GatThree

Inspiration

Life as a student is busy. People often stare at computer screens all day both for work and relaxation. However, we must prioritize our wellbeing. GatThree is a web app designed to run in the background to monitor the emotions of users from their facial expressions.

What it does

GatThree runs in the background of your browser, occasionally capturing an image from the computer's camera and classifying the user's face as an emotion. After a period of time, the app reacts based on the most frequent emotion classified. If the user's expression appears to be unhappy, the app will remind the user of the bright side of life. We hope this will remind users to take frequent breaks as well as use their computers in a healthy way.

How we built it

The frontend was built in JavaScript and HTML/CSS, the backend was built in Python and Flask, the ML model was a cv2 cascade classifier, the encouragement text generator used API to Google Gemini chatbot

Challenges we ran into

The model would often predict the facial expression to be happy (maybe because most of the data was happy faces so predicting happiness most of the time would lead to highest accuracy?), connecting frontend to backend model by making sure image gets sent correctly from frontend, Terraform not deploying app properly

Accomplishments that we're proud of

We were able to split into two teams that worked independently on frontend and backend, and we were able to successfully merge the two groups of code with little conflict.