Conventional Feature Engineering and Deep Learning Approaches to Facial Expression Recognition: A Brief Overview

Facial expression recognition (FER) is vital in pattern recognition, artificial intelligence, and computer vision. It has diverse applications, including operator fatigue detection, automated tutoring systems, music for mood, mental state identification, and security. Image data collection, feature engineering, and classification are vital stages of FER. A comprehensive critical review of benchmarking datasets and feature engineering techniques used for FER is presented in this paper. Further, this paper critically analyzes the various conventional learning and deep learning methods for FER. It provides a baseline to other researchers about future aspects with the pros and cons of techniques developed so far.