跳到正文
arXiv cs.AI· Tianci Xie, Li Chen, Pinxue Guo, Rui Zheng·· 2 小时前AI 评分42

NexSight:面向细粒度 Agentic 长视频理解的可验证任务

NexSight: Verifiable Tasks for Fine-Grained Agentic Long-Video Understanding

AI 导读

arXiv 论文提出 NexSight,构建可验证任务以评估和训练 agentic 长视频理解。其 SightForge 流水线产出 SightLine 基准,包含来自 108 个长视频的 310 个可溯源任务,最强模型仅得 54.52%,但给出标注证据后达 79.35%。

正文 · 原文

View PDF HTML (experimental)

Abstract:Fine-grained questions about long videos ask how many times an action occurs, in what order events unfold, or exactly when a state changes. Answering them can require finding brief moments scattered across hours of footage. Agentic models that decide when and where to look promise to find these moments, yet this ability is hard both to measure and to learn. Sparse supporting evidence may establish that an action occurred without establishing its total count; verifying timeline-wide claims requires checking for missed occurrences and double counting. Answer-only supervision, in turn, does not explicitly supervise the search for evidence. We present NexSight, which builds verifiable tasks to both evaluate and train agentic long-video understanding. Its agentic pipeline, SightForge, accepts a task only if a verifier blind to the draft answer can reconstruct it from evidence densely reacquired from the raw video and counterexample checks fail to refute it. This yields SightLine, 310 evidence-traceable tasks from 108 long videos, evaluated through SightLoop, a shared multiscale harness supporting temporal exploration, local inspection, evidence integration, and revisiting while recording returned observations. On disjoint videos, solver trajectories on verified tasks become step-level supervision for observation decisions. The strongest model reaches only 54.52% on SightLine, yet 79.35% when given the annotated evidence. With under 6K examples, Sight-35B nearly doubles its base model's interactive accuracy (21.29% to 42.26%), outperforms the 397B model of its family, begins using explicit spatial inspection after post-training, which its base model never uses, and gains over 8 points on two external long-video benchmarks.
Comments: 25 pages, including references and supplementary material
Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Cite as: arXiv:2611.22790 [cs.CV]
  (or arXiv:2611.22790v1 [cs.CV] for this version)
  https://doi.org/10.48550/arXiv.2611.22790

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Tianci Xie [view email]
[v1] Tue, 17 Nov 2026 13:52:28 UTC (14,704 KB)

来源:arXiv cs.AI · arxiv.org