基于深度学习的行人重识别研究综述
|
摘要
由于视角、背景、光照条件和相互遮挡等因素的变化,行人重识别是一个具有挑战性的问题.近年来,许多研究者将深度学习的方法引入到行人重识别研究中,并获得了较好的重识别结果.本文介绍了基于深度学习的行人重识别的主要研究方法(局部特征学习、距离度量学习、基于视频序列学习和生成对抗网络),并介绍目前常用的用于深度学习的行人重识别数据集(Duke MTMC-reID、CUHK03和Market1501)及其存在的问题,同时,对行人重识别提出了自己的理解和观点.最后指出了未来可能的研究方向.
Due to changes in perspectives,backgrounds,lighting conditions,and mutual occlusion,person re-identification is still a challenging issue.In recent years,many researchers have introduced deep learning methods into person re-identification research and obtained better re-identification results.This paper introduces the main research methods of person re-identification based on deep learning(local feature learning,distance metric learning,video sequence learning,and generation of confrontation networks),and introduces commonly used person re-identification data sets for deep learning(Duke MTMC-reID,CUHK03,and Market1501) and their existing problems.At the same time,it puts forward their own understanding and viewpoints on person re-identification,and finally points out possible future research directions.
Due to changes in perspectives,backgrounds,lighting conditions,and mutual occlusion,person re-identification is still a challenging issue.In recent years,many researchers have introduced deep learning methods into person re-identification research and obtained better re-identification results.This paper introduces the main research methods of person re-identification based on deep learning(local feature learning,distance metric learning,video sequence learning,and generation of confrontation networks),and introduces commonly used person re-identification data sets for deep learning(Duke MTMC-reID,CUHK03,and Market1501) and their existing problems.At the same time,it puts forward their own understanding and viewpoints on person re-identification,and finally points out possible future research directions.
引文
[1] BAI X,YANG M K,HUANG T T,et al. Deep-person:learning discriminative deep features for person re-identification[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1711.10658.pdf.
[2] VARIOR R R,SHUAI B,LU J W,et al. A siamese long short-term memory architecture for human re-identification[C]//Proceedings of the European Conference on Computer Vision. Cham:Springer,2016:135-153.
[3] ZHAO H Y,TIAN M Q,SUN S Y,et al. Spindle net:person re-identification with human body region guided feature decompo-sition and fusion[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Hawaii,2017:1077-1085.
[4] WEI L H,ZHANG S L,YAO H T,et al. Glad:global-local-alignment descriptor for pedestrian retrieval[C]//Proceedings ofthe 2017 ACM on Multimedia Conference. California,2017:420-428.
[5] ZHANG L,XIANG T,GONG S G. Learning a discriminative null space for person re-identification[C]//Proceedings of theIEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1239-1248.
[6] ZHENG L,YANG Y,HAUPTMANN A G. Person re-identification:past,present and future[DB/OL].[2018-10-22].https://arxiv.org/pdf/1610.02984.pdf.
[7] ZHOU S,WANG J J,WANG J Y,et al. Point to set similarity based deep feature learning for person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Hawaii,2017:3741-3750.
[8] HERMANS A,BEYER L,LEIBE B. In defense of the triplet loss for person re-identification[DB/OL].[2018-10-22].https://arxiv.org/pdf/1703.07737.pdf.
[9] CHENG D,GONG Y H,ZHOU S P,et al. Person re-identification by multi-channel parts-based CNN with improved triplet lossfunction[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1335-1344.
[10] CHEN W H,CHEN X T,ZHANG J G,et al. Beyond triplet loss:a deep quadruplet network for person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Hawaii,2017:403-412.
[11] XIAO Q Q,LUO H,ZHANG C. Margin sample mining loss:a deep learning based method for person re-identification[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1710.00478.pdf.
[12] MCLAUGHLIN N,RINCON J M D,MILLER P. Recurrent convolutional network for video-based person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1325-1334.
[13] ZHANG D Y,WU W X,CHENG H,et al. Image-to-video person re-identification with temporally memorized similaritylearning[J]. IEEE transactions on circuits&systems for video technology,2017,PP(99):1-1.
[14] MCLAUGHLIN N,RINCON J M D,MILLER P. Recurrent convolutional network for video-based person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1325-1334.
[15] HUANG W J,LIANG C,YU Y,et al. Video-based person re-identification via self paced weighting[C]//Proceedings of theThirty-Second Conference on Artificial Intelligence. Louisiana,2018:2273-2280.
[16] LIU H,JIE Z Q,JAYASHREE K,et al. Video-based person re-identification with accumulative motion context[J]. IEEEtransactions on circuits&systems for video technology,2017,PP(99):1-1.
[17] SONG G L,LENG B,LIU Y,et al. Region-based quality estimation network for large-scale person re-identification[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1711.08766.pdf.
[18] ZHENG Z D,ZHENG L,YANG Y. Unlabeled samples generated by gan improve the person re-identification baseline in vitro[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1701.07717.pdf.
[19] ZHONG Z,ZHENG L,ZHENG Z D,et al. Camera style adaptation for person re-identification[DB/OL].[2018-10-22].https://arxiv.org/pdf/1711.10295.pdf.
[20] WEI L H,ZHANG S L,GAO W,et al. Person transfer GAN to bridge domain gap for person re-identification[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1711.08565.pdf.
[21] QIAN X L,FU Y W,WANG W,et al. Pose-normalized image generation for person re-identification[DB/OL].[2018-10-22].https://arxiv.org/pdf/1712.02225.pdf.
[22] HE L X,LIANG J,LI H Q,et al. Deep Spatial feature reconstruction for partial person re-identification:alignment-free approach[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1801.00881.pdf.
[23] SUN Y F,ZHENG L,DENG W J,et al. Svdnet for pedestrian retrieval[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1703.05693.pdf.
[24] ZHONG Z,ZHENG L,CAO D,et al. Re-ranking person re-identification with k-reciprocal encoding[C]//2017 IEEEConference on Computer Vision and Pattern Recognition(CVPR). Hawaii:IEEE,2017:3652-3661.
[25] ZHONG Z,ZHENG L,KANG G L,et al. Random erasing data augmentation[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1708.04896.pdf.
[26] SARFRAZ M S,SCHUMANN A,EBERLE A,et al. A pose-sensitive embedding for person re-identification with expandedcross neighborhood re-ranking[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1711.10378.pdf.
[27] FAN H H,ZHENG L,YANG Y. Unsupervised person re-identification:clustering and fine-tuning[DB/OL].[2018-10-22].https://arxiv.org/pdf/1705.10444.pdf.
[28] BAK S,CARR P,LALONDE J F,et al. Domain adaptation through synthesis for unsupervised person re-identification[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1804.10094.pdf.
[29] JOSE C,FLEURET F. Scalable metric learning via weighted approximate rank component analysis[C]//European Conferenceon Computer Vision. Cham:Springer,2016:875-890.
[30] VARIOR R R,HALOI M,WANG G. Gated siamese convolutional neural network architecture for human re-identification[C]//European Conference on Computer Vision. Cham:Springer,2016:791-808.
[31] XIAO T,LI H S,OUYANG W L,et al. Learning deep feature representations with domain guided dropout for person re-identi-fication[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1249-1258.
[32] ZHOU Z,HUANG Y,WANG W,et al. See the forest for the trees:joint spatial and temporal recurrent neural networks forvideo-based person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Hawaii,2017:6776-6785.
[2] VARIOR R R,SHUAI B,LU J W,et al. A siamese long short-term memory architecture for human re-identification[C]//Proceedings of the European Conference on Computer Vision. Cham:Springer,2016:135-153.
[3] ZHAO H Y,TIAN M Q,SUN S Y,et al. Spindle net:person re-identification with human body region guided feature decompo-sition and fusion[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Hawaii,2017:1077-1085.
[4] WEI L H,ZHANG S L,YAO H T,et al. Glad:global-local-alignment descriptor for pedestrian retrieval[C]//Proceedings ofthe 2017 ACM on Multimedia Conference. California,2017:420-428.
[5] ZHANG L,XIANG T,GONG S G. Learning a discriminative null space for person re-identification[C]//Proceedings of theIEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1239-1248.
[6] ZHENG L,YANG Y,HAUPTMANN A G. Person re-identification:past,present and future[DB/OL].[2018-10-22].https://arxiv.org/pdf/1610.02984.pdf.
[7] ZHOU S,WANG J J,WANG J Y,et al. Point to set similarity based deep feature learning for person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Hawaii,2017:3741-3750.
[8] HERMANS A,BEYER L,LEIBE B. In defense of the triplet loss for person re-identification[DB/OL].[2018-10-22].https://arxiv.org/pdf/1703.07737.pdf.
[9] CHENG D,GONG Y H,ZHOU S P,et al. Person re-identification by multi-channel parts-based CNN with improved triplet lossfunction[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1335-1344.
[10] CHEN W H,CHEN X T,ZHANG J G,et al. Beyond triplet loss:a deep quadruplet network for person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Hawaii,2017:403-412.
[11] XIAO Q Q,LUO H,ZHANG C. Margin sample mining loss:a deep learning based method for person re-identification[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1710.00478.pdf.
[12] MCLAUGHLIN N,RINCON J M D,MILLER P. Recurrent convolutional network for video-based person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1325-1334.
[13] ZHANG D Y,WU W X,CHENG H,et al. Image-to-video person re-identification with temporally memorized similaritylearning[J]. IEEE transactions on circuits&systems for video technology,2017,PP(99):1-1.
[14] MCLAUGHLIN N,RINCON J M D,MILLER P. Recurrent convolutional network for video-based person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1325-1334.
[15] HUANG W J,LIANG C,YU Y,et al. Video-based person re-identification via self paced weighting[C]//Proceedings of theThirty-Second Conference on Artificial Intelligence. Louisiana,2018:2273-2280.
[16] LIU H,JIE Z Q,JAYASHREE K,et al. Video-based person re-identification with accumulative motion context[J]. IEEEtransactions on circuits&systems for video technology,2017,PP(99):1-1.
[17] SONG G L,LENG B,LIU Y,et al. Region-based quality estimation network for large-scale person re-identification[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1711.08766.pdf.
[18] ZHENG Z D,ZHENG L,YANG Y. Unlabeled samples generated by gan improve the person re-identification baseline in vitro[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1701.07717.pdf.
[19] ZHONG Z,ZHENG L,ZHENG Z D,et al. Camera style adaptation for person re-identification[DB/OL].[2018-10-22].https://arxiv.org/pdf/1711.10295.pdf.
[20] WEI L H,ZHANG S L,GAO W,et al. Person transfer GAN to bridge domain gap for person re-identification[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1711.08565.pdf.
[21] QIAN X L,FU Y W,WANG W,et al. Pose-normalized image generation for person re-identification[DB/OL].[2018-10-22].https://arxiv.org/pdf/1712.02225.pdf.
[22] HE L X,LIANG J,LI H Q,et al. Deep Spatial feature reconstruction for partial person re-identification:alignment-free approach[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1801.00881.pdf.
[23] SUN Y F,ZHENG L,DENG W J,et al. Svdnet for pedestrian retrieval[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1703.05693.pdf.
[24] ZHONG Z,ZHENG L,CAO D,et al. Re-ranking person re-identification with k-reciprocal encoding[C]//2017 IEEEConference on Computer Vision and Pattern Recognition(CVPR). Hawaii:IEEE,2017:3652-3661.
[25] ZHONG Z,ZHENG L,KANG G L,et al. Random erasing data augmentation[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1708.04896.pdf.
[26] SARFRAZ M S,SCHUMANN A,EBERLE A,et al. A pose-sensitive embedding for person re-identification with expandedcross neighborhood re-ranking[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1711.10378.pdf.
[27] FAN H H,ZHENG L,YANG Y. Unsupervised person re-identification:clustering and fine-tuning[DB/OL].[2018-10-22].https://arxiv.org/pdf/1705.10444.pdf.
[28] BAK S,CARR P,LALONDE J F,et al. Domain adaptation through synthesis for unsupervised person re-identification[DB/OL].[2018-10-22]. https://arxiv.org/pdf/1804.10094.pdf.
[29] JOSE C,FLEURET F. Scalable metric learning via weighted approximate rank component analysis[C]//European Conferenceon Computer Vision. Cham:Springer,2016:875-890.
[30] VARIOR R R,HALOI M,WANG G. Gated siamese convolutional neural network architecture for human re-identification[C]//European Conference on Computer Vision. Cham:Springer,2016:791-808.
[31] XIAO T,LI H S,OUYANG W L,et al. Learning deep feature representations with domain guided dropout for person re-identi-fication[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas,2016:1249-1258.
[32] ZHOU Z,HUANG Y,WANG W,et al. See the forest for the trees:joint spatial and temporal recurrent neural networks forvideo-based person re-identification[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Hawaii,2017:6776-6785.