How to Ensure Secure Online Tests with Facial Recognition

Article verified for Release 15.2 on April 15, 2026.

This article provides an overview of how facial recognition is used to maintain the integrity and security of online tests, including how the process currently operates.

Equipment Check

At this stage, a photo of the student is captured. For first-time sessions, this photo is stored in the student’s profile for future comparisons with subsequent session photos.

Test Phase

During the examination itself, facial recognition and comparison are facilitated by specific metrics and the “Verify” parameter:

Face Visibility – Face invisible or not looking into the camera (c2)

This operates in real-time on the front-end, checking every 10 seconds. It may trigger a false positive if the student wears glasses or there is glare.

Multiple Faces – Several faces in front of the camera (c3)

Also real-time on the front-end, checking every 10 seconds. This metric triggers when it detects more than one face in the frame.

Profile Mismatch – Face does not match the profile (c4)

Indicates when the detected face does not match the stored profile in case of the first session. Otherwise, it compares to the first photo of the session. Checking every 60 seconds, operates on the back-end.

Similar Profile – Found a similar profile (c5)

Detects if a closely matching profile exists. This metric compares the profile photo of the current participant with other profiles. Checking every 60 seconds, operates on the back-end.

The “Verify” Parameter

The “Verify” parameter reflects its outcome in the session list. It compares live session photos with the student’s profile image. Unlike c4, which strictly compares live photos with the initial session image, “Verify” assesses live photos against the profile image throughout the session.