College is hard. It's hard to keep track of all your activities, classes, and other commitments while maintaining your mental health. We've been through this ourselves and seen many of our friends suffer from being too stressed and sad. We want to fix that with Moody!
What it does
Moody is a mood tracker powered by AI. It's easy to use and fast -- so it won't interfere with your day to day life. Logging an emotion is as simple as taking a selfie! Keep logging your feelings, and you'll be able to see your mood history and statistics.
How we built it
Moody's core functionality -- the quick logging -- relies on a custom ML model, which was built with TensorFlow and Keras. We use a Deep Convolutional Neural Network with a custom architecture to predict the user's emotion from a picture of their face. This model was trained on Google Cloud Platform. The AI backend then tells the web backend what emotion the person in question is displaying, which then communicates with the front end. The web backend is written in python with Flask, and the front end is written in Bootstrap and Jinja2.
Challenges we ran into
It was extremely challenging to train such a large neural network in such a limited amount of time. We had to learn how to use TPUs and GPUs to process very quickly.
Accomplishments that we're proud of
We're proud of our model and the potential of our app.
What we learned
We learned how to use ML models with web frameworks.
What's next for Moody
Moody can be used to give students tips on how to be more happy and less stressed. It can also supply data to university administration so that officials can change policy to make their students happier.