Application utilizing a Convolutional Neural Network built with tensorflow and trained on a data set of 48X48 pixel images of peoples faces. The people in the data set are displaying an emotion that is one of 4 classes. Surprise, Happy, Sad, Neutral. After training the model it was converted into tensorflow.js format for deployment in a web application. The emotion detection portion of this appliation was part of the capstone project for MIT's Applied Data Science Program
The web application activates and uses either the forward or rear camera on mobile or primary web camera and additionally attached camera on desktop. Activate the emotion detection feature, keep your face 2 feet from the camera for best accuracy, and the predicted emotion will be displayed on the screen.