Facial Emotion Recognition Using Deep Learning

Abstract

Facial emotion recognition (FER), because of its significant academic and business potential, is an important subject in the fields of computer vision and artificial intelligence. The purpose of this project is to develop an emotion detection pipeline using video frames. In particular, we detect and analyse the faces of the video through deep neural networks for the recognition of emotions. We use a CNN and an RNN based on documents submitted in the Wild Challenge for emotional recognition. An input video is divided into small segments. We will detect, crop and align faces for each segment. This gives an image sequence. A CNN will extract relevant features in the sequence for each image. These features will be sequentially feed to an RNN that encodes emotional movement and facial expressions. The entire process is carried out as a Python.