Hierarchical Space-Time Attention for Micro-Expression Recognition

arXiv:2405.03202v1 Announce Type: new
Abstract: Micro-expression recognition (MER) aims to recognize the short and subtle facial movements from the Micro-expression (ME) video clips, which reveal real emotions. Recent MER methods mostly only utilize special frames from ME video clips or extract optical flow from these special frames. However, they neglect the relationship between movements and space-time, while facial cues are hidden within these relationships. To solve this issue, we propose the Hierarchical Space-Time Attention (HSTA). Specifically, we first process ME …

abstract arxiv attention cs.cv emotions extract flow hierarchical however micro movements optical optical flow recognition relationship space type video