All work — Computer vision — 2026
SAT0RU
Hand-tracked cursed-technique visualiser: MediaPipe landmarks drive a Three.js particle engine, with an ML classifier picking the technique in real time.
The problem
Gesture demos usually hard-code one pose and break the moment lighting changes.
How I approached it
A trained classifier reads 21-point hand landmarks per frame and votes across a rolling window; a deterministic geometric fallback takes over whenever confidence drops, so the visual never freezes.
The outcome
Techniques fire reliably at webcam frame rate on a laptop GPU, with graceful degradation instead of a dead screen.
Camera access: this one asks for your webcam and runs entirely in your browser — no video is uploaded or stored anywhere.
Next project