All projects

StayAwake

Real-time computer vision that detects lost focus from eye closure, yawning, and head orientation.

StayAwake watches for the physical signals that precede lost attention — sustained eye closure, yawning, and head orientation drifting away from the screen — and flags them in real time from a standard webcam feed.

The part that made it actually usable was per-user calibration. Baseline eye aperture, blink rate, and natural head position vary enough between people that a single fixed threshold produces constant false alarms for some users and silence for others. Calibrating against a short per-user baseline first, then measuring deviation from that, turned a noisy demo into something that could run in the background without being ignored.

Live detection firing as eyes close and a yawn is caught.