Driver Awareness

Problem

Show what the detection and facial-analysis APIs could do in one of the hardest environments for them: inside a moving vehicle, where lighting swings from direct sun to tunnel darkness in seconds and the subject is partly occluded by their own posture.

Approach

Combined gaze estimation with head pose, composed from the same APIs the product line already exposed, and tuned against real in-cabin footage rather than clean benchmark data.

Result

A working demonstration of the API stack applied to driver attention. It was built to show capability rather than to ship — the demo never became a product.