EgoReasoner：通过任务自适应结构化思维学习视角4D推理
EgoReasoner: Learning Egocentric 4D Reasoning via Task-Adaptive Structured Thinking
AI总结：
本文提出EgoReasoner框架，通过任务自适应结构化思维解决视角4D推理任务，提升空间锚定、时间跟踪和持续推理能力，模型在HD-EPIC基准上取得37.5%准确率。
摘要：
视角视频理解因环境的动态4D性质而复杂，相机运动和物体位移需要持续重新评估空间关系。本文针对未充分探索的视角4D推理任务，如固定物计数、视角相对固定物定位、物体运动路线跟踪和静止物定位，这些任务需要不同的认知操作：空间锚定、时间跟踪和持续推理。我们发现这些结构差异使得任务无关方法不足：通用链式思维方法缺乏任务合适的推理原语，统一强化学习在空间任务上主动破坏性能。为此，我们提出EgoReasoner，一个两阶段框架，将推理支架和奖励信号与每个任务的认知结构对齐。第一阶段，任务自适应思维模板指导结构化CoT轨迹的合成，通过监督微调使模型在不同任务类型上适应性推理。第二阶段，任务感知奖励函数验证实体定位、时间对齐和任务自适应逻辑一致性，通过GRPO强化学习选择性加强每个推理路径。我们的3B参数模型仅在16K样本上训练，达到37.5%的平均准确率，在挑战性的HD-EPIC基准上超越Qwen2.5-VL-7B（25.7%）超过10个百分点。
英文摘要：
Egocentric video understanding is inherently complex due to the dynamic 4D nature of the environment, where camera motion and object displacements necessitate a continuous re-evaluation of spatial relations. In this work, we target a suite of under-explored egocentric 4D reasoning tasks, including fixture interaction counting, viewpoint-relative fixture location, object movement itinerary tracking, and stationary object localization, that require fundamentally different cognitive operations: spatial anchoring, temporal tracking, and duration reasoning. We observe that these structural differences make task-agnostic approaches insufficient: generic Chain-of-Thought methods lack task-appropriate reasoning primitives, and uniform reinforcement learning actively destabilizes performance on spatial tasks. To address this, we propose EgoReasoner, a two-stage framework that aligns both the reasoning scaffold and the reward signal to each task's cognitive structure. In the first stage, Task-Adaptive Thinking Templates guide the synthesis of structured CoT traces that teach the model to reason adaptively across task types via supervised fine-tuning. In the second stage, task-aware reward functions verify entity grounding, temporal alignment, and task-adaptive logical consistency, selectively strengthening each reasoning pathway via reinforcement fine-tuning with GRPO. Our 3B-parameter model, trained on only 16K samples, achieves 37.5% average accuracy on the challenging HD-EPIC benchmark, surpassing Qwen2.5-VL-7B (25.7%) by over 10 points.
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。
