Design and application of smart proctoring model based on YOLOv5
Aiming at the problems of large workload and strong subjectivity of invigilators in the traditional invigilation mode,an intelligent proctoring model based on human motion detection is constructed.The model can be applied to the national standard-ized examination room to automatically monitor the cheating behavior of candidates in real time.The YOLOv5 algorithm is used to train a smart proctoring model to detect the cheating behavior of the candidates,and use the deep learning method to determine the cheating behavior.In the process of experimentation,by simulating the real examination room environment,the intelligent proctor-ing model has a high accuracy rate of detecting candidates'cheating behavior,with an accuracy of 96.3%on the training set,and real-time detection can be achieved with GPU support.At the same time,the recognition accuracy under different light and pixels is compared and analyzed,and it is proved that light and pixels will have a certain impact on accuracy.Experimental results show that the model can effectively reduce the work cost of invigilators and realize the fairness of invigilation in the examination room.
万方数据
维普
