The COVID-19 pandemic catalyzed a shift from traditional offline assessments to online examinations, driving technological advancements and widespread adoption of online education platforms. Online proctoring became essential to maintain assessment integrity. While offline proctoring is valued for its direct supervision and logistical control, this paper proposes an innovative approach to offline group examination proctoring using advanced image processing and machine learning techniques. Cameras installed in examination halls capture real-time video feeds, processed by TensorFlow.js models to detect anomalies such as unauthorized persons or materials. Object detection algorithms track students’ movements to ensure they remain within designated areas, leveraging techniques like Pose-Net for real-time behavior analysis and feature selection methods to identify suspicious activities. The system, powered by TensorFlow.js’s scalability and real-time processing capabilities, monitors multiple examination halls simultaneously. Additionally, traditional object detection models using the TensorFlow framework and the COCO dataset assist supervisors in identifying and mitigating cheating. This approach aims to enhance accuracy and efficiency in offline examination proctoring, upholding academic integrity with minimal manual intervention.