1 paper
Junshan Hu, Chaoxu guo, Liansheng Zhuang +4
Temporal action detection (TAD) aims to locate and recognize the actions in an untrimmed video. Anchor-free methods have made remarkable progress which mainly formulate TAD into tw…