【摘要】 针对校园封闭管理下的学生翻越围栏、偷取外卖等现象，提出一种基于YOLOv5的校园围栏场景下行人异常行为检测系统。该系统首先对监控视频中提取的图像进行网络训练，模型训练完成后以此来检测视频中的翻越、攀爬栅栏围墙等异常行为。当检测到与围栏距离过近的人员存在疑似异常行为时，系统触发警报模块，警示学生保持与围栏间的距离。更多还原

【Abstract】 In view of the phenomenon of students jumping over the fence and stealing takeout in the closed campus during the prevention and control period, a YOLOv5 based pedestrian abnormal behavior detection system under the campus fence scenario is proposed. Firstly, the system carries out network training on the images extracted from the surveillance video, and uses the trained offline network model to automatically monitor the abnormal behaviors of people in the surveillance video, such as jumping over and climbing. When detecting the suspected abnormal behavior of those closed to the fence, the system could issue an alarm to remind students to keep the distance from the fence.更多还原