计算机科学

计算机科学 ›› 2024, Vol. 51 ›› Issue (6A): 230600138-7.doi: 10.11896/jsjkx.230600138

• 图像处理&多媒体技术 • 上一篇    下一篇

一种单阶段无监督可见光-红外跨模态行人重识别方法

娄刃1, 和任强2, 赵三元2,3, 郝昕2, 周跃琪1, 汪心渊1, 李方芳4   

  1. 1 浙江省交通运输科学研究院 杭州 310000
    2 北京理工大学计算机学院 北京 100081
    3 北京理工大学长三角研究院(嘉兴) 浙江 嘉兴 314011
    4 浙江交投高速公路运营管理有限公司企业研究院 杭州 310000
  • 发布日期:2024-06-06
  • 通讯作者: 赵三元(zhaosanyuan@bit.edu.cn)
  • 作者简介:(179787711@qq.com)
  • 基金资助:
    浙江省交通运输厅科技计划项目:交通流雷视融合感知系统评测技术研究(202209);浙江省科学技术厅公益性项目:基于雷视一体设备的车辆轨迹数据质量评测技术研究(LGC22E080003)

Single Stage Unsupervised Visible-infrared Person Re-identification

LOU Ren1, HE Renqiang2, ZHAO Sanyuan2,3, HAO Xin2, ZHOU Yueqi1, WANG Xinyuan1, LI Fangfang4   

  1. 1 Zhejiang Academy of Transportation Sciences,Hangzhou 310000,China
    2 School of Computer Science,Beijing Institute of Technology,Beijing 100081,China
    3 Yangtze River Delta Research Institute(Jiaxing),Beijing Institute of Technology,Jiaxing,Zhejiang 314011,China
    4 Enterprise Institute of Zhejiang Communications Investment Expressway Operation Management Co.Ltd,Hangzhou 310000,China
  • Published:2024-06-06
  • About author:LOU Ren,born in 1982,bachelor,senior engineer.His main research interests include transportation Internet of Things and artificial intelligence.
    ZHAO Sanyuan,born in 1985,Ph.D,associate professor.Her main research interests include computer vision,deep learning and virtual reality.
  • Supported by:
    Research on Evaluation Technology of Traffic Flow based on Vision-Radar Fusion Perception System(202209) and Research on Vehicle Trajectory Data Quality Evaluation Technology Based on Radar-vision Integrated Equipment(LGC22E080003).

PDF (PC)

353

摘要： 无监督“可见光-红外”跨模态行人重识别任务能够缓解智能监控场景中需要大量人工标注的问题。常见多阶段模型用于处理不同模态数据。文中提出了一种有效的单阶段无监督跨模态行人重识别的方法,设计了基于置信因子的聚类算法和图嵌入的跨模态特征处理方法,分别用于解决无标签问题和跨模态问题。实验结果表明,相较于现有算法,所提方法在r=1时精度至少取得了7%的提高。

关键词: 跨模态学习, 无监督行人重识别, 可见光-红外行人重识别, 无监督学习, 跨模态特征处理

Abstract: The unsupervised visible-infrared multi-modal person re-identification can alleviate the problem that a lot of manual labeling is required in the intelligent monitoring scene.Common multi-stage models are used to process different modal data separately.This paper proposes an effective single-stage unsupervised cross-modal pedestrian recognition method,and designs a clustering algorithm based on confidence factor and a cross-modal feature processing method based on graph embedding to solve the unlabeled problem and cross-modal problem respectively.Experimental results show that compared with the existing algorithms,the proposed algorithm has achieved an improvement of at least 7% in the case of r=1.

Key words: Cross-modal learning, Unsupervised person re-identification, Visible-infrared person re-identification, Unsupervised learning, Cross-modal feature processing

中图分类号: 

  • TP391
[1]LIN Y,DONG X,ZHENGL,et al.A bottom-up clustering approach to unsupervised person reidentification[C]//AAAI.2019:8738-8745.
[2]LIN Y,XIE L,WU Y,et al.Unsupervisedperson re-identificationvia softened similaritylearning[C]//CVPR.2020:3390-3399.
[3]ZENG K,NING M,WANG Y,et al.Hierarchical ClusteringWith Hard-Batch Triplet Loss for Person Re-Identification[C]//CVPR.2020:13657-13665.
[4]YANG B,YE M,CHEN J,et al.Augmented Dual Contrastive Aggregation Learning for Unsupervised Visible-Infrared Person Re-Identification[C]//Proceedings of the 30th ACM International Conference on Multimedia(MM’22).2022.
[5]LIANG W,WANG G,LAIJ,et al.Homogeneous-to-Heterogeneous:Unsupervised Learning for RGB-Infrared Person Re-Identification[C]//IEEE Transactions on Image Processing.2021:6392-6407.
[6]HARDOON D R,SZEDMAK S,SHAWE-TAYLOR J.Canonical correlation analysis:An overview with application to learning methods[J].Neural Computation,2004,16(12):2639-2664.
[7]RASIWASIA N,PEREIRA J C,COVIELLO E,et al.A new approach to cross-modal multimedia retrieval[C]//ACM MM,2010:251-260.
[8]ANDREW G,ARORA R,BILMES J,et al.Deep canonical correlation analysis[C]//ICML,2013:1247-1255.
[9]FENG F,WANG X,LI R.Cross-modal retrieval with correspondence autoencoder[C]//ACM MM.2014:7-16.
[10]ZHAN Y,YU J,YU Z,et al.Comprehensive distance-preserving autoencoders for cross-modal retrieval[C]//ACM MM.2018:1137-1145.
[11]HE L,XU X,LU H,et al.Unsupervised cross-modal retrievalthrough adversarial learning[C]//ICME.2017:1153-1158.
[12]CHUNG Y A,WENG W H,TONG S,et al.Unsupervisedcross-modal alignment of speech and textembedding spaces[C]//Proceedings of the 32nd International Conference on Neural Information Processing Systems(NIPS’18).2018:7365-7375.
[13]LI C,DENG C,LI N,et al.Self-super visedadversarial hashing networks fo rcross modal retrieval[C]//CVPR.2018:4242-4251.
[14]WANG B,YANG Y,XU X,et al.Adversarial cross-modal re-trieval[C]//ACM MM,2017:154-162.
[15]YAN S,XU D,ZHANG B,et al.Graph embedding and extensions:A general framework for dimensionality reduction[J].IEEE TPAMI,2006,29(1):40-51.
[16]MORSING L H,SHEIKH-OMARO A,IOSIFIDIS A.Supervised domain adaptation:A graph embedding perspective and a rectified experimental protocol[J].arXiv:2004.11262,2020.
[17]MOON H,PHILLIPSP J.Computational and performance as-pects of PCA-based face-recognition algorithms[J].Perception,2001,30(3):303-321.
[18]YU S,LI S,CHEN D,et al.COCAS:A Large-Scale Clothes Changing Person Dataset for ReIdentification[C]//CVPR.2020:3400-3409.
[19]SONG L,WANG C,ZHANG L,et al.Unsupervised domainadaptive re-identification:Theory and practice[J].Pattern Re-cognition,2020,102:107173.
[20]ZHU J Y,PARK T,ISOLA P,et al.Unpaired image-to-imagetranslation using cycle-consistent adversarial networks[C]//ICCV.2017:2223-2232.
[21]FU Y,WEI Y,WANG G,et al.Self-similarity grouping:A simple unsupervised cross domain adaptation approach for person re-identification[C]//ICCV.2019:6112-6121.
[22]ZHONG Z,ZHENG L,LUO Z,et al.Invariance matters:Exemplar memory for domain adaptive person re-identification[C]//CVPR.2019:598-607.
[23]DALAL N,TRIGGS B.Histograms of oriented gradients forhuman detection[C]//CVPR.2005:886-893.
[24]XIONG F,GOU M,CAMPS O,et al.Person re-identificationusing kernel-based metric learning methods[C]//ECCV.2014:1-16.
[25]MATSUKAWA T,OKABE T,SUZUKIE,et al.Hierarchicalgaussian descriptor for person re-identification[C]//CVPR.2016:1363-1372.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发