Facial expressions display emotions and play a vital role in interpersonal communication. In this paper, We propose a multi-model method to track the real-time emotion of human subjects. We explore numerous training strategies and hyperparameters on our custom ensemble model, which achieves a 74.48% accuracy by integrating different techniques. In addition, we combine our ensemble model and a shrunk version of the YOLOv1 network to create a web application using python flask that achieves facial expression recognition in real-time.
