A gradual self distillation network with adaptive channel attention for facial expression recognition

Xin Zhang, Jinlin Zhu, Dongjing Wang, Yueyun Wang, Tingting Liang, Hongbo Wang, Yuyu Yin

Abstract: Highlights•A lightweight Gradual Self Distillation Network for accurate and efficient FER.•A gradual SD strategy transferring knowledge from deep to shallow layers gradually.•An adaptive channel attention to enhance the ability of capturing important features.