Education · Case study

AI remote exam proctoring with face and behavior analysis

A remote exam proctoring system for an education provider. It watches webcam, audio and typing behavior during the exam, checks identity with anti-spoofing face detection and keystroke biometrics, flags anomalies in real time, keeps an offline backup of answers and produces an integrity report with heatmaps afterwards.

The challenge

Moving exams online removed the invigilator from the room. The provider needed a way to confirm that the right person was sitting the exam and to spot suspicious behavior, without a human watching every candidate live.

Our approach

We combined webcam, audio and behavior signals in one proctoring pipeline. Face detection with anti-spoofing checks identity, keystroke patterns add a second biometric signal, anomalies are flagged in real time, and each session ends with an integrity report for reviewers.

What we built

  • Webcam and audio monitoring over WebRTC with real-time anomaly flags
  • Facial detection with anti-spoofing to confirm the candidate's identity
  • Keystroke biometrics as a second identity and behavior signal
  • Post-exam integrity reports with heatmaps for reviewers
  • Offline backup so an exam survives a dropped connection

Highlights

  • Multiple signals (face, audio, keystrokes) instead of a single webcam check
  • Reviewers see flagged moments and heatmaps rather than hours of footage

Technology

  • Python
  • OpenCV
  • TensorFlow
  • WebRTC
  • Flask

Online exams need two answers: is this the registered candidate, and did anything unusual happen? We built the proctoring pipeline around both questions. Face detection with anti-spoofing handles identity, keystroke biometrics add a signal that a photo cannot fake, and audio and behavior analysis flag anomalies while the exam is running.

After the exam, reviewers get an integrity report with heatmaps that points them to the moments worth checking. An offline backup keeps the candidate’s work safe if the connection drops mid-exam.

Client details are kept confidential. On a call we can walk you through this project in more depth, under NDA.

Have a project like this in mind?

Book a free 30-minute strategy call. We map where agents save you the most time, and you leave with a plan whether or not we work together.

NDA on request from the first call · You own 100% of the code and IP

More work

Before you go

What would your project cost?

Answer a few questions and get a price range, a timeline and what drives the cost. Two minutes, no email needed.

Estimate my projectOr start with the AI Opportunity Audit, $490