Inspiration
Our inspiration for FitVision came from personal experiences of working out at home during COVID lockdown. We remembered that getting started working out regularly at home was difficult without an educated trainer helping you out on your exercise, form, and counting reps.
What it does
FitVision is a computer vision Ai application that recognizes body movements to distinguish specific muscles being worked. It captures a user's live camera feed while performing a particular exercise. It aids them in completing the necessary reps and sets of the exercise, while providing feedback on the user's form real-time. The backend CV application marks key joints, like the shoulder and elbow, and provides feedback on optimal muscle activation, helping users understand their workout efficacy. It also tracks the user's meals, calories burned, weight goals, and weight gained/lost.
How we built it
We built FitVision using Python with libraries such as Mediapipe for pose detection and OpenCV for image processing. The application leverages Computer Vision and Ai techniques to identify and assist in the user's workouts, real-time through the user’s camera feed. We tied our front-end & back-end with API's we developed and integrated FireStore for User data storage, and Firebase Auth for User Authentication. Additionally, we developed a user-friendly web interface using React, Vite, and Tailwind CSS to ensure seamless user interaction.
Challenges we ran into
One of the main challenges we faced was ensuring our front-end and back-end was integrated correctly through APIs. We encountered difficulties with real-time processing speeds and integrating the computer vision component with the front-end application. Another speed bump we ran into was figuring out how to create the rep counter that keeps track of how many reps of the exercise the user has done in real-time.
Accomplishments that we're proud of
We successfully implemented real-time computer vision model, which worked effectively across various test cases with our front-end application. Additionally, we created a responsive web interface that allows users to interact with the application intuitively.
What we learned
Throughout the development of FitVision, we learned a great deal about computer vision techniques and their practical applications in fitness technology. We gained experience in setting up and configuring various libraries, troubleshooting integration issues, and the importance of user experience design in applications.
What's next for FitVision
Moving forward, we plan to refine the muscle recognition algorithms for greater accuracy and explore additional features, such as incorporating a feedback loop for users to track their progress over time. We also aim to enhance the web application with more interactive features, such as personalized workout plans and video demonstrations. Another goal of ours is developing a version of FitVision on the iOS App Store and Google Play Store.
Log in or sign up for Devpost to join the conversation.