Two-Stage Object-Centric Framework Automates Exam Cheating Detection at Scale
arXiv·low signal
Le et al. present a two-stage detection framework combining object detection with behavioral analysis for automated exam proctoring. While narrow in scope, the architecture demonstrates a reusable pattern for real-time multi-stage vision pipelines: detect objects of interest first, then classify behavioral patterns — applicable to broader compliance monitoring and activity recognition.