Gaze Tracking

Gaze tracking focuses on a subtle but telling signal: the direction of a candidate's attention. During a legitimate exam, a test-taker's eyes stay largely on the screen where the questions appear. Frequent or sustained glances elsewhere, especially toward a fixed point off to the side, can suggest that the candidate is reading from notes, a phone, or a second monitor.

The technique uses the webcam to estimate eye position and head orientation, building a picture of where the candidate is looking over the course of the session. Rather than flagging every glance, well-designed systems look for patterns, such as repeated or prolonged off-screen attention, that stand out from the natural eye movement of someone simply thinking or reading the question.

Its purpose is to catch a form of cheating that other signals might miss. A candidate could remain silent and appear alone on camera while still consulting hidden material, and the direction of their gaze may be the clearest indication that something is amiss. In this way gaze tracking complements face and audio monitoring.

Care is needed to use it fairly. People naturally look away to think, and disability, screen setup, or testing conditions can all produce innocent off-screen glances. Treating gaze signals as prompts for human review, rather than proof of misconduct, and being transparent with candidates helps avoid penalizing harmless behavior.

Within automated and AI-based proctoring, gaze tracking is one of several behavioral signals that together build a fuller account of a session. Combined with face detection, object detection, and audio analysis, it adds a useful dimension to the layered monitoring that protects online exam integrity. On its own a single off-screen glance means little, but considered alongside other evidence it can help a reviewer understand what was happening during a flagged moment, which is how the signal is best used.