slouchi

Inspiration

While brainstorming ideas of problems we wanted to tackle, we tried to focus on something that anyone attending this hackathon could relate to. As we spend hours each day on our laptops, especially during this event, we remind each other to stop slouching, and this sparked an idea. What could we build for people who don't have people around them constantly reminding them of their posture, and how can we make it engaging for the users?

What it does

The main function of slouchi is a posture tracker that uses live pose detection using Google's MoveNet model via TensorFlow.js, which tracks body keypoints at a real-time frame rate. Additionally, a skeleton overlay over the camera feed is colour-coded based on posture quality, and a posture scoring analyzes shoulder tilt, head-forward position, and spine alignment to generate a 0-100 score. The friends function allows users to monitor their friends' postures and bet against them having the worst posture for the month. Finally, there is a page that provides users with simple, daily stretches they can do to improve posture, in a checklist.

How we built it

The app was primarily designed in Figma, using CSS, TSX, and JSON.

Challenges we ran into

As we were all largely unfamiliar with the application, we had to learn how to use it from scratch and ran into various difficulties while navigating the platform.

What we learned

We learned how to create a functional website, implement camera functions, go through the design/brainstorming process efficiently, and effectively divide roles.

What's next for slouchi

In the future, we hope to expand to a 3d view, not only analyzing a front-on view of the user's posture, but also a side view, and implement a metric measuring if the user is too close to their screen. We also hope to further develop functions involving friends