This document summarizes research on developing an AI to detect facial emotions in images. The researchers tested several models including KNN, logistic regression, decision trees, neural networks, CNNs, and transfer learning. Their most accurate models were a pre-trained VGG model using transfer learning at 68.2% accuracy and their own trained CNN model at 68.0% accuracy. Potential applications of the emotion detection AI include helping children with autism and improving online learning environments. If given more time, the researchers would develop a live camera feature for the AI.