Abstract: Highlights•We propose the GACN model to recognize pose-invariant confusable facial expressions.•The proposed GACN includes two modules named PIFEE and GAFEC.•Our method improves FER performance via the integration of PIFEE and GAFEC.
Loading