[{"data":1,"prerenderedAt":298},["ShallowReactive",2],{"content-query-XoxhdvQkrH":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"body":13,"_type":292,"_id":293,"_source":294,"_file":295,"_stem":296,"_extension":297},"/technology-blogs/en/2949","en",false,"","Research on Person Re-identification and Gait Recognition Based on MindSpore to Address Cloth-Changing Challenges","Author: Li Ruifeng | Source: Zhihu","2023-12-01","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/01/19/07185c2122274635bfac29ae82cde1e8.png","technology-blogs",{"type":14,"children":15,"toc":289},"root",[16,24,33,38,43,48,53,58,69,74,83,96,101,106,114,122,127,135,143,148,160,168,176,181,189,194,201,206,211,218,223,231,239,244,251,258,263,268,276,284],{"type":17,"tag":18,"props":19,"children":21},"element","h1",{"id":20},"research-on-person-re-identification-and-gait-recognition-based-on-mindspore-to-address-cloth-changing-challenges",[22],{"type":23,"value":8},"text",{"type":17,"tag":25,"props":26,"children":27},"p",{},[28],{"type":17,"tag":29,"props":30,"children":31},"strong",{},[32],{"type":23,"value":9},{"type":17,"tag":25,"props":34,"children":35},{},[36],{"type":23,"value":37},"Paper Title",{"type":17,"tag":25,"props":39,"children":40},{},[41],{"type":23,"value":42},"An In-Depth Exploration of Person Re-Identification and Gait Recognition in Cloth-Changing Conditions",{"type":17,"tag":25,"props":44,"children":45},{},[46],{"type":23,"value":47},"Paper Source",{"type":17,"tag":25,"props":49,"children":50},{},[51],{"type":23,"value":52},"CVPR 2023",{"type":17,"tag":25,"props":54,"children":55},{},[56],{"type":23,"value":57},"Paper URL",{"type":17,"tag":25,"props":59,"children":60},{},[61],{"type":17,"tag":62,"props":63,"children":67},"a",{"href":64,"rel":65},"https://ieeexplore.ieee.org/document/10204291",[66],"nofollow",[68],{"type":23,"value":64},{"type":17,"tag":25,"props":70,"children":71},{},[72],{"type":23,"value":73},"Code URL",{"type":17,"tag":25,"props":75,"children":76},{},[77],{"type":17,"tag":62,"props":78,"children":81},{"href":79,"rel":80},"https://gitee.com/chunjie-zhang/ccpg-cvpr2023",[66],[82],{"type":23,"value":79},{"type":17,"tag":25,"props":84,"children":85},{},[86,88,94],{"type":23,"value":87},"As an open-source AI framework, MindSpore supports ultra-large-scale AI pre-training and brings excellent experience of device-edge-cloud synergy, simplified development, ultimate performance, and security and reliability for researchers and developers. Since it was open-sourced on March 28th, 2020, MindSpore has been downloaded for more than 6 million times. It has also been the subject of hundreds of papers presented at premier AI conferences. Furthermore, it has a large community of developers and has been introduced in over 100 universities and 5000 commercial apps. Being widely used in scenarios such as AI computing centers, finance, smart manufacturing, cloud, wireless, datacom, energy, \"1+8+",{"type":17,"tag":89,"props":90,"children":91},"em",{},[92],{"type":23,"value":93},"N",{"type":23,"value":95},"\" consumer, and smart automobiles, MindSpore has emerged as one of the leading open-source software on Gitee. The MindSpore community extends a warm welcome to all who wish to contribute to open-source development kits, models, industrial applications, algorithm innovations, academic collaborations, AI-themed book writing, and application cases across the cloud, device, edge, and security.",{"type":17,"tag":25,"props":97,"children":98},{},[99],{"type":23,"value":100},"Thanks to the support from scientific, industry, and academic circles, MindSpore-based papers account for 7% of all papers about AI frameworks in 2023, ranking No. 2 globally for two consecutive years. The MindSpore community is thrilled to share and interpret top-level conference papers and is looking forward to collaborating with experts from industries, academia, and research institutions, so as to yield proprietary AI outcomes and innovate AI applications. In this blog, I'd like to share the paper of the team led by Prof. Zhang Chunjie, School of Computer and Information Technology at Beijing Jiaotong University.",{"type":17,"tag":25,"props":102,"children":103},{},[104],{"type":23,"value":105},"This blog mainly focuses on person re-identification (ReID) and gait recognition. To explore the cloth-change problem in an actual scenario, we published the first cloth-changing video dataset for person ReID, and proposed a gait recognition method for the cloth-changing scenario as the standard method for the dataset. The code for person ReID and gait recognition can be easily implemented based on MindSpore official documents or object detection code and models provided by the community.",{"type":17,"tag":25,"props":107,"children":108},{},[109],{"type":17,"tag":29,"props":110,"children":111},{},[112],{"type":23,"value":113},"01",{"type":17,"tag":25,"props":115,"children":116},{},[117],{"type":17,"tag":29,"props":118,"children":119},{},[120],{"type":23,"value":121},"Research Background",{"type":17,"tag":25,"props":123,"children":124},{},[125],{"type":23,"value":126},"The rapid development of deep learning promotes the technologies used for person recognition and grounds them from theory to application, contributing to social stability and harmonious development. Specifically, person ReID and gait recognition are attracting attention. The goals of the two technologies are consistent, that is, recognition and matching of target persons using image and video data provided by surveillance cameras. However, in practice, person ReID and gait recognition are challenged with problems such as person cloth-changing, blocking, and low-quality images. Although applied to practical scenarios and achieving certain recognition accuracy, the two technologies still lack researches on the recognition accuracy. This is due to the gap in a proper video dataset for person ReID, which limits the volume of related researches. Meanwhile, gait recognition is usually conducted indoors because most gait datasets are collected in a lab environment with controlled collection conditions. To study the cloth-changing problem in reality, we need a more suitable dataset to implement systematic researches.",{"type":17,"tag":25,"props":128,"children":129},{},[130],{"type":17,"tag":29,"props":131,"children":132},{},[133],{"type":23,"value":134},"02",{"type":17,"tag":25,"props":136,"children":137},{},[138],{"type":17,"tag":29,"props":139,"children":140},{},[141],{"type":23,"value":142},"Team Introduction",{"type":17,"tag":25,"props":144,"children":145},{},[146],{"type":23,"value":147},"Li Weijia, the first author of the paper, graduated in 2023 with a master's degree from the School of Computer and Information Technology, Beijing Jiaotong University. His research focuses on computer vision and gait recognition.",{"type":17,"tag":25,"props":149,"children":150},{},[151,153,158],{"type":23,"value":152},"The Center of Digital Media Information Processing (MePro) of Beijing Jiaotong University started in 1998 and was selected as the Innovation Team Development Plan of the Ministry of Education in 2012. MePro consists of 14 teachers and more than 100 master's and PhD students. Its research field focuses on digital media information processing, including image/video coding and transmission, digital watermark and forensics, and media content analysis and understanding. In the year 2022, the lab made significant contributions to the field by publishing 61 high-impact papers, including 38 papers in the esteemed international journal ",{"type":17,"tag":89,"props":154,"children":155},{},[156],{"type":23,"value":157},"IEEE Trans",{"type":23,"value":159}," and 23 papers presented at top-tier international conferences like NeurIPS, CVPR, ECCV, and ACM MM.",{"type":17,"tag":25,"props":161,"children":162},{},[163],{"type":17,"tag":29,"props":164,"children":165},{},[166],{"type":23,"value":167},"03",{"type":17,"tag":25,"props":169,"children":170},{},[171],{"type":17,"tag":29,"props":172,"children":173},{},[174],{"type":23,"value":175},"Introduction to the Paper",{"type":17,"tag":25,"props":177,"children":178},{},[179],{"type":23,"value":180},"The paper introduces research on video-based person ReID and gait recognition. The goal of ReID and gait recognition is to match the target pedestrian under surveillance cameras. The two technologies are thus vitally important in surveillance analysis. However, cloth-changing affects their recognition accuracy. There was little work on this problem because of the lack of datasets. To tackle this problem, the authors propose a Cloth-Changing benchmark for Person re-identification and Gait recognition (CCPG) which provides 200 identities and over 16,000 sequences. RGB and silhouette version data involving various indoor and outdoor cloth-changing situations are available in this dataset for research purposes. In this way, the dataset becomes a benchmark for comparing person ReID and gait recognition in such conditions. Compared with other datasets, this dataset features richer samples in cloth-changing types and scenes.",{"type":17,"tag":25,"props":182,"children":183},{},[184],{"type":17,"tag":185,"props":186,"children":188},"img",{"alt":7,"src":187},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/01/19/981b9bde80664d7791a12fb9266a1522.png",[],{"type":17,"tag":25,"props":190,"children":191},{},[192],{"type":23,"value":193},"comparison between CCPG and other datasets",{"type":17,"tag":25,"props":195,"children":196},{},[197],{"type":17,"tag":185,"props":198,"children":200},{"alt":7,"src":199},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/01/19/2c0e866f8f63485baaa5b4c8cc9321c9.png",[],{"type":17,"tag":25,"props":202,"children":203},{},[204],{"type":23,"value":205},"CCPG and sub-datasets",{"type":17,"tag":25,"props":207,"children":208},{},[209],{"type":23,"value":210},"It should be noted that, based on the availability of information of the shoe and face areas in practice, the authors mask the two areas and obtain three sub-datasets for further experiments. Finally, a series of experiments on person ReID and gait recognition in the cloth-changing condition are carried out based on CCPG, proving the advantage of gait recognition in solving the cloth-changing problem. In addition, AUG-OGBase, a simple and efficient gait recognition method, achieves satisfactory results in the CCPG dataset, which further demonstrates the potential of gait recognition in tackling this issue.",{"type":17,"tag":25,"props":212,"children":213},{},[214],{"type":17,"tag":185,"props":215,"children":217},{"alt":7,"src":216},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/01/19/1f8ffd0282e64ebaa182186e4efca795.png",[],{"type":17,"tag":25,"props":219,"children":220},{},[221],{"type":23,"value":222},"AUG-OGBase network design",{"type":17,"tag":25,"props":224,"children":225},{},[226],{"type":17,"tag":29,"props":227,"children":228},{},[229],{"type":23,"value":230},"04",{"type":17,"tag":25,"props":232,"children":233},{},[234],{"type":17,"tag":29,"props":235,"children":236},{},[237],{"type":23,"value":238},"Experimental Result",{"type":17,"tag":25,"props":240,"children":241},{},[242],{"type":23,"value":243},"On the CCPG dataset, the authors compare the conventional video-based ReID method and the gait recognition method. The experimental results are further analyzed and studied.",{"type":17,"tag":25,"props":245,"children":246},{},[247],{"type":17,"tag":185,"props":248,"children":250},{"alt":7,"src":249},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/01/19/dacf6b4e1b1941c4ac604fb45cc92664.png",[],{"type":17,"tag":25,"props":252,"children":253},{},[254],{"type":17,"tag":185,"props":255,"children":257},{"alt":7,"src":256},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2024/01/19/4d84aea1456a40a0a09e175b0c199621.png",[],{"type":17,"tag":25,"props":259,"children":260},{},[261],{"type":23,"value":262},"Experiments show that the ReID method can achieve high recognition accuracy when the shoe and face area information is available. However, if the information is unavailable, the performance of ReID will be greatly affected. In other words, ReID still identifies a person based on appearance information as it lacks identification of dynamic information (for example, gait features). Therefore, identification accuracy decreases in the cloth-changing condition, posing limits on the ReID method in solving the cloth-changing problem.",{"type":17,"tag":25,"props":264,"children":265},{},[266],{"type":23,"value":267},"Gait recognition outperforms ReID mainly because it focuses more on the spatial statics and temporal dynamics of the human body than on the appearance features. As a result, gait recognition can better utilize information irrelevant to clothes for recognition, thus achieving higher recognition accuracy. Moreover, it is further demonstrated that gait recognition carries specific potential for addressing the cloth-changing challenge.",{"type":17,"tag":25,"props":269,"children":270},{},[271],{"type":17,"tag":29,"props":272,"children":273},{},[274],{"type":23,"value":275},"05",{"type":17,"tag":25,"props":277,"children":278},{},[279],{"type":17,"tag":29,"props":280,"children":281},{},[282],{"type":23,"value":283},"Summary and Prospects",{"type":17,"tag":25,"props":285,"children":286},{},[287],{"type":23,"value":288},"This paper mainly introduces the CCPG dataset used to study the performance of person ReID and gait recognition in the case of cloth-changing. The dataset contains 200 persons and more than 16,000 pedestrian sequences for person ReID and gait recognition. This paper also analyzes the experiments conducted based on the CCPG dataset. The results show that gait recognition has higher potential in solving the cloth-changing issue. The CCPG dataset and experimental results provide a new scheme to tackle the cloth-changing issue, which is conducive to the development of social security technologies around the world.",{"title":7,"searchDepth":290,"depth":290,"links":291},4,[],"markdown","content:technology-blogs:en:2949.md","content","technology-blogs/en/2949.md","technology-blogs/en/2949","md",1776506108298]