Joint image-instance spatial-temporal attention for few-shot action recognition

Open Webpage

Zefeng Qian, Chongyang Zhang, Yifei Huang, Gang Wang, Jiangyong Ying

Published: 01 Jan 2025, Last Modified: 18 Apr 2025Comput. Vis. Image Underst. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•A novel FSAR method that explicitly utilize both instance-level and image-level cues.•Instance-enhanced prototype construction and global–local matching strategy.•Experiments across five widely used benchmarks validate the our effectiveness.