Automated Remote Proctoring and Malpractice Detection Using Facial Tracking and Object Recognition Models
Nijampatnam Hari Krishna,
Surapaneni Thanmai Sai,
Sayyad Nagul Saida,
Sajja Baji Babu,
Saranu Rama Krishna
Examinations are widely used to assess student knowledge, skills, and academic performance. However,
main- taining examination integrity in remote and large-scale envi- ronments is a major challenge.
Traditional monitoring methods such as manual invigilation and simple camera observation are timeconsuming,
inconsistent, and prone to human error. This paper presents an automated remote proctoring
and mal- practice detection system that integrates facial tracking and object recognition in a unified
computer vision framework. The proposed system monitors candidates through live video streams to
detect prolonged face absence, multiple faces, suspicious head movements, gaze deviation, and prohibited
objects such as mobile phones, books, and paper notes. Instead of directly accusing students, the system
generates an exam integrity report with a cumulative suspicion score to support fair decision-making by
invigilators. The proposed framework improves monitoring consistency, reduces human workload, and
provides scalable support for examination halls and supervised remote assessments.
