arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.16155cs.CV
收藏

通过学习动作和上下文的显式子空间的弱监督时间动作定位

Weakly Supervised Temporal Action Localization Through Learning Explicit Subspaces for Action and Context

University of Illinois at Chicago(伊利诺伊大学芝加哥分校);The State University of New York at Buffalo(纽约州立大学布法罗分校);Wormpex AI Research(Wormpex AI研究院)

Ziyi Liu, Le Wang, Wei Tang, Junsong Yuan, Nanning Zheng, Gang Hua

摘要:

弱监督时间动作定位(WS-TAL)方法学习仅在视频级监督下定位视频中动作实例的时间开始和结束。现有的WS-TAL方法依赖于为动作识别而学习的深度特征。然而,由于分类和定位之间的不匹配,这些特征不能区分频繁共现的上下文背景,即,上下文和实际行动实例。我们将这种挑战称为动作上下文混淆,它将对动作定位的准确性产生不利影响。为了解决这一挑战,我们引入了一个框架,分别为动作和它们的上下文学习两个特征子空间。通过明确地考虑动作视觉元素,动作实例可以更精确地定位,而不会从上下文中分心。为了便于学习这两个特征子空间,只有视频级别的分类标签,我们利用预测从空间和时间流的片段分组。此外,还引入了一个无监督学习任务,使所提出的模块专注于挖掘时态信息。所提出的方法在三个基准上优于最先进的WS-TAL方法,即,THUMOS 14,ActivityNet v1.2和v1.3数据集。

英文摘要:

Weakly-supervised Temporal Action Localization (WS-TAL) methods learn to localize temporal starts and ends of action instances in a video under only video-level supervision. Existing WS-TAL methods rely on deep features learned for action recognition. However, due to the mismatch between classification and localization, these features cannot distinguish the frequently co-occurring contextual background, i.e., the context, and the actual action instances. We term this challenge action-context confusion, and it will adversely affect the action localization accuracy. To address this challenge, we introduce a framework that learns two feature subspaces respectively for actions and their context. By explicitly accounting for action visual elements, the action instances can be localized more precisely without the distraction from the context. To facilitate the learning of these two feature subspaces with only video-level categorical labels, we leverage the predictions from both spatial and temporal streams for snippets grouping. In addition, an unsupervised learning task is introduced to make the proposed module focus on mining temporal information. The proposed approach outperforms state-of-the-art WS-TAL methods on three benchmarks, i.e., THUMOS14, ActivityNet v1.2 and v1.3 datasets.

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。

↑