BriefGPT.xyz
Dec, 2021
时间等变对比视频表示学习
Time-Equivariant Contrastive Video Representation Learning
HTML
PDF
Simon Jenni, Hailin Jin
TL;DR
介绍了一种新的自监督对比学习方法,旨在从未标记的视频中学习表示。该方法利用新的约束条件,以建立对时间变换等价的表示,并更好地捕捉视频动态。实验表明,时间等变表示在UCF101,HMDB51和Diving48的视频检索和动作识别基准测试中实现了最先进的结果。
Abstract
We introduce a novel
self-supervised contrastive learning
method to learn representations from unlabelled videos. Existing approaches ignore the specifics of input distortions, e.g., by learning invariance to
temporal t
→