Haar Cascade algorithm detects face in the image or live feed, crops it and then sends it to our CNN model which detects emotion. Model is trained on FER-2013 dataset on kaggle. Dataset consists of 7 classes, the training set consists of 28,709 examples and the test set consists of 3,589 examples. The detected emotion is used as key to search for a gif using giphy api, the api then sends a gif to be displayed.