以行人再识别技术为核心的多算法融合模型在网格化管理中的应用

Multi-algorithm fusion model with Re-identification pedestrian recognition technology as the core application in grid management

近年来,随着城市的发展,居民购房迁移、外地人口大量涌入,流动人口规模急剧膨胀,人户分离情况尤为突出,给城市社区的管理带来新的挑战.文章以社区网格化管理中的人员管理为切入点,利用深度学习方法下的ReID行人再识别、人脸识别算法与多算法融合技术,对社区内的人员进行人脸、人体图像采集,并进行特征比对、图像聚类处理,形成一人一档数据,实现对社区人员活动轨迹的刻画.某住宅小区自应用以来,已新增登记常住人口561人、暂住人口298人,小区实有人口登记率从41%提升至80%以上.

In recent years,with the development of the city,local residents have purchased houses and migrated,a large influx of foreign population,the size of the floating population has expanded rapidly,and the separation of households is particularly prominent,which has brought new challenges to the management of urban communities.Taking the personnel management in community grid management as the entry point,this paper uses Re-identification pedestrian rerecognition,face recognition algorithm and multi-algorithm fusion technology under the deep learning method to collect face and human image of the people in the community,and carry out feature comparison and image clustering processing,so as to form one-person-one-file data,so as to realize the characterization of the activity trajectory of community personnel.Since the application of residential quarters,561 permanent residents and 298 temporary residents have been newly registered.The actual population registration rate in residential quarters has increased from 41%to over 80%.