This document discusses facial expression recognition and the challenges that remain. It provides an overview of the current state-of-the-art techniques for facial expression recognition, which still struggle with accuracy when tested on naturalistic data rather than posed images. The document outlines a proposed pipeline for facial expression recognition that combines deep learning techniques for feature fusion and representation learning to help address these challenges and improve recognition accuracy on real-world data. Samples of datasets used for training and evaluating facial expression recognition systems are also presented.