教室监控下学生异常行为检测系统

针对教室监控中学生异常行为无法实时检测并反馈的现状,设计了一套基于YOLO v3算法的教室监控学生异常行为检测系统,包括摄像头硬件采集、异常行为识别和响应三个模块.其中采用基于数据标签的随机擦除预处理方法模拟图像中的目标被遮挡的情形,提高网络的泛化能力,使得网络仅通过学习局部特征即可完成目标的检测和识别;其次改进了YO...

Classroom Monitoring Students Abnormal Behavior Detection System

Aiming at the status quo that abnormal behaviors of students in classroom monitoring cannot be detected and feedback in real time, a classroom monitoring student abnormal behavior detection system based on YOLO v3 algorithm is designed, which includes three modules：camera hardware acquisition, abnormal behavior recognition and response. Among them, the random erasure preprocessing method based on data labels is used to simulate the situation where the target in the image is occluded, and the generalization ability of the network is improved, so that the network can complete the detection and recognition of the target only by learning local features. Secondly, it improves YOLO v3 algorithm’s backbone network, Darknet, expands the shallow network so that it is not easy for the network to ignore the edges of pictures or small target objects. The improved algorithm effectively increases the speed and accuracy of student abnormal behavior detection and reduces the missed detection rate, basically meeting the requirements of real-time detection tasks, which reduces the teachers’ workload to a certain extent and improving classroom efficiency.