Synergic learning for noise-insensitive webly-supervised temporal action localization

作者：

Highlights：

• We propose a framework for Webly-supervised Temporal Action Localization (WebTAL).

• We introduce a synergic task called STOP for spatio-temporal feature learning.

• WST is proposed to iteratively generate better features and improve TAL results.

• Experiments on benchmarks show that our method achieves state-of-the-art results.

摘要

•We propose a framework for Webly-supervised Temporal Action Localization (WebTAL).•We introduce a synergic task called STOP for spatio-temporal feature learning.•WST is proposed to iteratively generate better features and improve TAL results.•Experiments on benchmarks show that our method achieves state-of-the-art results.

论文关键词：Temporal action localization,Web supervision,Spatio-temporal representation

论文评审过程：Received 31 May 2021, Revised 24 June 2021, Accepted 29 June 2021, Available online 2 July 2021, Version of Record 13 July 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104247