arXiv cs.AI· Zhenqi He, Dunyuan Xu, Yang Bo, Kien T. Pham, Yanghao Wang, Shuhan Li, Xintong Guo, Yuyang He, Fang Liu, Huankang Guan, Jingyan Chen, Rui Liu, Long Chen·· 1 天前AI 评分26
MemInStream：流式视频理解该“记住”还是仅“认出”？重思记忆评测
MemInStream: Memorizing or Merely Recognizing? Rethinking Streaming Video Memory Evaluation
AI 导读
研究者提出 MemInStream，一个专门评估流式视频理解记忆能力的基准，通过证据可控验证协议为每题绑定时间局部化的证据区间，只保留模型能凭标注证据作答、却无法靠近期观察、证据后观察或纯文本回答的问题。该基准覆盖情景检索、状态跟踪与更新、多证据整合、上下文记忆四大类，旨在把记忆局限与视觉理解失败区分开，减少不依赖记忆的捷径。
正文 · 原文
Authors:Zhenqi He, Dunyuan Xu, Yang Bo, Kien T. Pham, Yanghao Wang, Shuhan Li, Xintong Guo, Yuyang He, Fang Liu, Huankang Guan, Jingyan Chen, Rui Liu, Long Chen
Abstract:Streaming video understanding aims to continuously process incoming visual observations, making the retention and retrieval of relevant information increasingly important as the video history grows. Although existing benchmarks include memory-related questions, question-answering accuracy alone does not reliably measure memory capabilities. Incorrect predictions may arise from failures in visual understanding rather than forgetting, whereas correct predictions may rely on linguistic priors or recent observations without requiring historical memory. To address these ambiguities, we introduce MemInStream, a benchmark dedicated to evaluating memory capabilities in streaming video understanding. We develop an evidence-controlled validation protocol that associates each question with temporally localized evidence intervals. The protocol validates evidence solvability and memory necessity by retaining questions that models can answer from the annotated evidence but cannot resolve using recent observations, post-evidence observations, or textual input alone. This helps disentangle memory limitations from understanding failures while reducing memory-independent shortcuts. MemInStream provides a systematic evaluation of memory capabilities through four major categories, covering Episodic Retrieval, State Tracking \& Updating, Multi-Evidence Integration, and Contextual Memory across diverse video scenarios. Overall, \textsc{MemInStream} provides a comprehensive and reliable benchmark for systematically evaluating memory capabilities in streaming video understanding.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.15640 [cs.CV] |
| (or arXiv:2611.15640v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2611.15640 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zhenqi He [view email]
[v1]
Wed, 11 Nov 2026 15:52:19 UTC (1,349 KB)
来源：arXiv cs.AI · arxiv.org