Computer vision / 2026 / Live in the browser
Finger Monkey Lab
A gesture control layer for hand tracking, built as an Electron app over Python ML backends and now running its tracking front end in a browser too.
01
The problem
Browser hand tracking hits a ceiling on model choice and system access.
02
How I approached it
Electron handles the interface while Python owns the models, connected over a local bridge so heavier detectors can run natively.
03
The outcome
Browser-grade UI with native-grade models behind it on the desktop. The web build runs the same interface and the same MediaPipe tracking with no sidecar, so the gesture layer works without installing anything and the heavier Python detectors are what the desktop adds.
Related work
ASL fingerspelling to text from a webcam. It reads hand geometry instead of pixels, so all 26 letters work with no training, and two seconds of calibration per letter fits it to your own hand.
A webcam focus monitor that watches your head and posture while you study, and when you keep drifting off it opens the McDonald's careers page for you.
A desktop app that watches the webcam and, once you stop facing the screen, pulls Chrome to the front, launching it on Gmail if it is not open.