TRACE (Temporal Audit and Condition-aware Evaluation) is a new benchmark and evaluation framework for streaming video understanding that explicitly records when evidence becomes valid, how visual history is maintained, and how responses are triggered. It combines temporally audited visual tasks, evidence timing, instruction-dependent trigger annotations, a unified causal Core–Adapter protocol, and multidimensional reporting of answer quality, timeliness, response-selection behavior, workload, completion, and reliability. Using 1,240 records from 517 videos, TRACE evaluated eight publicly available models in eight configurations, revealing that similar QA accuracy can hide significant differences in completion, answer validity, generation workload, and proactive performance metrics such as response delay, false alarms, and missed target windows.
By Yibo Ma, Qianqian Zhang, Peng Liu, Tiancheng Zhao
arXiv:2607. 09322v1 Announce Type: new Abstract: In this work, we introduce LongMedBench, a real-world EHR-based benchmark for long-horizon clinical decision-making.
By Yanzhen Chen, Zihan Xu, Xiaocheng Zhang, Zhiting Fan, Weiqi Zhai, Hongxia Xu, Zuozhu Liu
arXiv:2608. 14015v1 Announce Type: cross Abstract: Understanding tens-of-minutes surgical videos requires long-horizon temporal reasoning, answering what happens before, after, or across stages of a procedure by grounding the question in visual evidence spread across time.
By Yingying Fan, Penghui Du, Leyan Zhu, Runze He, Zimeng Wu, Yuxuan Zhang, Liang Chen, Jiahao Xie, Jiangtang Wang, Shuai Shao, Anchao Yang, Yutong Bai, Yan Wang
arXiv:2606. 01961v1 Announce Type: new Abstract: Autonomous agents are increasingly expected to support end-to-end medical-AI research workflows, moving beyond isolated prediction tasks or short-form clinical question answering.
By Junqi Liu, Salena Song, Yuhan Wang, Jiawei Mao, Hardy Chen, Xiaoke Huang, Tianhao Qi, Pengfei Guo, Yucheng Tang, Yufan He, Can Zhao, Andriy Myronenko, Dong Yang, Daguang Xu, Yuyin Zhou
The paper introduces Watch-Think-Interact (WTI), a closed-loop framework for multi-question streaming video reasoning that maintains compact natural-language memory entries linked to video time ranges. WTI decides whether to answer, continue watching, or recall relevant past intervals for each question, avoiding replay of the full history. The authors build a large dataset, WTI-82K, and a training method, Stream-GDPO, achieving state‑of‑the‑art performance on StreamingBench and OVO-Bench.
By Ziheng Huang, Yicheng Bao, Xueheng Li, Zhenkun Gao, Bangwei Liu, Kunquan Li, Yuxiang Shen, Bangyan Li, Xuejiao Wang, Changbo Wang, Gaoqi He
ProactiveBench evaluates streaming video models on their ability to interact proactively, rather than reactively. It tests models at one‑second intervals without explicit cues, using six subtasks that vary trigger ambiguity, timing tolerance, and response patterns. The benchmark measures both response and silence rates, distinguishing early, in‑window, and missed responses, and penalizes omissions and repetitions.
By Kaixuan Du, Xin Wan, YuKun Wang, Hang Zhang, Meng Cao, Dai Guan, Ming Chen, Ni Li
arXiv:2605.11533v4 Announce Type: replace
Abstract: Routine clinical check-up reports combine laboratory measurements, physiological assessments, imaging findings and visually structured information,...
By Sike Xiang, Shuang Chen, Kevin Qinghong Lin, Jialin Yu, Yijia Sun, Philip Torr, Amir Atapour-Abarghouei
In clinical practice, patients often undergo multiple imaging examinations over successive visits, yielding longitudinal data. Modeling such temporal information is crucial for reliable assessment of disease progression and treatment response.
TimeBlind is a diagnostic benchmark designed to evaluate fine‑grained spatio‑temporal compositionality in video large language models (LLMs). It categorizes temporal understanding into three levels—atomic event recognition, event property characterization, and reasoning about event interdependencies—and uses a minimal‑pairs paradigm where video pairs share identical static content but differ only in temporal structure. Across 20 state‑of‑the‑art MLLMs tested on 600 curated instances, the best model achieved only 48.2% instance accuracy, far below human performance of 98.2%, highlighting a reliance on static visual shortcuts rather than true temporal reasoning.
By Baiqi Li, Kangyi Zhao, Ce Zhang, Chancharik Mitra, Jean de Dieu Nyandwi, Gedas Bertasius
OVIBench introduces the first standardized benchmark for evaluating vision‑language models on Online Video Question Answering under Interruption, a realistic setting where users can interrupt the model during answer generation. The benchmark categorizes interruptions into Cancellation, False Trigger, and Correction, supports both open‑ended and multiple‑choice tasks, and provides an offline simulation protocol plus a multi‑dimensional metric suite. Experiments show that OVIBench can distinguish models’ interruption‑handling abilities, particularly in following correction requests, and that fine‑tuning on the newly created OVI‑Train dataset yields significant performance gains.
By Naiming Liu, Zhiheng Wu, Shuning Wang, Tie Zhang, Bowen Liu, Tong Wang
AgentVidBench is a new multi‑hop video question‑answering benchmark designed to evaluate spatial, temporal, and causal reasoning in multimodal large language models (MLLMs). Unlike existing tests that focus on simple scene queries or global summaries, AgentVidBench includes step‑by‑step solution traces to assess whether agents gather the necessary evidence to justify their answers. Experiments with 12 MLLMs show limited single‑turn performance, but integrating these models into agentic workflows improves both accuracy and trajectory scores, establishing AgentVidBench as a comprehensive testbed for future research on agentic video understanding.
By Seoyeon An, Hyeonseo Jang, Minsu Kim, Chanho Lee, Younghan Park, Kangwook Lee
The paper introduces SurgFUTR, a state‑change learning framework for predicting future events in endoscopic videos. Instead of forecasting raw observations, it classifies transitions between current and future states using a teacher‑student architecture and an Action Dynamics module. The authors also present SFPBench, a benchmark with five short‑ and long‑term prediction tasks, and demonstrate consistent improvements across multiple datasets and procedures, including cross‑procedure transfer.
By Saurav Sharma, Chinedu Innocent Nwoye, Didier Mutter, Nicolas Padoy