IRL Fruit Ninja

Main Menu

Inspiration

We wanted to bring the nostalgic mobile gaming experience into the physical world by using computer vision to turn real-life hand gestures into a high-speed arcade game.

What it does

IRL Fruit Ninja is a web-based game that uses your webcam to track hand movements, allowing players to slice virtual fruit and dodge bombs in both single-player and two-player split-screen modes.

How we built it

The game was developed using MediaPipe for real-time hand tracking, HTML Canvas for the frontend rendering, and a FastAPI backend to manage the global leaderboard.

Challenges we ran into

The biggest hurdles were optimizing the computer vision model to maintain a for smooth slicing and managing the complex state logic required for a synchronized two-player split-screen experience. In addition to deployment with many dependencies and a complex file management system.

Accomplishments that we're proud of

We are particularly proud of successfully implementing a low-latency blade-swipe effect that feels responsive and the wood-themed leaderboard UI that integrates seamlessly with our backend.

What we learned

During this project, we deepened our understanding of browser-based machine learning and learned how to efficiently handle real-time coordinate mapping between webcam feeds and canvas elements.

What's next for IRL Fruit Ninja

We plan to refine our Vercel deployment, introduce more diverse power-ups beyond the star multiplier, and integrate gesture-based menu navigation for a completely touchless experience.