TLG: 通过源标注重建和类别目标推理实现视频问答的时间逻辑基础
TLG: Temporal-Logic Grounding for Video Question Answering via Source-Annotation Reconstruction and Category-Targeted Reasoning
AI总结：
提出TLG三阶段系统，通过重建动作时间线、解析问题为时间逻辑程序并确定性执行，结合强视觉语言模型和前沿推理模型，将视频问答准确率从46.9%提升至71.37%。
摘要：
TimeLogic挑战评估对视频的形式时间逻辑推理——包括16个算子（之前、之后、直到、自从、总是、共现、排序等），采用布尔和四选一形式。端到端视频语言模型在此任务上接近随机水平，因为它们将视频视为帧的集合，无法定位动作发生的时间。我们提出TLG（时间逻辑基础），一个三阶段系统：（i）从生成基准测试的公共源数据集标注中重建每个视频的动作时间线，将每个问题解析为时间逻辑程序，并确定性执行；（ii）在没有标注的情况下回退到强大的开放视觉语言模型；（iii）仅将视觉语言模型经验上最弱的问题类别路由到前沿推理模型。TLG将测试准确率从46.9%的视觉语言模型基线提升到71.37%，绝对增益+24.5，达到排行榜前三名3分以内。我们报告了广泛的消融实验，包括三种基于模型的时间线重建变体，它们都低于整体视觉语言模型，将时间基础隔离为不可约的瓶颈，并表明真正的标注——而非更大的模型——驱动准确率。
英文摘要：
The TimeLogic Challenge evaluates formal temporal-logic reasoning over video - 16 operators (before, after, until, since, always, co-occur, ordering, ...) in boolean and 4-way multiple-choice form. End-to-end video-language models (VLMs) hover near chance on this task because they treat video as a bag of frames and cannot localize when actions occur. We present TLG (Temporal-Logic Grounding), a three-tier system that (i) reconstructs each video's action timeline from the public source-dataset annotations the benchmark was generated from, parses every question into a temporal-logic program, and executes it deterministically; (ii) falls back to a strong open VLM where no annotation exists; and (iii) routes only the question categories where the VLM is empirically weakest to a frontier reasoning model. TLG raises test accuracy from a 46.9% VLM baseline to 71.37%, a +24.5 absolute gain, reaching within 3 points of the leaderboard top. We report extensive ablations, including three model-based timeline-reconstruction variants that all underperform a holistic VLM, isolating temporal grounding as the irreducible bottleneck and showing that real annotations - not larger models - drive accuracy.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
