Facial Expression Recognition Using Deep Learning and Graph Convolution Networks

Facial expression recognition is an important for the cognitive psychology research and artificial intelligence. This study focuses on recognizing facial expressions in difficult situations like different angles of face, lighting, and partial face visibility. To analyze facial expression the Graph Convolution Network (GCN) method. This method first identifies the important areas from the video frames using a bottom-up attention module. These areas become nodes in a graph. This graph captures the connection between the nodes and whether the image's parts are related. Action units are highlighted using an attention map to find important details related to expression. The action map combines different parts of the face.