An emotion recognition "try not to laugh" style online game. Play by yourself or challenge your friends to see who can resist the urge the longest. Players select a video and press play, and for as long as they do not express the emotion the video is intended to illicit, they gain points. As soon as a player smiles or laughs to a funny video, for example, the game is ended and their point total is displayed. Players can play with funny, sad, and scary videos.
AI Model Features
- Machine learning model trained using the FER 2013 dataset. We used Fear, Happy, Sad, and Neutral. The model analyzes facial expression and makes a prediction.
- CNN model created with Keras, 6 2D convolutional layers, 3 fully connected layers, softmax output.
- Hosted by a flask application with facial recognition, facial landmarking, and image transformation.
Other details
- Fully functioning backend, complete with mySQL database, API to interact with said database, complete with necessary docker containers
- Fully functioning React.js frontend, with profiles, leaderboards, gameplay pages, and more
This project was created by Nic Ballesteros, Dom Miller, Sri Potturu, and Ashton Statz for CS 30700 at Purdue University, Fall Semester 2021