Face Recognition: How It Works

Face Recognition confirms that the person tracking time is the employee you hired. It is optional and consent-first. It runs only in the current version of the tracking app: employees still on the legacy app or on the Chrome extension must install the current version from the Downloads section of their dashboard first, and the projects page tells you when that is the case.

The flow has four parts:

• Consent. The employee grants consent for your company — per employer, and withdrawable at any time. Nothing runs until they agree.
• Photo. The employee submits a face photo, which you approve or reject from the Face Recognition Setup window in PROJECTS.
• Verification. Once enabled, SCREENish checks the camera during tracked sessions against the approved photo. Each check also runs liveness detection: several frames are tested for a live person in front of the webcam rather than a photo or a video held up to it, and a check counts as a match only when the face matches and the frames are live.
• Match rate. You set the minimum acceptance rate; the per-hour match result appears in WORK LOGS.

A low match rate is a prompt to talk, not proof of anything — cameras and lighting fail far more often than people do.

A sustained run of checks in which a face is present but keeps failing the liveness test shows in WORK LOGS as a “Spoof suspected” notice. What it means and what to do about it is covered in Work Logs: Camera Notices During Face Verification; what liveness detection catches and what it does not is set out on the liveness detection page.