Skip to main content

Get Video Face Analysis

GET 
/v2/videos/:video_id/face_analysis

Per-frame gaze, eye, blink, liveness, and lip signals — no frame images.

Same authorization as other video GETs (owner, admin, shared, or sample).

Each frames[] row is one face on one frame. Fields the pipeline did not compute are omitted.

Blink rate is not returned. Count False→True transitions on blink and divide by the face's duration in minutes. The eye-tracking heatmap is not a stored dataset — it is a density view of gaze_yaw × gaze_pitch.

Page with start_frame + length on long videos. length=0 (default) pulls every frame from frames-service, same as GET /v1/videos/{id}/frames.

Request​

Responses​

Successful Response