ViewBridge：用于极端视角变化下的活动视角不变性的课程知识蒸馏
ViewBridge: Curriculum Knowledge Distillation for Activity View-Invariance Under Extreme Viewpoint Changes
AI总结：
本文提出ViewBridge框架，通过课程学习和知识蒸馏方法，在严重视角遮挡下学习丰富的视频表示，提升了极端视角变化下的活动识别性能。
摘要：
传统视角不变学习方法依赖受控的多视角训练数据，但难以处理野外视频中的极端视角差异和低视觉内容。我们引入ViewBridge框架，学习在严重视角遮挡下丰富的视频表示。我们提出一个保留动作中心语义的知识蒸馏目标，以及一种新的课程学习过程，逐步增加挑战性视角，从而实现对极端视角差异的平滑适应。为排序训练视频片段，我们定义了基于几何的指标，反映其可能的遮挡水平。在训练时利用多视角数据，而在推理时输入是未经校准的单视角视频。在两个任务——时间键步定位和细粒度键步识别上，我们在三个数据集（Ego-Exo4D、LEMMA、EPFL-Smart-Kitchen-30）上优于现有最佳方法。项目页面：https://vision.cs.utexas.edu/projects/learning_view_distill/
英文摘要：
Traditional methods for view-invariant learning rely on controlled multi-view training data with minimal scene clutter. However, they struggle with in-the-wild videos that exhibit extreme viewpoint differences and share little visual content. We introduce ViewBridge, a framework for learning rich video representations in the presence of severe view-occlusions. We introduce a knowledge distillation objective that preserves action-centric semantics, together with a novel curriculum learning procedure that pairs incrementally more challenging views over time, thereby allowing smooth adaptation to extreme viewpoint differences. To sort training video segments for the proposed curriculum, we define a geometry-based metric that reflects their likely occlusion level. While training leverages multi-view data, at inference time, the input is an uncalibrated, single-viewpoint video. Evaluating our approach on two tasks -- temporal keystep grounding and fine-grained keystep recognition -- we outperform SOTA approaches across three datasets (Ego-Exo4D, LEMMA, EPFL-Smart-Kitchen-30). Project page: https://vision.cs.utexas.edu/projects/learning_view_distill/ .
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。
