行人属性识别研究

Research on Pedestrian Attribute Recognition

随着大量监控摄像头的出现,视频监控在刑侦破案、安防、智慧城市等领域的地位越来越重要,而行人作为视频监控的主体,对其特征的分析就显得格外重要,但由于摄像头距离行人较远,外部环境的影响和分辨率不足等原因,造成行人图像质量不高,单纯依靠人脸检测已经无法满足需求,所以对行人的多种属性分析十分必要。总结行人属性识别的现有工作,介绍多标签学习和传统的识别算法以及深度学习的方法,并对未来的工作做出展望。

In recent years,with the emergence of a large number of surveillance cameras,video surveillance is playing an increasingly important role in criminal investigation,security,smart city and other fields.As the main body of video surveillance,it is very important to analyze the characteristics of pedestrians.However,due to the fact that the camera is far away from pedestrians,the influence of external environment and insufficient resolution,the image quality of pedestrians is not high,and the simple face detection cannot meet the demand.Therefore,it is very necessary to analyze multiple attributes of pedestrians.The existing work of pedestrian attribute identification is summarized.This paper mainly introduces the multi-tag learning,the traditional recognition algorithm and the deep learning method,and makes a prospect for the future work.