TAR:面向视频时序定位的时序锚点约束推理框架
TAR: Temporal Anchor-Constrained Reasoning for Video Temporal Grounding
AI 导读
论文提出 TAR(Temporal Anchor-Constrained Reasoning),针对视频时序定位(VTG)中纯结果监督强化学习导致的推理幻觉问题,引入时序锚点(T-anchor)作为可审计的检查点机制,强制模型在 CoT 中持续依据视觉证据校准时序预测。该方法仅用标准 7B 模型即可自动收集高质量 CoT 数据,无需依赖超大规模模型,实验显示取得 SOTA 性能。
正文 · 原文
Abstract:Video Temporal Grounding (VTG) aims to localize specific video segments corresponding to natural language queries. While recent Large Vision-Language Models (LVLMs) employ Reinforcement Learning to generate Chains-of-Thought (CoT), they typically rely solely on outcome-based supervision. Consequently, this often leads to hallucinations, where the reasoning process becomes disconnected from the visual content and the final prediction. Existing attempts to mitigate this by relying on external supervision from larger models or separate reward models are computationally expensive and prone to rigid patterns. To address these challenges, we propose TAR (Temporal Anchor-Constrained Reasoning), a framework that introduces the temporal anchor (T-anchor) as a transparent and auditable checkpoint mechanism. T-anchor enforces progressive refinement within the CoT, compelling the model to continuously ground its intermediate thoughts in visual evidence and iteratively calibrate temporal predictions, thereby significantly enhancing the faithfulness and autonomy of the reasoning process and final accuracy. Furthermore, we introduce a bootstrapping paradigm that automatically harvests high-quality CoT data using only a standard 7B model, eliminating the dependency on ultra-large models. Extensive experiments demonstrate that TAR achieves state-of-the-art performance and generates faithful, autonomous, and progressively refined reasoning traces.
| Comments: | Accepted by ECCV2026 |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2508.07683 [cs.CV] |
| (or arXiv:2508.07683v3 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2508.07683 arXiv-issued DOI via DataCite |
Submission history
From: Chaohong Guo [view email]
[v1]
Mon, 11 Aug 2025 06:59:32 UTC (2,172 KB)
[v2]
Mon, 29 Jun 2026 02:15:03 UTC (1,192 KB)
[v3]
Tue, 17 Nov 2026 07:55:57 UTC (1,196 KB)
来源:arXiv cs.AI · arxiv.org