Flow analysis method based on human and object weight identification

The invention discloses a traffic analysis method based on person and object weight identification. The method comprises the following steps: S1, obtaining visitor original video data collected by a camera installed in a shop; s2, extracting data containing figure images from the original video data; s3, performing de-weighting processing based on the recognition of the human face and the corresponding worn clothes; the feature values of the human face and the corresponding worn clothes are not repeatedly stored any more, and the repeated feature values are deleted; s4, after duplicate removal, performing character labeling and statistics, and establishing a database; and S5, in the target time period, calculating and analyzing the flow according to the statistical data. The accuracy of passenger flow volume analysis can be improved. 本发明公开了一种基于人与物重识别的流量分析方法，其包括如下步骤：S1.获取安装在店铺内摄像头采集的访客原始视频数据；S2.从原始视频数据中提取包含人物图像的数据；S3.基于人脸和对应身穿的衣物的识别，进行去重处理；不再重复保存人脸和对应身穿的衣物的特征值，删除重复的特征值；S4.去重后，进行人物标注与统计，建立数据库；S5.在目标时间段，根据统计的数据计