I’m working on a real-time Computer Vision project for an online assessment system.
The requirement is to detect when a candidate’s face becomes completely absent or partially outside the camera frame during a test. If it continues for a certain duration, the system should show a warning and save an evidence frame with the timestamp.
I’m currently considering Python + OpenCV, YOLO for face detection, MediaPipe for facial landmarks, and FastAPI/WebSockets for real-time communication.
I’d love to hear from anyone who has experience with real-time Computer Vision or proctoring systems. What approach and technologies would you recommend, and how would you handle false positives and real-time performance?