Dual-STI: Dual-path spatial-temporal interaction learning for dynamic facial expression recognition

Min Li, Xiaoqin Zhang, Chenxiang Fan, Tangfei Liao, Guobao Xiao

Abstract: Highlights•We propose a distinct framework to concurrently extract spatial-temporal features for dynamic facial expression recognition.•We introduce a module for spatial-temporal interaction learning and comprehensive facial expression feature extraction.•Our method achieves state-of-the-art results on three benchmarks: DFEW, AFEW, and FERV39k.