Arise

Train. Progress. Arise.

Comment

Inspiration

Arise was inspired by the idea of combining fitness, mindfulness, and gamification into a single, personal experience that runs entirely on your own device. Many fitness apps focus on workouts or just meditation, but few bring everything together in a cohesive, interactive way. The goal was to create something that feels like a personal wellness dashboard with part trainer, part coach, and part game while also exploring the potential of computer vision and local AI tools.

What it does

Arise is a local wellness and fitness website that helps users stay active and mindful through:

Real time exercise tracking using a webcam (push-ups, squats, etc.) Yoga pose guidance with feedback and similarity scoring Guided meditation with customizable breathing patterns Daily quests, XP, streaks, and a local leaderboard for motivation Voice coaching for real-time feedback and encouragement

All of this runs locally on localhost, meaning user data stays on their machine and the experience is fast, private, and self contained.

How we built it

Arise was built using a full-stack approach:

Backend using Flask (Python) handles API endpoints, logic for exercise tracking, yoga feedback, rewards, and voice generation. Computer Vision created with MediaPipe and OpenCV for power pose detection and movement tracking. Frontend made with React and Vite, which provides a fast, responsive UI with multiple sections (exercise, yoga, meditation, quests). Local JSON files store player progress, leaderboard data, and messages. The architecture separates concerns into services (e.g., tracker_service.py, yoga_service.py, rewards_service.py) to keep the system extendible.

Challenges we ran into

Pose detection accuracy for exercises and yoga tracking is highly sensitive to camera angle, lighting, and body positioning, which makes consistency difficult. Real time feedback was difficult. Ensuring smooth, low latency updates between the backend and frontend required careful polling and session handling. Handling browser permissions and ensuring stable webcam access across environments was tricky.

Accomplishments that we're proud of

Successfully integrating AI-powered pose tracking into a full web app experience while creating a multi-feature platform (exercise, yoga, meditation, gamification) rather than a single demo. We built a fully local system with progress and no external dependencies. Implementing a gamified reward system with quests, XP, and streaks. Finally, adding voice coaching for a more immersive and interactive experience

What we learned

How to combine computer vision with web development in a practical application, and the importance of user experience when working with real time systems. Tradeoffs between local architecture vs. cloud-based solutions. How to structure a project into fast services for scalability. Challenges of working with hardware (webcam) and system level tools in web apps. The value of gamification in keeping users engaged

What's next for Arise

Expanding supported exercises and yoga poses. Improving pose detection accuracy with better models or calibration. Adding user profiles and optional cloud sync. Enhancing the leaderboard into a multi-user or online system. Introducing new wellness features like stretching routines or guided programs.

We believe that Arise has strong foundations as a local first wellness platform, and future iterations can evolve it into a more robust, scalable, and widely usable application.

Share this project:

Updates