FormTrackr

Inspiration

Many inexperienced gym-goers perform their exercises with poor form. Not only does this lead to sub-optimal muscle gain, but also can cause pain in the back and joints. We want to create an app that helps with this, which allows users to not have to hire a personal trainer.

What it does

The web-app mainly acts as a gym progress tracker, where users can create progress entries that show how much weight and how many reps and sets they did for a particular exercise at a particular date and time. Users can also choose to record a video of themselves doing the workout. The plan was to send this video over to an AI backend to analyze the form and return a "Form Match Percentage", but we ran out of time before the AI could be developed. So current features only include:

• User sign in/sign up
• Progress creation/deletion
• Exercise recording

How we built it

The web app was built using Next.js and we used a Postgres database to store all our data - we used Prisma as our ORM. The AI (or what we have of it so far) was built using python and yolov8.

Challenges we ran into

Developing the AI was harder than expected. Although we did manage to get proper tracking of joints (as in body joints), we couldn't find a way to properly analyze each joint and decide if it was too low/too high etc.

Accomplishments that we're proud of

The web app works perfectly fine, although it is a bit slow.

What we learned

AI is pretty hard to develop in 2 days!!

What's next for FormTrackr

Hopefully get a working AI model

Built With

• nextjs
• postgresql
• python
• tailwindcss
• typescript
• yolov8