Guide
Face Detection vs Face Recognition in Online Proctoring
6 minute read - Updated: November 2026
Two phrases that sound alike describe very different things, and the difference decides what an institution owes its students under biometric privacy law. Face detection asks whether a face is present and whether it is the same face throughout a session. Face recognition asks who that face is by matching it against a stored identity. Online proctoring services use one, the other, or both, and their marketing rarely makes the distinction clear. This guide explains it.
Face detection
Face detection finds a face in a video frame and tracks it. In proctoring it answers practical questions: is there a face in the frame, is there a second face, did the face leave for a long period, is the face looking away from the screen for an extended time, and does the face at minute forty appear to be the same face as at minute one. It does not need to know who the person is, and it does not need a stored reference image of the person. The data it produces is a set of events and, where recording is on, the video itself.
Face recognition
Face recognition identifies a person by converting facial features into a template and comparing it to a stored template or reference image: a photo ID, a prior session, or an enrollment picture. In proctoring it answers one question, is this the enrolled student, at the start of a session or continuously. It requires collecting and storing a biometric identifier, which is the thing biometric privacy laws regulate.
Why the difference matters legally
Illinois, Texas and Washington have biometric privacy statutes, several other states regulate biometric identifiers through consumer privacy laws, and the GDPR treats biometric data used to identify a person as a special category. The trigger in most of these laws is the creation or use of a biometric identifier for identification. Face detection that tracks presence and consistency without identifying the person generally sits outside that definition; face recognition sits inside it and brings notice, consent, retention and destruction obligations with it. Institutions should read their vendor's data processing agreement to find out which they are using, and should not rely on this guide as legal advice.
What to ask a proctoring vendor
Do you identify the test-taker against a reference image, or detect and track a face without identifying it. If you identify, what biometric data do you store, where, and for how long, and can the feature be turned off. If you detect only, do you store any template beyond the session. Do you send video or images to any third-party AI provider. Does any of this data train your models.
What ScreenComply does by default
ScreenComply uses face detection, not face recognition, by default. When webcam recording is enabled by the institution, the session is checked for a second face, extended absence and face consistency, and the events are recorded in the session report for a reviewer. No identity match is made and no biometric template is kept beyond the session. An institution can enable an optional photo ID match at the start of a session with student notice and consent; when it does, the comparison data is deleted with the session, and the institution's retention setting governs the recording. No video, audio or images are sent to third-party model providers, and no student data trains models without written authorization.
Frequently asked questions
Is face detection a biometric identifier?
Generally not when it does not identify the person, but definitions vary by statute. Ask your counsel with the vendor's data processing agreement in hand.
Does online proctoring have to use face recognition?
No. Many integrity outcomes depend on presence, consistency and events, not on identification. Identity verification can be done with a document check or kept off entirely.
Can a student refuse a face scan?
The institution sets the alternative. ScreenComply supports a configured fallback for a student who declines consent.
Does face detection work for students of all skin tones and in poor light?
Detection performance varies with lighting and camera quality for everyone, which is one reason flags are reviewed by a person rather than acted on automatically.
What does ScreenComply store from the webcam?
The recording, if the institution enables it, for the institution's retention period, plus the events detected. No biometric template beyond the session.
See a session, a detection, and a report
A short walkthrough of a live exam, a flagged moment, and the finding your integrity office would receive.