模型（VLM）的无训练视频异常检测（VAD）中，答案接口（包括答案量表和读出规则）对检测性能的关键影响。研究发现，使用基于完整概率分布的’概率读出’方法，在多个VLM、基准测试和指标上，均显著优于仅使用最可能答案的’生成读出’方法，平均性能提升5到13个百分点。

Details

Motivation: VLM可用于无训练视频异常检测，但VAD基准测试要求为每个视频片段输出一个标量异常分数，并基于此排序进行评估。然而，如何将VLM的答案输出映射为异常分数（即答案接口的设计）会影响最终的排序结果，目前这一关键组件常被忽视或未明确指定。

Result: 在四个7-8B参数的VLM上，对多种答案量表、基准测试（如UCF-Crime, ShanghaiTech）和评估指标（AUROC, AP）的组合测试表明，概率读出方法在所有测试条件下均优于生成读出方法，平均性能提升显著。生成读出方法由于仅保留一个答案值，导致分数分辨率极低（即使有91个可能答案，也只产生4-18个不同分数），从而压缩了排序信息。

Insight: 论文的核心创新点在于明确指出并系统分析了’答案接口’（包括答案量表和读出规则）是VLM-based VAD检测器的重要组成部分，而非简单的格式化细节。其关键发现是’生成答案排序压缩’现象：仅使用最可能答案的生成读出会丢失答案分布信息，导致排序分辨率严重下降，而利用完整分布的概率读出能保留更精细的排序信息。这为未来设计基于VLM的VAD方法提供了重要的设计准则：必须明确指定并评估答案接口。

Abstract: Vision-language models enable training-free video anomaly detection by answering questions about video segments. VAD benchmarks, however, require a scalar anomaly score for each segment and evaluate the resulting ranking using the AUROC or AP. A VLM-based detector should therefore define an answer interface: the answer scale specifies the admissible answers, and the readout rule maps the model’s output distribution to a score. Because this interface can change the evaluated ranking, it is part of the detector rather than a formatting detail. The generated readout uses only the most likely answer, whereas the probability readout uses the full distribution over admissible answers. Across four 7-8B VLMs, the probability readout outperforms the generated readout for every tested combination of answer scale, benchmark, and metric, with average gains ranging from 5 to 13 points across the four benchmark-metric pairs. The gap arises because the generated readout keeps only one answer value per segment, so segment with different answer distributions can receive the same score and lose their relative order. We call this loss of relative order generated-answer rank compression. Even when the answer scale allows 91 answers, the generated readout produces only 4-18 distinct scores, whereas the probability readout retains substantially finer score resolution. The advantage persists under every decoding strategy, prompt wording, and joint scoring-explanation prompt we test. The answer interface is therefore a consequential component of VLM-based VAD and should be explicitly specified and evaluated.

[79] Just Noticeable Difference Modeling for Token Compression in Vision-Language-Action Models cs.CV | cs.ROPDF

Zhuoyuan Li, Rui Zhao, Jin Wang, Hanwei Zhu, Cong Zhang

TL;DR: 本文提出了Action-JND，一种用于具身智能体（vision-language-action models）中令牌压缩的新方法。该方法将经典的’恰可察觉差异’（JND）概念扩展到闭环控制中，通过评估令牌变化对下游机器人动作预测的影响来指导压缩。作者开发了一个轻量级的JND估计器来预测令牌的最大可容忍扰动，并将其作为即插即用的标准，应用于KV缓存重用和令牌剪枝等压缩方案。

Details

Motivation: 现有令牌压缩方案（如基于视觉相似性或注意力分数）仅间接衡量令牌的重要性，未能直接评估压缩对下游关键任务（如机器人动作预测）的影响。在延迟敏感的具身智能体中，确保压缩不会导致不可接受的动作偏差至关重要。

Result: 在LIBERO基准测试上，使用OpenVLA和OpenVLA-OFT模型进行的实验表明，Action-JND方法能持续提升压缩的可靠性，尤其是在高压缩比下效果更为显著。

Insight: 核心创新在于将JND原则从人类感知系统扩展到机器响应，提出了以动作容忍度（action tolerance）作为令牌压缩的直接指导准则。这为具身AI模型的效率优化提供了一个新的、任务感知的视角，其轻量级JND估计器设计也具有良好的可部署性。

Abstract: Token compression has become a key technique for reducing the inference cost of large foundation models, with approaches such as token pruning and KV-cache reuse widely adopted in vision-language models and recently explored for embodied agents. In embodied agents, tokens not only support perception and semantic understanding but also directly affect latency-sensitive closed-loop robot action prediction. Existing schemes typically guide compression using redundancy or importance cues, such as visual similarity, attention scores, and saliency. However, these cues only indirectly measure the key factor for safe compression: how much a token can change before causing an unacceptable deviation in downstream actions. This receiver-dependent tolerance is closely related to the principle of just noticeable difference (JND). Classical JND characterizes signal tolerance in the human visual system, while machine-oriented JND extends this concept to downstream machine responses. Building on this progression, we introduce Action-JND, which extends JND modeling to embodied perception by defining noticeability through the language-conditioned action response of a vision-language-action (VLA) policy in closed-loop control. A token change is considered admissible only when the induced action deviation remains within a tolerated margin. To realize this concept, we develop a lightweight token-wise JND estimator in deep visual-feature space to predict the maximum tolerable perturbation while preserving policy responses. The resulting action-tolerance score serves as a plug-and-play criterion for VLA compression paradigms, including stale-KV reuse and token pruning, prioritizing action-tolerant tokens for compression. Experiments on the LIBERO benchmark with OpenVLA and OpenVLA-OFT demonstrate that Action-JND consistently improves compression reliability, especially under aggressive compression ratios.

[80] On the Transferability of Agricultural Weed Detection Under Cross-Field Distribution Shift cs.CV | cs.LGPDF

Nikhilesh Prabhakar, Pranuthi Tenali, Wilfredo Abudeye Fernandez, Shekhar Borah, Athresh Karanam

TL;DR: 本文研究了农业杂草检测模型在跨作物和跨田块分布偏移下的可迁移性，通过新收集的棉花田无人机图像数据集和现有大豆数据集，评估了多种迁移策略，发现少量目标域标注样本的微调优于无监督域自适应目标检测。

Details

Motivation: 现有杂草检测方法通常在单一作物和田间评估，缺乏模型在新田块或作物类型上泛化能力的证据，本文旨在探究跨数据集性能下降的原因及恢复策略，减少新部署场景的重复标注需求。

Result: 在跨作物比较中，仅使用25个目标域标注样本的少样本微调策略超越了无监督域自适应目标检测方法，表明源域选择结合适度目标监督比复杂自适应算法更有效。

Insight: 创新点在于系统评估了农业杂草检测的跨域迁移性能，并提出了源域选择与少量目标监督结合的实用策略，为实际部署提供了高效解决方案，而非依赖复杂的无监督自适应算法。

Abstract: Accurate agricultural weed detection in real-world field conditions is essential for precision agriculture, enabling targeted intervention and reducing yield loss. Recent work has reported strong detection performance from UAV-based imagery across a range of crops, yet existing approaches evaluate within a single crop and field, leaving practitioners with little evidence that a model trained on one crop will generalize to a new field or crop type. In this work, we characterize where cross-dataset weed-localization performance degrades and which modeling choices recover it, reducing the need to relabel every new deployment field. We introduce a newly collected and annotated UAV image dataset for agricultural weed detection in cotton fields and use it alongside an existing soybean dataset collected under a similar protocol. Using these datasets, we evaluate the performance of several strategies for transferring a detector trained on one crop to another, comparing unsupervised domain adaptive object detection (DAOD) against pretraining on a domain-adjacent source dataset followed by few-shot fine-tuning on the target dataset. Our analysis spans target-domain label budgets from zero to the full target dataset, characterizing the trade-off between adaptation strategy and annotation effort. We find that few-shot fine-tuning with as few as 25 labeled target examples outperforms unsupervised DAOD in our cross-crop comparison, suggesting that source domain selection combined with modest target supervision is more productive than algorithmic sophistication in adaptation.

[81] WildFin: An In-the-Wild Dataset for Fish Behavioral Recognition cs.CVPDF

Abigail G. Grassick, Jerome Tze-Hou Hsu, Ethan Lin, Ziang Liu, Max Whitton

TL;DR: 该论文介绍了WildFin，一个用于鱼类行为识别的野外视频数据集，旨在解决生态学中野外视频数据利用的瓶颈问题。数据集包含两种关键的真实世界场景：固定摄像头监控鱼群和潜水员动态跟踪个体鱼类。研究通过基准测试现代视觉基础模型，揭示了当前模型能力与真实世界水下行为分析需求之间的巨大差距。

Details

Motivation: 生态学领域野外视频数据激增，但专家标注成本高昂成为主要瓶颈；现有计算机视觉模型在复杂海洋环境中部署时经常失败，需要新的基准来表征这些失败并推动模型改进。

Result: 在WildFin数据集上对现代视觉基础模型进行了基准测试，量化了静态架构与时空架构之间的权衡，结果表明当前模型能力与真实世界水下行为分析需求之间存在巨大差距。

Insight: 创新点在于构建了一个由生态学家收集和标注的大规模、高质量的野外鱼类行为识别基准数据集，涵盖了静态和动态两种关键的真实采集范式，为评估和改进模型在复杂真实环境中的鲁棒性提供了重要资源。

Abstract: Recent advances in field technology have led to a massive influx of in-the-wild video data for ecological science. The primary bottleneck in leveraging this data is the high cost of expert annotation. While computer vision offers a potential solution, current models frequently fail when deployed in complex marine environments. To characterize these failures, we introduce WildFin, a novel benchmark for fish behavior recognition collected and annotated by ecologists.WildFin spans two critical real-world paradigms: stationary cameras monitoring groups of fish and dynamic divers following individual subjects. The dataset represents a massive curation effort, involving 1,350 hours of fieldwork and 600 hours of expert annotation to produce 9 hours of behavioral data with over 2 million frame-by-frame labels. We benchmark modern vision foundation models and quantify tradeoffs between static and spatiotemporal architectures, revealing the substantial gap that remains between current model capabilities and the demands of real-world underwater behavioral analysis. Project website: https://team-wildfin.github.io/.

Haonan Jia, Shichao Dong, Zenghui Sun, Jiawen Zheng, Ziqi Miao

TL;DR: 本文提出了Re$^3$Cap方法，一种基于检索引导的推理策略，用于增强图像描述生成。该方法通过检索机制识别描述中的幻觉和遗漏，并利用强化学习进行优化，无需额外标注即可生成更准确、详细的图像描述。

Details

Motivation: 现有基于强化学习的图像描述方法在鼓励大型视觉语言模型探索新颖推理策略方面存在局限，导致其性能与监督微调存在差距。本文认为多模态检索可作为有效的描述优化信号。

Result: 在COCO-LN500基准测试中，Re$^3$Cap在关系推理任务上平均比GRPO方法提升了8.64%，实验表明其性能甚至优于监督微调方法。

Insight: 创新点在于将多模态检索作为强化学习的推理引导信号，通过Caption Refinement Suggester和Caption Quality Assessor模块实现无监督的描述质量评估与优化，为缓解描述幻觉问题提供了新思路。

Abstract: Reinforcement Learning (RL) has demonstrated significant gains in image captioning, yet it is still limited in encouraging Large Vision-Language Models (LVLMs) to explore novel reasoning strategies. This limitation leads to a performance gap between RL and Supervised Fine-Tuning (SFT). In this paper, we argue that multi-modal retrieval can serve as an effective reasoning signal for caption refinement. Based on this insight, we present the Retrieval-Guided Refinement for Image Captioning (Re$^3$Cap), a retrieval-guided reasoning strategy that enhances image captioning without requiring additional annotations. Instantiated by Caption Refinement Suggester (CRS) and Caption Quality Assessor (CQA), this strategy identifies hallucinations and omissions in image captions, leading to more accurate and detailed descriptions. Extensive experiments demonstrate the superiority of our method in image captioning, even compared with Supervised Fine-Tuning. Especially, Re$^3$Cap outperforms GRPO with an average improvement of 8.64% in relation reasoning on the COCO-LN500 benchmark.

[83] OmniAssistBench: Assistant-style Interaction Benchmark for Omni-LLMs cs.CVPDF

Xianyun Sun, Chaoyou Fu, Zhengye Zhang, Feiyang Duan, Qingyuan Cao

TL;DR: 本文提出了OmniAssistBench，这是一个用于评估全模态大语言模型（Omni-LLMs）作为实时视频助手交互能力的基准测试。它通过逆向工程现有互联网视频，构建了模拟连续交互的多轮对话数据集，以解决静态离线数据集无法适应动态交互评估的挑战。

Details

Motivation: 动机在于评估Omni-LLMs作为主动交互式视频助手的潜力，这类助手需要结合视觉状态、用户目标和先验知识来提供有效帮助，而传统的被动视频理解评估方法无法适应动态、不可预测的交互路径。

Result: 在提出的基准测试上，专有模型Gemini-3-Pro得分为66.4（满分100），开源模型Qwen3-Omni-Instruct得分为51.2。结果表明，当前模型虽然能理解用户输入，但在处理视觉提示（如手势）、维持多轮对话的历史上下文以及延迟响应直到目标事件出现等方面存在明显不足。

Insight: 创新点在于构建了一个通过逆向工程视频来模拟真实、动态交互的评估基准，并引入了预定义先验来约束交互路径，从而实现对助手式交互能力的系统化、可重复评估。这为未来开发更可靠的交互式AI助手指明了改进方向。

Abstract: Recent omni-modal large language models (Omni-LLMs) show great potential as real-time video assistants, which continuously perceive environments and guide users to achieve specific goals. Unlike traditional passive video understanding, interactive assistants should actively combine visual states, user goals, and prior knowledge to provide effective help. Evaluating this is rather challenging, as the model’s unpredictable response dynamically changes the user’s subsequent actions, which static offline datasets cannot accommodate. To address this bottleneck, we introduce OmniAssistBench. To solve the issue of diverging interaction paths where the same user goal can be achieved through various methods, we provide models with predefined priors derived from the source video, requiring them to guide users along the exact same routes. Since real interaction videos are rare, we construct the dataset by reverse-engineering existing Internet videos. We deduce logical user goals and segment the videos into multi-turn clips to simulate continuous interactions. This rigorous pipeline required over 1000 expert person-hours to build the dataset. Results show that the proprietary Gemini-3-Pro reaches 66.4 out of the max point of 100, while the open-source Qwen3-Omni-Instruct achieves 51.2. Although current models generally understand user inputs, they frequently provide incorrect or incomplete answers. Specifically, they struggle with visual prompts (e.g., hand gestures), fail to maintain historical context during multi-turn interactions, and fail to delay response until the target event. Results indicate substantial room for improvement before models can become reliable assistants.

cs.MM [Back]

[84] When Generated Images Look Right and Retrieve Wrong: Coverage-Guided Cross-Scale Re-Indexing for Knowledge-Faithful Generative Perception cs.MM | cs.AI | cs.CV | cs.GRPDF

Guangyuan Dong, Chuang Liu, Yangchen Zeng, Haoyu Wang, Xiaoyang Yu

TL;DR: 本文提出CERES框架，针对多模态信息系统中生成图像因尺度差异导致的语义崩溃问题，通过构建三级语义金字塔、基于共现感知的路由器挖掘隐式概念、执行尺度路由交叉注意力生成，并利用冻结视觉语言模型重索引验证覆盖度，实现了在保持像素保真度的同时提升概念查询检索能力。

Details

Motivation: 解决现有语言引导生成模型在处理多尺度场景实体时，因使用全局池化文本嵌入而丢失尺度特定概念，导致生成图像虽像素保真度高但无法被原始查询检索的问题，即语义崩溃现象。

Result: 在四个全色锐化基准测试的七个设置中，CERES取得了新的最先进性能，尤其在尺度变化最极端的场景下提升最大；概念查询检索Recall@5提升14.0个百分点，图文平均倒数排名提升0.19，优于最强基线。

Insight: 创新点在于提出闭环多模态索引框架，通过语义金字塔和尺度路由交叉注意力机制显式处理多尺度概念，并利用可微软Jaccard覆盖度目标在非退化条件下优化轻量生成器，确保生成内容保持可查询性而非仅自指征一致性。

Abstract: Multimodal information systems increasingly route generated visual content back through the same vision-language index that informed its production, so the output must remain retrievable by the queries it was meant to serve. When the scene contains entities at vastly different scales, existing language-guided generators condition on a single, globally pooled text embedding and quietly drop scale-specific concepts, breaking concept-query retrieval even when pixel fidelity is high. We formalise this failure as semantic collapse and propose CERES, a closed-loop multimodal indexing framework that builds a three-level semantic pyramid, mines implicit concepts via a co-occurrence-aware router, performs scale-routed cross-attention into a lightweight U-Net generator, and verifies coverage by re-indexing the generated image with the same frozen VLM. A continuously differentiable soft-Jaccard coverage objective returns dense gradients to the 0.39M-parameter generator under explicit non-degeneracy conditions, and coverage is verified by an independent DINOv2 linear probe trained only on external scene and object labels. On four pansharpening benchmarks across seven settings, CERES delivers the new state of the art with the largest gains where scale variation is most extreme. It also improves concept-query retrieval Recall@5 by +14.0 points and image-text mean reciprocal rank by 0.19 over the strongest baseline, showing that the closed loop preserves queryable content rather than self-referential feature consistency.

eess.IV [Back]

[85] Frozen CLIP Priors for Robust Self-Supervised Poisson Inverse Problems eess.IV | cs.CVPDF

Laura C. Diaz-Delgado, Emmanuel Martinez, Henry Arguello

TL;DR: 本文提出了一种基于冻结CLIP先验的自监督Poisson逆问题求解方法，通过ADMM启发的展开式即插即用求解器，将闭式数据一致性更新与参数高效的先验解耦。该方法利用冻结的CLIP RN50密集多尺度特征构建轻量级解码器作为先验，并结合GR2R测量域重破坏和等变成像正则化进行自监督训练。在Poisson CFA去马赛克和去模糊任务上，该方法展示了有竞争力的重建质量、改进的鲁棒性，且自监督性能接近监督训练水平。

Details

Motivation: 解决光子受限场景下成像逆问题的自监督学习挑战，其中获取干净真实数据不切实际，且重建必须在数据集和采集偏移下保持稳定。Poisson噪声的信号相关统计特性与采样算子（如CFA马赛克）交互，进一步放大了这一挑战。

Result: 在Poisson CFA去马赛克和去模糊实验中，该方法表现出有竞争力的重建质量，在分布偏移下具有改进的鲁棒性，且自监督性能接近监督训练水平。

Insight: 创新点包括：利用冻结的CLIP基础视觉编码器（RN50）的密集多尺度特征构建参数高效的先验，无需昂贵微调即可实现跨域泛化；结合GR2R测量域重破坏和等变成像正则化进行自监督训练，增强稳定性；ADMM启发的展开式求解器将数据一致性与先验解耦，提升可解释性和效率。

Abstract: Self-supervised learning for imaging inverse problems is increasingly important in photon-limited settings, where acquiring clean ground truth is impractical and reconstruction must remain stable under dataset and acquisition shifts. This challenge is amplified under Poisson noise, whose signal-dependent statistics interact with sampling operators (e.g., CFA mosaicing). Meanwhile, foundation vision encoders trained at web scale offer distortion-invariant, content-related representations that generalize well across domains, suggesting a promising route to build priors that transfer beyond the training distribution without expensive fine-tuning. This paper proposes an ADMM-inspired unrolled plug-and-play solver for Poisson inverse problems that decouples a closed-form data-consistency update from a parameter-efficient prior. The prior is implemented as a lightweight decoder operating on frozen CLIP RN50 dense multi-scale features, adapting foundation representations with less trainable parameters. For self-supervision, the method integrates GR2R measurement-domain re-corruption with an Equivariant Imaging regularizer via virtual acquisitions. Experiments on Poisson CFA demosaicing and deblurring show competitive quality, improved robustness under shifts, and self-supervised performance approaching supervised training.

[86] Sparse Light Field Sampling Improves Casual 3D and 4D Reconstruction eess.IV | cs.CVPDF

Shamus Li, Ruiming Cao, Laura Waller, Kristina Monakhova, Sara Fridovich-Keil

TL;DR: 本文探讨了消费级多视角相机（如智能手机、立体相机和光场相机）在单次曝光中同步捕获多个视点的优势，并分析了传感器受限和曝光受限两种多视角场景。研究通过新构建的数据集评估了稀疏视角3DGS和4DGS基线，发现即使基线较短，使用多摄像头也能显著提升单次曝光、少样本和日常视频场景下的重建质量。

Details

Motivation: 当前新视角合成流程通常仅使用单目流，依赖相机运动或学习先验来获取角度覆盖，而忽略了消费设备上可用的多视角数据。本文旨在探究为何不使用多个视点，并分析多视角采样在重建中的价值。

Result: 实验表明，在固定传感器预算下，尽管空间分辨率较低，但角度采样在曝光稀缺时仍能改善重建质量；单次曝光和动态场景中，多视角带来的角度多样性对恢复场景几何和运动尤为关键，重建质量显著提升。

Insight: 创新点在于系统分析了消费级多视角相机的稀疏采样潜力，并验证了角度多样性对3D和4D重建的重要性，特别是在单次曝光和动态场景中，这为利用现有设备提升重建性能提供了新思路。

Abstract: Many consumer smartphones, stereo cameras, and light field cameras record multiple synchronized viewpoints in a single exposure event. However, novel view synthesis pipelines commonly use only a monocular stream and rely on camera motion or learned priors to obtain angular coverage. In this paper, we ask: why do we use only one viewpoint? We analyze sensor-limited multi-view, where one sensor trades off spatial and angular resolution, and exposure-limited multi-view, where multiple sensors on one commodity device observe each event simultaneously. We introduce a new dataset incorporating three types of commodity multi-view cameras, and evaluate sparse-view 3DGS and 4DGS baselines measuring reconstruction quality as a function of number of exposures and angle between extreme views. Our results demonstrate that using multiple cameras, even with a low baseline, significantly improves reconstruction quality in single-shot, few-shot, and casual video settings. In addition, under a fixed sensor budget, angular sampling improves reconstruction when exposures are scarce despite lower spatial resolution. The gains are most pronounced for single-shot and dynamic scenes, where a stationary monocular camera lacks the angular diversity to recover scene geometry and motion.

[87] Toward Vision Language Model-based Assessment of Clinical Quality and Usability of LGE-MR Images for Cardiac Ablation Planning eess.IV | cs.CVPDF

Bipasha Kundu, Abhishek Chaturvedi, Axel W. E. Wismueller, Richard Simon, Cristian A. Linte

TL;DR: 本文提出了一种基于视觉语言模型（VLM）的两阶段框架，用于对左心房晚期钆增强磁共振成像（LGE-MRI）进行临床图像质量评估。该框架首先生成结构化的放射学风格质量报告，预测五个放射科医生定义的标准，然后通过GPT推理模块将报告映射为结构化质量评分和用于消融规划的二元临床可用性决策。

Details

Motivation: 解决在房颤患者消融规划中，LGE-MRI图像质量（如噪声、运动伪影）的自动化、可解释评估缺失问题。当前手动评估主观且难以扩展，而现有自动化方法仅提供无临床解释的标量分数。

Result: 在包含60个标注图像-文本对的数据集上对四种SOTA VLM架构进行了基准测试。InternVL2在标准级别准确率上表现最佳（平均ACC=0.65，PLCC=0.79），而DeepSeek在临床可用性决策上与专家达成完美一致（Acc=1.00，kappa=1.00）。

Insight: 创新点在于将图像质量评估（IQA）分解为可解释的放射学标准预测和后续的临床决策推理两阶段流程，结合了VLM的视觉理解和LLM的推理能力，为安全关键的医疗决策提供了透明、结构化的自动化评估方法。

Abstract: LGE cardiac MRI is widely used for left atrial fibrosis assessment and ablation planning in atrial fibrillation patients as knowledge of fibrotic tissue regions identified from LGE-MRI is critical for catheter ablation. Often, poor quality images used during ablation planning can cause mis-localization of ablation targets, directly impacting procedure safety and outcome. The decision of whether a scan meets the minimum quality threshold for ablation planning is currently made informally by the reviewing radiologist and is not captured by any automated system, yet it is arguably the most safety-critical output of the image quality assessment (IQA) process. However, variations in image quality caused by noise, motion artifacts, and poor boundary definition significantly compromise the reliability of downstream segmentation and clinical decision-making tasks. Manual quality assessment by expert radiologists is subjective and difficult to scale, while existing automated methods produce scalar scores without interpretable clinical reasoning. In this work, we propose a two-stage vision language model (VLM) framework for clinically grounded image quality assessment of left atrial LGE-MRI. In the first stage, a fine-tuned VLM generates structured radiology-style quality reports predicting five radiologist-defined criteria: Noise, Motion Artifact, LA Boundary Accuracy, PV Region Accuracy, and Under-segmentation Severity. In the second stage, a GPT-based reasoning module maps the predicted quality and reports to a structured quality scores and binary clinical usability decision for ablation planning. We curate a dataset of 60 annotated image slice-text pairs from 20 patients and benchmark four state-of-the-art VLM architectures. InternVL2 achieves the highest criterion-level accuracy (Avg ACC=0.65, PLCC=0.79), while DeepSeek achieves perfect clinical usability agreement (Acc=1.00, kappa=1.00).

cs.CR [Back]

[88] Privacy-Preserving Object Detection for Vision Transformer-Based Models cs.CR | cs.CVPDF

Homare Sueyoshi, Kiyoshi Nishikawa, Hitoshi Kiya

TL;DR: 本文提出了一种新颖的隐私保护目标检测方法，首次将感知加密技术应用于目标检测任务，以保护测试图像中的敏感视觉信息。该方法利用Vision Transformer的嵌入结构和带密钥的域适应技术，在ViTdet模型上验证了其有效性，能在保护隐私的同时达到与无保护模型几乎相同的检测精度。

Details

Motivation: 解决现有视觉信息保护研究主要集中于图像分类任务，而缺乏针对目标检测任务的隐私保护方法的问题。

Result: 在基于ViT的目标检测模型ViTdet上进行的实验表明，该方法在准确性和视觉保护方面均有效，其检测精度几乎与未加保护的模型相当。

Insight: 创新点在于首次将感知加密引入目标检测领域，并巧妙结合Vision Transformer的嵌入结构和带密钥的域适应技术来实现隐私与性能的平衡。

Abstract: We propose a novel object detection method that enables us to protect sensitive visual information of test images. Previous studies considering visual information protection focus on image classification tasks. This paper proposes an object detection method using perceptual encryption for the first time. The proposed method can achieve almost the same accuracy as that of models without any protection by utilizing the embedding structure of the Vision Transformer (ViT) and a domain adaptation technique with keys. In experiments, the effectiveness of the proposed method is verified in terms of accuracy and visual protection under the use of ViTdet, which is a ViT-based object detection model.

cs.IR [Back]

[89] KoViDoRe: Korean Visual Document Retrieval cs.IR | cs.CVPDF

Yongbin Choi, Yongwoo Song, Mujeen Sung

TL;DR: 本文介绍了KoViDoRe，一个专门针对韩语视觉文档检索的基准数据集，旨在解决现有基准在韩语复杂结构文档和多页检索场景上的不足。该数据集通过多阶段数据构建流程创建，包含多样化布局的文档，并评估了多种多模态检索模型，发现现有模型在韩语视觉文档检索上表现不佳。为此，作者还构建了一个大规模训练数据集Ko-VDR Train Public，以支持相关模型的开发。

Details

Motivation: 现有多模态检索基准主要集中于英语，对韩语视觉文档（如PDF和报告）的覆盖有限，且现有韩语资源多评估单页检索，无法捕捉需要跨多页证据聚合的真实场景。

Result: 在KoViDoRe基准上评估了多种多模态检索模型，发现当前模型在处理韩语视觉文档检索时表现不佳，尤其是在涉及结构化内容和多样化查询类型的情况下。

Insight: 创新点在于构建了首个专门针对韩语视觉文档检索的基准数据集KoViDoRe，并通过多阶段数据构建流程（包括结构化文档解析、基于摘要和上下文的合成查询生成以及人工验证的相关性映射）确保数据质量；同时，还提供了大规模训练数据集Ko-VDR Train Public，为韩语视觉文档检索模型的开发提供了统一资源。

Abstract: Recent advances in multimodal retrieval have improved the ability to retrieve information from visually rich documents such as PDFs and reports. However, existing benchmarks remain largely centered on English and provide limited coverage of Korean visual documents with complex structures. Furthermore, most existing Korean resources primarily evaluate single-page retrieval, failing to capture realistic scenarios that require evidence aggregation across multiple pages. To address these gaps, we introduce KoViDoRe, a benchmark for Korean visual document retrieval. The dataset is constructed from publicly available Korean documents with diverse layouts, including tables, figures, and multi-column structures. We develop a multi-stage data curation pipeline consisting of structured document parsing, synthetic query generation using both summary-based and context-based strategies, and relevance mapping with human verification. Using KoViDoRe, we evaluate a wide range of multimodal retrieval models and observe that current models struggle to effectively handle Korean visual document retrieval, particularly in settings involving structured content and diverse query types. Motivated by this finding, we further curate a large-scale training dataset, Ko-VDR Train Public, to support the development of retrieval models tailored to Korean visual documents. Together, KoViDoRe and Ko-VDR Train Public provide a unified benchmark and training resource for Korean visual document retrieval.

cs.AI [Back]

[90] Why2Speak: Faithful Reasoning for Abstaining Action Policies cs.AI | cs.CLPDF

Shreya Mendi, Brinnae Bent

TL;DR: 本文研究了在多轮对话中智能体决定是否发言的干预时机问题，探讨了在行动与弃权之间做出选择时，如何确保其推理过程的忠实性（即解释真实反映决策计算）。通过使用Qwen3-8B模型，比较了直接决策、推理策略、监督微调和强化学习等方法，发现存在能力与可审计性之间的权衡：最强直接策略性能更高但无推理可检查，而推理策略提供可追溯性却以性能（特别是对真实干预机会的召回率）下降为代价。

Details

Motivation: 解决智能体系统在反复选择行动或弃权时，其推理过程的忠实性问题，以确保对决策的监督有效，因为解释只有在反映实际计算过程时才有用。

Result: 在基于Qwen3-8B的实验中，直接决策策略（无思维链推理）实现了更高的决策质量，但缺乏可检查的推理；推理策略（带思维链）提供了可审计的推理痕迹，但性能较低，尤其是对真实干预机会的召回率下降。监督微调要么抑制了推理，要么保留了推理但未改善决策质量；强化学习也未能改进推理策略。

Insight: 揭示了能力与可审计性之间的根本权衡，并指出暴露推理可能改变智能体的行动策略而非仅仅使其可观察。研究还发现标准忠实性评估方法（如基于概率的指标、受控激活探针和行为消融）可能存在局限性，例如在置信决策下饱和、易受类别不平衡和文本泄漏影响，以及混淆推理内容与推理模式变化。为此，论文提供了用于评估可行动或弃权智能体的基于推理的监督控制方法。

Abstract: Many agentic systems must repeatedly choose between acting and abstaining, making faithful reasoning important for oversight: an explanation is useful only if it reflects the computation that produced the action. We study this problem through intervention timing in multi-party conversation, where an assistant must decide whether to speak or remain silent. This setting exposes class imbalance, asymmetric action costs, and the possibility that exposing reasoning changes the policy being audited. Using Qwen3-8B, decoded with or without chain-of-thought reasoning, we compare direct decision policies, reasoning policies, supervised fine-tuning, and reinforcement learning. We find a capability-auditability tradeoff: the strongest direct policy achieves higher quality but exposes no reasoning to inspect, while the reasoning policy provides a trace at the cost of lower performance, particularly recall of true intervention opportunities. Supervised fine-tuning either suppresses reasoning or preserves it without improving decision quality, while reinforcement learning also fails to improve the reasoning policy. We identify one mechanism underlying this failure: group relative objectives provide no learning signal on confidently wrong prompts when sampled rollouts all select the same action. Controlled activation probes and behavioral ablations show that standard faithfulness methods can overstate evidence that exposed reasoning reflects the underlying decision process. Probability-based metrics saturate under confident decisions, probes are vulnerable to class imbalance and textual leakage, and reasoning ablations can confound reasoning content with changes in inference mode. Together, these results show that exposing reasoning can change an agent’s action policy rather than simply make it observable. We provide controls for evaluating reasoning-based oversight of agents that can act or abstain.

[91] Calibrating Criterion Revision in LLM Agents: Failure Modes and a Trace-Anchored Protocol cs.AI | cs.CLPDF

Guodong Xu

TL;DR: 本文研究了LLM智能体中准则修订（criterion revision）的归因问题，即当初始准则K0接受了一个违反更广泛承诺B的结果时，如何判断系统是否形成并持续使用了新准则K1。作者提出了一个包含五个非补偿性条件的评估框架，并在CMB-0.1基准上对多个模型和机制进行了测试，发现没有模型能完全满足所有条件。基于这些失败模式，论文进一步提出了一个更严格的、基于执行痕迹的CMB-0.4前瞻性协议。

Details

Motivation: 动机在于解决LLM智能体在失败后或跨任务迁移时，如何归因其准则修订行为的问题。具体来说，当智能体最初的评估准则（K0）接受了一个违反其更广泛承诺（B）的结果时，需要确定哪些观察能证明系统确实形成并持续使用了修订后的新准则（K1）。

Result: 在CMB-0.1基准上，对12个跨领域案例和4种机制（无状态推理、仅追加历史、模型生成但由框架提交的状态、评估者编写的预言机状态）进行了评估。共进行了84次确定性评分试验和192次模型-案例-机制试验。结果显示，没有任何模型试验能同时满足所有五个条件。具体地，Qwen2.5-7B模型在没有任何修订状态的情况下回答了所有迁移和保留项目，暴露了零状态重构的问题。

Insight: 论文的主要创新点在于提出了一个用于评估LLM智能体准则修订能力的、包含五个非补偿性条件的严格测量框架（CMB-0.1），并基于首次实证诊断的失败模式，设计了一个更具区分度的、基于执行痕迹的前瞻性协议（CMB-0.4）。该协议要求隐藏式迁移、显式的WRITE/NO-WRITE/ESCALATE动作、独立记录的策略选择提交、匹配的干预、重复的隐藏项目以及冻结的可执行预言机，为未来测试提供了更严谨的方法论。从客观角度看，这项工作将LLM智能体的“准则修订”这一抽象能力操作化为可测量的具体条件，并揭示了当前模型在此能力上的普遍不足，推动了该领域评估方法的发展。

Abstract: Language-model agents can improve after failure or carry text across episodes without revising what counts as success. We study the narrower attribution problem of criterion revision: when criterion K0 accepts an outcome violating a broader commitment B, what observations justify saying that the system formed and persistently used K1? We require five non-compensatory conditions: criterion-failure detection, a model-emitted proposal, new-episode transfer, intervention sensitivity on the claimed carrier, and preservation. We evaluate CMB-0.1 on twelve cross-domain cases and four arms: stateless inference, append-only history, model-generated but harness-committed state, and evaluator-written oracle state. Seven mechanism fixtures yield 84 deterministic scorer trials; four local quantized artifacts yield 96 calls and 192 model-case-arm trials. No model trial satisfies all five conditions, but this zero does not establish general capability absence. Eleven calls remain invalid after one retry; several commitments disclose the target distinction; the harness performs commits; deletion reuses a stateless call; and conflict changes multiple factors. Qwen2.5-7B answers every transfer and preservation item without revision state, exposing zero-state reconstruction. These failures make CMB-0.1 an instrument-calibration result rather than a model ranking. We derive a prospective, trace-anchored CMB-0.4 protocol requiring concealed transfer, explicit WRITE/NO-WRITE/ESCALATE actions, a separately logged policy-selected commit, matched interventions, repeated hidden items, and a frozen executable oracle. It is a successor design, not a completed confirmatory result. The paper contributes a measurement chain, an empirical diagnosis of its first implementation, and a more discriminating protocol for future tests of criterion revision.

[92] Enhancing LLMs in Predictive Political QA with Semi-Structured Data cs.AI | cs.CL | cs.IRPDF

Yinan Liu, Zihan Zhou, Zichun Jin, Xinyu Wang, Bin Wang

TL;DR: 本文提出了一种名为PSL的双视角框架，旨在增强大型语言模型（LLMs）在预测性政治问答（如预测政治人物投票行为）中的表现。该框架通过将半结构化的政治记录转化为面向推理的证据，利用语义视角提取问题相关人物的立场信号，并通过向量视角从人物交互图中学习结构感知的人物表征，从而有效建模预测相关的信号。

Details

Motivation: 现有LLM增强方法（如基于人物档案的模拟和知识图谱证据注入）主要将外部资源视为知识证据，未能充分建模预测性政治问答中的关键信号，特别是人物立场和人物间的高阶结构依赖关系。

Result: 在三个真实世界数据集和多个LLM上的实验表明，PSL框架持续优于基线方法，消融实验证实了立场信号和结构信号带来的互补性增益。

Insight: 创新点在于识别并建模了预测性政治问答中两个互补的信号：捕捉特定议题偏好的立场信号和捕捉政治人物间间接依赖关系的高阶结构信号，并提出了一个结合语义和向量双视角的框架来有效提取和融合这些信号，超越了传统仅将外部资源作为知识证据的方法。

Abstract: Predictive political question answering (QA), such as predicting how a political actor will vote, goes beyond factual lookup. External political resources offer rich historical evidence, but rarely contain the answer itself. Existing LLM augmentation methods, including actor-profile-based simulation and knowledge graph evidence injection, improve political reasoning but largely treat external resources as knowledge-based evidence, leaving prediction-relevant signals under-modeled. We identify two complementary signals for predictive political QA: actor stances that capture issue-specific preferences, and high-order structure signals that capture indirect dependencies among political actors. We propose PSL, a dual-view framework that converts semi-structured political records into inference-oriented evidence for LLMs. PSL extracts stance signals from question-relevant actor records in a semantic view, and learns structure-aware actor representations from an actor interaction graph in a vector view. Across three real-world datasets and multiple LLMs, PSL consistently outperforms baselines, with ablations confirming the complementary gains of stance and structure signals.

[93] StateSight: Benchmarking Latent Spatial-State Reconstruction in Vision-Language Models cs.AI | cs.CVPDF

Michelle Lin

TL;DR: 本文介绍了StateSight基准测试，用于评估视觉语言模型从单张图像重建潜在空间状态的能力。该基准包含三个任务：立方体网对立面推理、遮挡立方体塔计数和四邻域连通分量计数。实验结果显示，当前领先的VLM模型（如GPT-5.5和Claude Sonnet 5）在这些任务上的表现远低于人类基线，表明模型在空间结构重建和推理方面存在显著缺陷。

Details

Motivation: 当前视觉语言模型的多模态问答评估往往混合了感知、OCR、领域知识、语言先验和推理等多种能力，难以孤立地评估其从单张图像重建潜在空间结构这一核心能力。

Result: 在StateSight基准上，GPT-5.5在三个任务上的准确率分别为59.3%、33.3%和28.3%，Claude Sonnet 5为53.3%、18.7%和7.3%，而30人的人类基线平均准确率分别为80.8%、68.8%和64.3%，人类表现全面超越模型。所有模型输出均无格式错误，但准确率揭示了其空间推理的不足。

Insight: 论文的创新点在于提出了一个程序化生成的、专注于隔离评估VLM空间状态重建能力的基准（StateSight）及其附带的分步推理数据集（StateSight-Steps）。其核心洞察是，格式正确的模型响应可能掩盖了其在恢复可验证视觉推理所需的空间结构方面的根本性失败，这为诊断和改进VLM的空间推理能力提供了新的工具和视角。

Abstract: Vision-language models are increasingly used for multimodal question answering, yet their ability to reconstruct latent spatial structure from a single image remains difficult to isolate. Broad benchmarks often combine perception, optical character recognition, domain knowledge, linguistic priors, and reasoning in the same evaluation. We introduce StateSight, a procedurally generated benchmark for cube-net opposite-face reasoning, occluded cube-tower counting, and 4-neighbor connected-component counting. Each task family contains 300 single-image prompts with deterministic oracle labels and exact-match scoring. OpenAI GPT-5.5, using the API model identifier gpt-5.5, achieved 59.3%, 33.3%, and 28.3% accuracy across the three tasks, while Claude Sonnet 5 achieved 53.3%, 18.7%, and 7.3%. All final direct runs had zero format errors. A 30-participant human baseline on 60 items exceeded both models on every task, with mean accuracies of 80.8%, 68.8%, and 64.3%. Visible-derivation analysis identified recurring errors in image-state reconstruction and reasoning procedure. We also introduce StateSight-Steps, a companion dataset of 900 interleaved image-text examples and 3,600 deterministic intermediate visual states. The results show that format-valid responses can mask failures to recover the spatial structure required for verifiable visual inference.

Yibo Hu, Yu Qian, Mao Gu, Yingfan Tao, Yuhao Chen

TL;DR: TLive-Omni是一个专为电商直播场景设计的全模态理解模型，能够处理语音、视频帧、产品图像、叠加文本和用户查询等多种模态的嘈杂、长时数据流。它通过Per-vGrid时间戳令牌组织实现时序对齐，并采用三阶段监督训练（从全模态感知到指令跟随响应）和Faithful-RFT强化微调来提升答案忠实度和表达质量。模型还配备了面向场景的原子能力分类法和紧凑数据生产引擎，以支持可扩展训练。

Details

Motivation: 电商直播需要理解分布在多种模态（如语音、视频、文本）中的产品信息，且数据流具有噪声大、时序长的特点，现有模型难以有效处理这种复杂的全模态、长时分析任务。

Result: 在电商直播基准测试中，模型在直播电商领域任务上表现出色，同时在通用基准测试上也展现出优秀的泛化能力。

Insight: 创新点包括：1) Per-vGrid时间戳令牌组织，将视频网格与对应时间音频分组以促进时序对齐；2) 三阶段渐进式训练流程（全模态感知到指令跟随）结合Faithful-RFT强化微调，直接利用任务可验证反馈优化最终响应而非推理探索；3) 面向场景的原子能力分类法与紧凑数据生产引擎，高效生成多任务训练信号；4) 可扩展训练技术如同步长度分组采样器和轻量动态采样策略，减少填充并维持GRPO的有效相对优势。

Abstract: E-commerce live streaming requires omni-modal understanding of noisy, temporally extended streams, where product facts are distributed across speech, video frames, product images, overlaid text, and user queries. We present TLive-Omni, an omni-modal understanding model tailored to live-commerce scenarios. It maps image, video, audio, and text inputs into a unified representation space. For long-form live streaming analysis, we introduce Per-vGrid, a timestamped token organization that groups each video grid with its temporally corresponding audio within explicit boundary tokens to facilitate temporal alignment. We design a three-stage supervised training recipe that progressively develops live-commerce understanding, from omni-modal perception to instruction-following responses. We then propose Faithful-RFT, a reinforcement fine-tuning stage that further improves answer faithfulness and expression quality while meeting real-time demands, scoring final responses directly with task-verifiable feedback rather than optimizing for reasoning-style exploration during rollout. Moreover, TLive-Omni is supported by a scenario-oriented atomic capability taxonomy and a compact data production engine that converts live-commerce audio, image, and video streams into training signals for speech recognition, speaker analysis, product visual grounding, text recognition, temporal grounding, video dense caption, and omni-modal QA, etc. For scalable training, a synchronized length-grouped sampler reduces padding while preserving comparable workloads across workers, while a lightweight dynamic sampling strategy regenerates rollout groups with near-zero reward variance to maintain meaningful relative advantages for GRPO. Experiments on e-commerce live streaming benchmarks demonstrate strong performance across live-commerce domain tasks, together with excellent generalization on general benchmarks.

[95] Anatomy-Informed Neural Networks: Encoding Anatomic Priors in Loss and Architecture, with an SE(3) Formulation of Guidewire-Induced Aortoiliac Deformation cs.AI | cs.CV | cs.ROPDF

David P. Stonko

TL;DR: 本文提出解剖学信息神经网络（AINN），通过将软解剖学先验作为损失函数中的惩罚项（如分支惩罚），并将硬解剖学先验（如血管连续性）融入网络架构和状态表示，确保模型预测在解剖学上合理。该方法应用于临床数据有限的场景：模拟导丝引起的主动脉-髂动脉树形结构变形，利用SE(3)李群表示血管中心线和导丝路径，结合Cosserat杆模型和解剖锚定血管的接触力学，通过Wasserstein-2最优传输损失从二维血管造影训练三维预测。

Details

Motivation: 解决深度学习解剖模型在数据稀缺时可能产生解剖学上不可能预测、泛化能力差的问题，为血管手术和自主血管内导航提供可靠变形模拟。

Result: 方法在已知真实数据上验证了运动学、损失和投影部分；力学求解器仅基于自身最优性条件验证，预测位移尚未达到网格收敛，目前未训练网络，未来计划迁移到真实CT扫描以评估预测准确性和数据需求。

Insight: 创新点在于将解剖学先验分为软硬两类分别嵌入损失函数和网络架构，利用SE(3)李群和Cosserat杆模型耦合力学表示，以及通过二维血管造影监督训练三维预测的跨维度学习框架。

Abstract: Deep-learning models of anatomy can be numerically plausible yet anatomically impossible, and they generalize poorly when data are scarce. We introduce Anatomy-Informed Neural Networks (AINN), in which soft anatomic priors enter as penalty terms in the loss (e.g., a branching penalty that treats a renal transplant artery off the iliac instead of the aorta as unexpected rather than impossible), in direct analogy to a physics-informed neural network, and hard anatomic priors (e.g., continuity of the vessel) are built into the architecture and state representation, making such invalid predictions impossible by construction wherever the prior admits architectural enforcement. We develop it on a clinical test case with limited data: how the aortoiliac tree deforms when a stiff wire is introduced endoluminally. This is important to contemporary aortic surgery and will matter to autonomous endovascular navigation. We lift the vessel centerline and the wire path from R^3 to curves of frames in the Lie group SE(3), and couple a Cosserat-rod wire to a tortuosity-modulated, anatomically anchored vessel through a unilateral lumen-contact inequality. The prediction is a constrained minimizer of the coupled elastic energy, with contact forces as its Lagrange multipliers. Supervision is a Wasserstein-2 optimal-transport loss between the predicted projection through the C-arm geometry and the observed angiogram, so a 2D angiogram can train a 3D prediction. The kinematics, loss and projection are verified against known ground truth; the mechanics solver only against its own optimality conditions, and predicted displacement is not yet mesh-converged. Here, no network is trained. Future work will transfer this in silico model to real CT scans and test whether it improves predictive accuracy and reduces the training data required.

cs.RO [Back]

[96] IMU-Free Body-Frame State Estimation with Sparse Scene Flow for Quadcopters cs.RO | cs.CVPDF

Daniel Grønhaug, Sofie Markeset, Mathias Kolberg

TL;DR: 本文提出了一种仅依赖视觉的X型四旋翼无人机状态估计系统，该系统无需惯性传感器，仅使用同步立体图像和电机推力指令。系统在机体坐标系下运行，通过连续-离散扩展卡尔曼滤波器估计姿态、速度、角速度、重力和扰动，并利用静态场景点作为隐式惯性参考。同时，系统还生成携带位置、速度和联合协方差的稀疏3D点云，用于下游世界模型。

Details

Motivation: 解决四旋翼无人机在缺乏GPS、IMU或任何世界坐标系基础设施情况下的状态估计问题，旨在实现仅依赖视觉的自主导航，降低对硬件传感器的依赖。

Result: 系统在机体坐标系下实现了状态估计，并生成了稀疏场景流和3D点云。虽然没有提及具体定量结果或基准测试，但方法设计为下游世界模型提供测量源，支持未来与GPS、IMU等传感器的集成。

Insight: 创新点包括：使用静态场景点作为隐式惯性参考的视觉状态估计方法；结合4视图（两个时间戳的立体对）完整束调整联合估计位置和速度；以及通过外部焦点实现空间自适应的点云密度分配，以优化计算资源。

Abstract: We present a vision-only state estimation system for X-configuration quadcopters equipped with a canonical stereo camera pair and no inertial sensors. The system operates entirely in the body frame, requiring only synchronised stereo images and motor thrust commands. A continuous-discrete extended Kalman filter on a composite manifold state $\langle SE(3), \mathbb{R}^3, \ldots \rangle$ maintains estimates of body-frame pose, velocity, angular velocity, gravity, and disturbances, using stationary scene points as implicit inertial references. Feature points are detected (FAST, Shi-Tomasi), tracked temporally (SSD, Lucas-Kanade) and matched across cameras (NCC), with search regions predicted from filter-derived pose and point uncertainty. Chi-squared gating on the normalised innovation admits only stationary points to the filter. The system also produces a sparse 3D point cloud carrying per-point position, velocity and joint covariance. These come from a 4-view (two stereo pairs at two timestamps) full bundle adjustment that jointly estimates position and velocity from stereo disparity and temporal parallax, with the filter-derived relative pose as a prior. Feature points in the EKF do not enter the solver; their information is reflected through the pose prior. Point cloud density is spatially adaptive: an external focus point directs allocation, producing dense coverage in the region of attention and sparse coverage elsewhere. The output is a body-frame state estimate, a calibrated pose change, and a sparse scene flow. It is intended as a measurement source for a downstream world model anchored in the current body frame, without dependence on GPS, IMU, or any world-frame infrastructure, though the architecture accommodates their future integration.

[97] VT-MUSE: Multimodal Unified Sequential Visuotactile Representation Learning for Manipulation cs.RO | cs.CVPDF

Congsheng Xu, Qiaochu Yang, Fangyuan Shi, Yifan Han, Baijun Chen

TL;DR: VT-MUSE是一个用于视触觉操作的多模态统一序列表示学习框架。它通过两阶段学习解决现有方法中视觉与触觉模态独立编码、缺乏细粒度跨模态依赖以及忽略接触时间演化的问题。第一阶段通过跨模态时间对齐和掩码视图一致性联合适应模态特定编码器；第二阶段使用条件变分潜变量模型处理掩码视觉序列和完整触觉历史，并通过辅助解码器重建视觉观测和预测触觉深度变化，以保留全局视觉上下文和局部接触动态。学习到的表示通过门控交叉注意力集成到轻量级Transformer策略中。

Details

Motivation: 现有方法通常在融合前独立编码视觉和触觉观测，限制了捕捉细粒度跨模态依赖的能力，且大多关注当前时间步的观测而忽略了接触的时间演化。

Result: 在仿真基准测试中，VT-MUSE在所有任务上比最强基线高出11个百分点，并在真实世界实验中实现了显著改进。

Insight: 创新点包括两阶段表示学习框架：第一阶段通过跨模态时间对齐和掩码视图一致性进行联合适应；第二阶段利用条件变分潜变量模型和辅助重建任务（视觉重建和触觉深度预测）来学习同时保留全局视觉上下文和局部接触动态的统一表示，并通过门控交叉注意力将表示集成到策略中。

Abstract: We propose VT-MUSE, a Multimodal Unified SEquential representation learning framework for visuotactilemanipulation. Existing approaches often encode visual and tactile observations independently before fusion, limiting their ability to capture fine-grained cross-modal dependencies. Moreover, most methods focus on observations at the current time step and overlook the temporal evolution of contact. VT-MUSE addresses both limitations through a two-stage representation learning framework. In Stage I, modality specific encoders are jointly adapted via cross-modal temporal alignment and masked-view consistency. In Stage II, a conditional variational latent model processes masked visual sequences together with full tactile histories. Auxiliary decoders reconstruct the masked recent visual observations and predict tactile depth changes, encouraging the latent representation to retain both global visual context and local contact dynamics. The learned representation is subsequently integrated into a lightweight Transformer policy through gated cross-attention. On the simulation benchmark, VT-MUSE outperforms the strongest baseline evaluated on all tasks by 11 percentage points and also achieves substantial improvements in real-world experiments.

cs.LO [Back]

[98] ProofJudge: Tool-Grounded LLM Evaluation of Formal Proof Quality in Mathlib cs.LO | cs.AI | cs.CLPDF

Shane Caldwell

TL;DR: 本文介绍了ProofJudge，一个基于工具增强的LLM评估系统，用于评估Lean 4中形式化证明的质量，涵盖正确性之外的五个维度：库利用、自动化适配、结构清晰度、陈述质量和Mathlib规范。该系统在一个包含218个声明的数据集上进行了评估，结果显示所有六个法官模型都能显著高于随机水平地恢复人类评审者的偏好。

Details

Motivation: 动机在于，尽管通过Lean 4内核类型检查的形式化证明在正确性上没有问题，但其质量仍存在显著差异，需要一种超越正确性的自动化评估方法来衡量证明的多个质量维度，以支持数学库（如Mathlib）的开发和维护。

Result: 在从Mathlib拉取请求中提取的218个声明数据集上，所有六个法官模型恢复人类评审者偏好的准确率均显著高于随机水平，范围从80.8%到63.5%，其中两个开源权重模型以最佳模型十分之一的成本达到了约70%的准确率。

Insight: 创新点在于将LLM作为法官，并通过工具访问（如查询库状态）来增强其评估能力，从而实现对形式化证明多维度质量的自动化评分；这为形式化数学和代码质量评估提供了可扩展且成本效益高的解决方案，相关数据集和工具已开源以促进进一步研究。

Abstract: Formal proofs in Lean 4 that pass the kernel’s type checker can nonetheless vary widely in quality. We introduce ProofJudge, an agentic LLM-as-judge system that scores formal proof quality along five dimensions beyond correctness: library leverage, automation fit, structural clarity, statement quality, and Mathlib conventions. We evaluate ProofJudge on a novel dataset of 218 declarations drawn from distinct Mathlib PRs. The judge agent is grounded by tool access to the commit the PR is applied to, enabling it to query the library state when scoring. A judge is considered aligned with human preferences when it rates the version of the PR Mathlib accepted above the initial version that was sent back for revision. All six judge models evaluated recover the reviewers’ preference well above chance, from 80.8% to 63.5%, and two open-weight judges reach roughly 70% at a tenth of the best judge’s cost. We release the judge harness, evaluation dataset, and evaluation traces as open-source artifacts to support further research.

Table of Contents

cs.CL [Back]

[1] When Vocabulary Comprehension Fails Clinical Reasoning: Evaluating Therapy Bots’ Safety Risks for Generation Alpha cs.CL | cs.AI | cs.CYPDF

[2] Beyond Prompt Engineering: A Systematic Analysis of Prompt Lexical Sensitivity and Its Impacts on Quality cs.CL | cs.AIPDF

[3] Inhibitory Attention for Clinical Long-Context Reasoning: Characterizing and Mitigating Lost-in-the-Middle Effects in EHR Processing cs.CL | cs.AIPDF

[4] How to Train a Real-World Silicon Concierge? Internalizing Complex Business Workflow to Only OneModel cs.CL | cs.AIPDF

[5] The Divergence Hypothesis: Unmasking Lexical Interference and Label Bias in Mental Health NLP cs.CL | cs.AIPDF

[6] Self-Speculation for Faster Reasoning Models cs.CLPDF

[7] Toward Auto-Research: Mining Falsifiable Research Ideas from Paper Knowledge Graphs with Categorical Structure cs.CL | cs.AIPDF

[8] Multilingual Verifier Bias in RLVR: Benchmark, Rollout Diagnosis, and the Cross-Lingual Selection Bottleneck cs.CL | cs.LGPDF

[9] Hadith computational science in the age of large language models: a critical narrative review cs.CL | cs.AIPDF

[10] Trilingual Topic Modeling of Sri Lankan Parliamentary Debates cs.CL | cs.AIPDF

[11] An ambiguity taxonomy for evaluating large language model performance on clinical registry abstraction: a multi-site prospective study cs.CLPDF

[12] Decoupled Vision-Language System for Multimodal Understanding and Generation cs.CL | cs.CVPDF

[13] Poly-InstructTTS: Learning In-the-Wild Expressive Speech Synthesis from Open-Ended Instructions cs.CL | cs.AIPDF

[14] Ansari: A Retrieval-Grounded Islamic AI Assistant – Architecture, Deployment, and Lessons from 140,000 Conversations cs.CL | cs.AI | cs.CYPDF

[15] ImmigrationReason: A Structured Dataset of U.S. Immigration Appeals for Legal Reasoning Research cs.CLPDF

[16] Evaluation-as-Search: Adaptive Discovery of Grounding Failures in Meeting Assistants cs.CL | cs.AIPDF

[17] LingShu: A Large-Scale Symptom-Centric Contextualized Knowledge Graph Bridging Traditional Chinese Medicine and Modern Biomedicine cs.CL | cs.AIPDF

[18] When Failures Propagate: Causal Failure Attribution in Agentic Retrieval-Augmented Generation cs.CL | cs.AIPDF

[19] AgentMercury: Your Agent Can Synthesize Verifiable Environments for Business Scenarios at scale cs.CL | cs.AIPDF

[20] STAR-OPD: Structured Aspect-Cascade-Aware On-Policy Reward Distillation for ABSA Quadruple Extraction cs.CL | cs.AIPDF

[21] Ontology-Driven Structural Regularization for Document-Level Relation Extraction cs.CLPDF

[22] KREL: Automatic Medical Coding via Knowledge-Guided Reasoning over Clinical Evidence with LLMs cs.CL | cs.AIPDF

[23] Quantization-Aware Healing: A Practical Recipe for Recovering Compressed, 4-Bit LLMs cs.CL | cs.AI | cs.LG | cs.PFPDF

[24] Evidence-Consistent Generative Detection under Scenario-Level Distribution Shift cs.CLPDF

[25] Free-Text Evaluation of LLMs for 5G Domain Knowledge and Fault Analysis using LLM-as-Judge cs.CL | cs.AI | cs.NIPDF

[26] EnSI-RAG: Entity-Structure-Indexed Retrieval-Augmented Generation for Long-Document Question Answering cs.CL | cs.AI | cs.DB | cs.IRPDF

[27] Memory Augmentation Unlocks Efficient Chain-of-Thought Reasoning cs.CLPDF

cs.CV [Back]

[28] Aggregating Visual Information with Optimal Transport for VideoLM Token Compression cs.CVPDF

[29] RISE: Adaptive Imagination for World Action Models cs.CVPDF

[30] DiffVC-ONE: Diffusion-based Generative Video Compression with One-Step Video Diffusion Transformer cs.CVPDF

[31] Annotations as Rollouts: Efficient and Scalable Reinforcement Learning for Video MLLMs cs.CVPDF

[32] Grounded-Exo2Ego: Structured Semantic Grounding for Robust Exocentric-to-Egocentric Video Generation cs.CVPDF

[33] Keep Your Friends Close, and the Right Neighbours Closer: Disaster-Conditioned Kernel-Regularized Graph Attention for Building Damage Classification cs.CV | cs.LGPDF

[34] Learning Prostate Anatomy at Test Time for Cancer Detection in Micro-Ultrasound cs.CV | cs.LGPDF

[35] Bridging Language and Spherical Space: Object-Centric Control for Text-to-Panorama Generation cs.CVPDF

[36] Identity-Aware Human-Object Interaction Motion Captioning cs.CV | cs.AIPDF

[37] ArtiMo: Agent-Driven Articulated Mesh Animation cs.CVPDF

[38] AGIDefect-4K: A Richly Annotated Dataset for AI-Generated Image Defect Detection, Localization and Explanation cs.CVPDF

[39] AffordAny: Open-World 3D Affordance Grounding from Monocular RGB Images via Vision-Language-Guided Geometric Reasoning cs.CVPDF

[40] Generating Multi-view Adversarial Examples for Visual Geometry Grounded Transformer cs.CVPDF

[41] VisTa3D: A Dataset and Benchmark for Thin Object Reconstruction from Vision, Tactile, and 3D Point Clouds cs.CVPDF

[42] Identity-Preserving Text-to-Video Generation via Agentic Enhancement and Semantic Repair cs.CVPDF

[43] SPARK-SAM: Self-Prompt Adaptation with Response Knowledge for SAM in Infrared Small Target Segmentation cs.CVPDF

[44] Vis-Poison: Poisoning Visual Knowledge in Multimodal Retrieval-Augmented Generation cs.CV | cs.AIPDF

[45] CARD: Diagnosing Belief to Action Routing Failures in Vision Language Models cs.CV | cs.AIPDF

[46] MotionPhys: Detecting AI-Generated Videos via Physical Consistency of Optical-Flow Trajectories cs.CVPDF

[47] CertVLA: Certified Defense against Physical Visual Attacks for Vision-Language-Action Models cs.CV | cs.AIPDF

[48] TRACE: Training-time Report-guided and Clinically Ordered Concept Editing cs.CV | cs.AIPDF

[49] Routing Before Looking: Query-Adaptive Evidence Acquisition for Long-form Video Understanding cs.CVPDF

[50] Enhancing Localized Reasoning for Long Video Understanding via Efficient Segment-to-Video Supervision cs.CVPDF

[51] Identify, Locate, Link: End-to-End Key-Value Extraction from Document Images cs.CV | cs.CLPDF

[52] Multi-Modal Traffic Sign Detection with Semantic Attributes for Autonomous Driving cs.CV | cs.ROPDF

[53] EviRank: Structured Relevance Evidence for Multimodal Image Re-ranking cs.CV | cs.LGPDF

[54] A Collaborative Multi-Modality Interaction for VLA-based End-to-End Autonomous Driving cs.CV | cs.ROPDF

[55] EmotionDialogCN: A Spontaneous Multimodal Dataset for Mandarin Emotional Dialogue cs.CVPDF

[56] InfinityEdit: Infinite Video Editing with a Lightweight Edit-Ignition Adapter cs.CVPDF

[57] Semantically Compatible Knowledge Distillation for Cross-Domain Object Detection with Vision Foundation Models cs.CVPDF

[58] OccluRank: Controllable Occlusion-Aware Layout-to-Image Generation by Adding Just an Ordinal Rank cs.CVPDF

[59] GAP-SAM: A Global Artifact Prior for Generalizable AI-Generated Image Manipulation Localization cs.CVPDF

[60] SuppreSensing: Expert-Guided Feature Recalibration and Discrepancy Augmentation for Multimodal Object Detection cs.CVPDF

[61] Kinematic Knowledge Maps for Pattern Alignment: Structured Latent Representational Learning in Multimodal Gait Analysis cs.CVPDF

[62] MigrationNarrate: A Dataset for Detection of Migration Narratives in YouTube Videos cs.CV | cs.CL | cs.CYPDF

[63] WA-JEPA: Rethinking the Video JEPA Paradigm for World-Action Modeling in Autonomous Driving cs.CV | cs.AIPDF

[64] Latent Ordinal Evidence, Misaligned Outputs: Inference-Time Ordinal Lens Alignment for Multimodal LLMs cs.CVPDF

[65] Recognition-Conditioned Reasoning: A Training-Free Multimodal-LLM Pipeline for Fine-Grained Micro-Action Understanding cs.CV | cs.MMPDF

[66] CoAnchor: Robust Collaborative Perception under Spatio-Temporal Misalignment via Object-Level Anchors cs.CV | cs.AIPDF

[67] COMET: Contrastive Motion-Enhanced Temporal Reasoning for Video Multimodal Large Language Models cs.CV | cs.CL | cs.LGPDF

[68] AT-ViT: Area-Targeted Multi-View Vision Transformer with Cross-Attention and Multi-Scale Patching for Plant Trait Recognition in Herbarium Images cs.CV | cs.AIPDF

[69] Robust Validation to Geometric Perturbations for Autonomous Pose Estimation cs.CVPDF

[70] A2DINOv3: Rethinking Multi-Modal Object Detection via Socialized Collaboration cs.CV | cs.AIPDF

[71] Masking Is Not Enough: Generative Restoration for Multimodal De-Identification in Medical AI cs.CV | cs.CRPDF

[72] Llama-Mobile: Efficient 2.7-Bit Quantization of VLMs cs.CV | cs.LGPDF

[73] Stream3Dv2: Geometric-Semantic Fusion Enhanced Streaming Zero-Shot 3D Scene Understanding cs.CVPDF

[74] Human-JEPA: A Human-Centric Vision Model that Perceives and Anticipates cs.CV | cs.LGPDF

[75] Is Visual Prompting All You Need? Studying VLM Spatial Reasoning under Progressive Visual Scaffolds cs.CV | cs.AIPDF

[76] A Modular Agent for Reliable and Auditable Spatial Relation Verification in CT Scans cs.CV | cs.AIPDF

[77] Towards Investigating Residual Hearing Loss: Quantification of Fibrosis in a Novel Cochlear OCT Dataset cs.CV | cs.AIPDF