Unsupervised Person Re-Identification: A Systematic Survey of Challenges and Solutions

Lin, Xiangtan; Ren, Pengzhen; Yeh, Chung‐Hsing; Yao, Lina; Song, Andy; Chang, Xiaojun

doi:10.48550/arxiv.2109.06057

Cited by 3 publications

(3 citation statements)

References 142 publications

(210 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…State-of-the-art unsupervised person Re-ID is based on clustering methods [23][24][25]. Many techniques have been offered to improve the estimation of pseudo-labels.…”

Section: Related Work 21 Unsupervised Person Re-idmentioning

confidence: 99%

UnA-Mix: Rethinking Image Mixtures for Unsupervised Person Re-Identification

Liu,

Sun,

Liu

et al. 2024

Processes

View full text Add to dashboard Cite

With the development of ultra-long-range visual sensors, the application of unsupervised person re-identification algorithms to them has become increasingly important. However, these algorithms inevitably generate noisy pseudo-labels, which seriously hinder the performance of tasks over a large range. Mixup, a data enhancement technique, has been validated in supervised learning for its generalization to noisy labels. Based on this observation, to our knowledge, this study is the first to explore the impact of the mixup technique on unsupervised person re-identification, which is a downstream task of contrastive learning, in detail. Specifically, mixup was applied in different locations (at the pixel level and feature level) in an unsupervised person re-identification framework to explore its influences on task performance. In addition, based on the richness of the information contained in the person samples to be mixed, we propose an uncertainty-aware mixup (UnA-Mix) method, which reduces the over-learning of simple person samples and avoids the information damage that occurs when information-rich person samples are mixed. The experimental results on three benchmark person re-identification datasets demonstrated the applicability of the proposed method, especially on the MSMT17, where it outperformed state-of-the-art methods by 5.2% and 4.8% in terms of the mAP and rank-1, respectively.

show abstract

“…State-of-the-art unsupervised person Re-ID is based on clustering methods [23][24][25]. Many techniques have been offered to improve the estimation of pseudo-labels.…”

Section: Related Work 21 Unsupervised Person Re-idmentioning

confidence: 99%

UnA-Mix: Rethinking Image Mixtures for Unsupervised Person Re-Identification

Liu,

Sun,

Liu

et al. 2024

Processes

View full text Add to dashboard Cite

show abstract

“…This is challenging since the source and target domain can have two extreme distributions, making the label space non-overlapped. Existing UDA re-ID methods can be roughly divided into three categories [6], namely mid-level feature alignment, domain style transfer, and clustering-based approach. In this paper, we focus on the last one, which can achieve superior performance in general.…”

Section: Introductionmentioning

confidence: 99%

Unsupervised Domain Adaptation Person Re-Identification by Camera-Aware Style Decoupling and Uncertainty Modeling

Guo

Liu

Shi

et al. 2022

2022 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

Unsupervised domain adaptation (UDA) person re-identification (re-ID) aims to transfer knowledge learned from labeled source domain to unlabeled target domain and has been successfully applied into a wide range of real-world scenarios. However, existing methods are mainly ineffective at handling domain shift as well as being sensitive to camera styles due to the unannotated target domain. In this paper, we propose a Camera-style Separation and Uncertainty Estimation (CSUE) model to address the problem from two perspectives. To alleviate the negative effect of cross-camera variation, we introduce the Camera-aware Style Decoupling module to impose inter-and-intra camera constraints on the feature extracting stage. It can better mine and describe the latent camera invariant features. Moreover, to avoid the inherent defect of clustering, an Uncertainty Modeling module is constructed via estimating the certainty, which helps progressively refine the pseudo labels. Extensive experiments on widely used datasets demonstrate the state-of-the-art performance of our model under the UDA re-ID setting.

show abstract

“…We adopted UDA because re-id model initialized with pre-trained weights from the published/source dataset tends to perform better than those using pre-trained weights from the ImageNet (for methods using fully unsupervised learning). This is illustrated by [51]. Our domain adaptation method utilizes an iterative process, in which pseudo labels are generated by density-based spatial clustering of applications with noise (DBSCAN) algorithm, and used as ground truth for the target domain training.…”

Section: Collaborative Learning Mutual Network For Target Domain Adap...mentioning

confidence: 99%

Person re-identification using visual analytic and deep learning

Tay¹

View full text Add to dashboard Cite

First and foremost, I would like to express my most sincere gratitude to my supervisor Prof. Yap Kim Hui, who not only accepted me as his Ph.D student, but also provides constant guidance during the time of my research in Nanyang Technological University. Prof. Yap is professional and knowledgeable, and he regularly challenges and stretches me to do better at each research work. This thesis is completed with his regular support and encouragement.My sincere appreciation also goes to Dr. Sharmili Roy for her advice, motivation, and encouragement. Her insightful views and precious suggestions not only broaden my horizons, but also helped me cross many hurdles during my research time. She is both a good mentor and a good friend that can share life experiences.Thank you to my thesis advisory committee (TAC) members Prof. Lin Wei Si and A/Prof. Chau Lap Pui for their great support and encouragement.

show abstract

Unsupervised Person Re-Identification: A Systematic Survey of Challenges and Solutions

Cited by 3 publications

References 142 publications

UnA-Mix: Rethinking Image Mixtures for Unsupervised Person Re-Identification

UnA-Mix: Rethinking Image Mixtures for Unsupervised Person Re-Identification

Unsupervised Domain Adaptation Person Re-Identification by Camera-Aware Style Decoupling and Uncertainty Modeling

Person re-identification using visual analytic and deep learning

Contact Info

Product

Resources

About