Post by Aryan Vyahalkar

In another life, I’m pretty sure I’d be a ninja instead of a programmer. But that's in another life. Here, I have a webcam, Python, and just enough boredom and curiosity to perform the next best thing. I turned my laptop camera into a computer vision clone of Fruit Ninja by Halfbrick. Under the hood, it's way more math than anything else. I'm using MediaPipe's HandLandMarker, which gives me 21 3-Dimensional landmarks per hand in real time (basically a live wireframe of my fingers). Those landmarks come in as normalized coordinates, so the first step is turning abstract numbers into something the screen understands. A few multiplies later, they’re in pixel space and ready for collisions. From there, OpenCV takes over: drawing the hands, tracking motion frame by frame, pawning fruits with simple physics (upward launch + gravity), and checking whether my fingertip trail gets close enough to a fruit’s center to count as a slice. It’s part geometry, part computer vision, and part me waving my hands around hoping the model keeps up. Not a swordfighter. Not even a game dev (might need to rotate after this). Just a bored, teenage programmer building out cool stuff. Check it out: GitHub repo 👉 https://lnkd.in/gpWPGAex Project case study 👉 https://lnkd.in/d_5hwk6Z (where I talk about the math) #ComputerVision #Python #OpenCV #MediaPipe #PyAutoGUI #GestureControl #FruitNinja #Innovation #LearningByBuilding #GitHub

Video Content