Inspiration
Remember how you wanted to start working out during quarantine, but you have no motivation to do it? Welcome to FitSense, where we strive to provide an interactive and entertaining workout experience to users. People often shy away from the gym, because of the fear of being judged by others (especially beginners). With FitSense, you can get the full gym experience from the comfort of your home.
What it does
FitSense provides a workout routine and uses computer vision to track progress. The computer vision model was trained to classify a certain range of motion as a specific exercise. FitSense will keep a record of repetition in a given exercise, and slowly progress the user from one exercise to another.
How we built it
Our project began with training a computer vision model to recognize different workouts and store them as numeral data. We can then feed the data through a program that will return the output displayed on the website.
Challenges we ran into
Initially, the model had trouble differentiating exercises that assumes similar posture. It would often treat two different movements as the same exercise due to the lack of simple size.
Accomplishments that we're proud of
Interactive design and high accuracy pose recognition, and great synergy as a team.
What we learned
In order to improve FitSense to be more accurate and compatible with different camera angles, we would need a lot more data and time to train the computer vision model. Another thing we learn is that training a computer vision model is much more effective on a plain background.
Built With
- computer-vision
- css
- html
- javascript
- machine-learning
Log in or sign up for Devpost to join the conversation.