基于智能视觉物联网的教室人数跟踪系统

People Tracking System Based on Intelligent Vision Internet of Things for Classroom

摘要: 为实时统计教室人数，利用教室现有的视频监控系统获取教室实时图像，搭建可靠、稳定的人数跟踪应用平台。结合教室内人员分布特点及教室建筑风格，通过检测学生头部特征，使用AdaBoost级联分类器对提取的视频帧HOG样本特征进行训练，从而实现头部检测模型用于检测人数。最后，开发基于Android平台的手机客户端，供用户实时查询教室的人数情况。系统测试和结果分析表明：该系统对教室人数检测有良好的识别率。在不遮挡头部特征的条件下，系统识别准确率高于80%，基本可满足教室人数实时统计的需求。

Abstract: To count the number of people in the classroom in real time, the classroom video monitoring system is used to generate the real-time images in the classroom. A reliable and stable people tracking application platform is realized in this paper. Taking into account the people distribution characteristics in the classroom and the architectural style of the classroom, this paper adopts a head detecting algorithm. The extracted video frame HOG sample features are then trained using the AdaBoost cascade classifier to implement a head detection model for detecting the number of people. Finally, the mobile client based on the Android platform is developed for the user to query the number of people in the classroom in real time. System testing results show that the system has an excellent recognition rate. In the case of not hiding the head characteristics, the system recognition rate is more than 75%, which can basically satisfy the requirements of people counting in real time.