SSD-Based Intelligent Invigilation System for Multi-Modal Behavioral Features in Examination Scenarios

Against the background of low efficiency and lack of accuracy of traditional invigilation methods, this paper proposes a solution to design and implement an intelligent invigilation system to improve the level of examination invigilation intelligence. The system includes three user roles: administrators, invigilators, and candidates. Its functional modules include user management, data acquisition, and preprocessing. The core algorithm adopts a Single Shot MultiBox Detector(SSD) target detection algorithm, which is used to analyze and judge the abnormal behaviors of the candidates intelligently. The experimental results show that this intelligent invigilation system solves the problems of traditional invigilation well and effectively improves the intelligence level and management efficiency of exam invigilation.