Computer Visionaries: Roger De Mello Koch, Seik Oh, Ziyan Liu, Alex Ker
Emotion recognition is important in various fields, such as psychology and medicine, helping to diagnose, predict and analyze behavior. In this project, we implement deep learning models that utilizes Convolutional Neural Networks (CNNs) and Transformer architectures. We compare these models to fine-tuned pre-trained models to evaluate their efficacy in emotion detection. Our best performing model, fine-tuned ViT, achieved ~70% validation accuracy and we constructed a real-time emotion classification pipeline.