Can this AI see
if you laugh?

The classic Try Not To Laugh Challenge with an AI to catch your laugh!

For best results, ensure good lighting and center only yourself in frame.

to start the AI!

0s

How To Play

Enable Camera 📸

Give webcam access to the AI to track your facial expressions.

Choose a Video 🎥

Search for a funny video or let us pick one at random for you from YouTube.

Let's Play! ▶️

Activate the AI and begin the timer with the 'Start' button and try not to laugh!

About

Built with modern languages to provide a modern experience

Employing the FER-2013 dataset from Kaggle, Python and TensorFlow are used to configure a convolutional neural network (CNN) model for facial emotion recognition. Using data augmentation techniques during preprocessing, the CNN model achieves an 84% testing accuracy.

Deployed on AWS EC2, the backend relies on a Flask and Nginx reverse proxy server hosting the CNN model, and for cross-communication between the front-end deployed on Vercel, respectively.

Built on Next.js, the front-end utilizes modern languages & libraries such as React.js, TypeScript, Chakra UI, Tailwind CSS, and Framer Motion to leverage static rendering.

Inspired by the recent surge of AI, this project was created to capitalize on these emerging technologies.