Filters








310 Hits in 5.9 sec

Style Normalization and Restitution for Generalizable Person Re-identification [article]

Xin Jin, Cuiling Lan, Wenjun Zeng, Zhibo Chen, Li Zhang
2020 arXiv   pre-print
To achieve this goal, we propose a simple yet effective Style Normalization and Restitution (SNR) module.  ...  Existing fully-supervised person re-identification (ReID) methods usually suffer from poor generalization capability caused by domain gaps.  ...  Camstyle: A novel data augmentation method for person re-identification. TIP, 2018. [75] Kaiyang Zhou, Yongxin Yang, Andrea Cavallaro, et al. Omni-scale feature learning for person re-identification.  ... 
arXiv:2005.11037v1 fatcat:6ord6xtutnem5gqak2nm4tm4fa

Style Normalization and Restitution for Generalizable Person Re-Identification

Xin Jin, Cuiling Lan, Wenjun Zeng, Zhibo Chen, Li Zhang
2020 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)  
To achieve this goal, we propose a simple yet effective Style Normalization and Restitution (SNR) module.  ...  Existing fully-supervised person re-identification (ReID) methods usually suffer from poor generalization capability caused by domain gaps.  ...  Acknowledgments This work was supported in part by NSFC under Grant U1908209, 61632001 and the National Key Research and Development Program of China 2018AAA0101400.  ... 
doi:10.1109/cvpr42600.2020.00321 dblp:conf/cvpr/JinLZ0Z20 fatcat:objraw6snnb6ngkh5wtatucj6i

Style Normalization and Restitution for Domain Generalization and Adaptation [article]

Xin Jin, Cuiling Lan, Wenjun Zeng, Zhibo Chen
2022 arXiv   pre-print
In this paper, we design a novel Style Normalization and Restitution module (SNR) to simultaneously ensure both high generalization and discrimination capability of the networks.  ...  An effective domain generalizable model is expected to be able to learn feature representations that are both generalizable and discriminative.  ...  This work is an extension of our conference paper [37] which is specifically designed for person re-identification.  ... 
arXiv:2101.00588v3 fatcat:q63qby7hijd67jznllgxsz7pwy

A Novel Mix-normalization Method for Generalizable Multi-source Person Re-identification [article]

Lei Qi, Lei Wang, Yinghuan Shi, Xin Geng
2022 arXiv   pre-print
Person re-identification (Re-ID) has achieved great success in the supervised scenario.  ...  In this paper, we aim to tackle the generalizable multi-source person Re-ID task (i.e., there are multiple available source domains, and the testing domain is unseen during training) from the data augmentation  ...  Generalizable Person Re-identification Currently, several methods have been developed to solve the generalizable (i.e., domain generalization) person Re-ID task, including the instance normalization based  ... 
arXiv:2201.09846v2 fatcat:nrvslpf7wzeitfvl3ymp24ukpm

Meta Batch-Instance Normalization for Generalizable Person Re-Identification [article]

Seokeon Choi, Taekyung Kim, Minki Jeong, Hyoungseob Park, Changick Kim
2020 arXiv   pre-print
Although supervised person re-identification (Re-ID) methods have shown impressive performance, they suffer from a poor generalization capability on unseen domains.  ...  In this paper, we propose a novel generalizable Re-ID framework, named Meta Batch-Instance Normalization (MetaBIN).  ...  Related Work Generalizable person re-identification: Domain generalizable person re-identification (DG Re-ID) aims to learn a robust model for obtaining good performance on an unseen target domain without  ... 
arXiv:2011.14670v1 fatcat:glt4gvide5dltepbq5bqcyuqfi

Adaptive Domain-Specific Normalization for Generalizable Person Re-Identification [article]

Jiawei Liu, Zhipeng Huang, Kecheng Zheng, Dong Liu, Xiaoyan Sun, Zheng-Jun Zha
2021 arXiv   pre-print
In this work, we propose a novel adaptive domain-specific normalization approach (AdsNorm) for generalizable person Re-ID.  ...  Although existing person re-identification (Re-ID) methods have shown impressive accuracy, most of them usually suffer from poor generalization on unseen target domain.  ...  [19] proposed a Style Normalization and Restitution module, which utilizes the IN layers to filter out style variations and compensates the identity-relevant features discarded by IN layers.  ... 
arXiv:2105.03042v2 fatcat:e4ukjdtdnrh7faizuljzjx3fri

Label Distribution Learning for Generalizable Multi-source Person Re-identification [article]

Lei Qi, Jiaying Shen, Jiaqi Liu, Yinghuan Shi, Xin Geng
2022 arXiv   pre-print
Person re-identification (Re-ID) is a critical technique in the video surveillance system, which has achieved significant success in the supervised setting.  ...  In this paper, we propose a novel label distribution learning (LDL) method to address the generalizable multi-source person Re-ID task (i.e., there are multiple available source domains, and the testing  ...  Generalizable Person Re-ID The goal of domain generalizable person re-identification is to learn a robust model in the source domain that can directly perform well in the target domain without additional  ... 
arXiv:2204.05903v1 fatcat:zzdl2trl7fbzhdggyrawzjumea

Towards Generalizable Person Re-identification with a Bi-stream Generative Model [article]

Xin Xu, Wei Liu, Zheng Wang, Ruiming Hu, Qi Tian
2022 arXiv   pre-print
Generalizable person re-identification (re-ID) has attracted growing attention due to its powerful adaptation capability in the unseen data domain.  ...  Extensive experiments demonstrate that our method outperforms the state-of-the-art methods on the large-scale generalizable re-ID benchmarks, involving domain generalization setting and cross-domain setting  ...  Then the Style Normalization and Restitution (SNR) [12] is also proposed to filter out the style shift by IN which is called Style Normalization (SN).  ... 
arXiv:2206.09362v2 fatcat:hgq7sahbsfh5lpvj67gb5dspty

Mimic Embedding via Adaptive Aggregation: Learning Generalizable Person Re-identification [article]

Boqiang Xu, Jian Liang, Lingxiao He, Zhenan Sun
2022 arXiv   pre-print
Domain generalizable (DG) person re-identification (ReID) aims to test across unseen domains without access to the target domain data at training time, which is a realistic but challenging problem.  ...  Meanwhile, META considers the relevance of an unseen target sample and source domains via normalization statistics and develops an aggregation module to adaptively integrate multiple experts for mimicking  ...  Related Work Domain Generalizable Person Re-identification. Person ReID has made great progress in recent years.  ... 
arXiv:2112.08684v3 fatcat:fzjkzrhji5dt7mohjr2nzals7i

TransMatcher: Deep Image Matching Through Transformers for Generalizable Person Re-identification [article]

Shengcai Liao, Ling Shao
2021 arXiv   pre-print
The proposed method, called TransMatcher, achieves state-of-the-art performance in generalizable person re-identification, with up to 6.1% and 5.7% performance gains in Rank-1 and mAP, respectively, on  ...  However, existing studies mostly use Transformers for feature representation learning, e.g. for image classification and dense predictions, and the generalizability of Transformers is unknown.  ...  Acknowledgements The authors would like to thank Yanan Wang who helped producing Fig. 1 in this paper, and Anna Hennig who helped proofreading the paper, and all the anonymous reviewers for the valuable  ... 
arXiv:2105.14432v2 fatcat:dyte26en5begle6l72nylur4za

Graph Sampling Based Deep Metric Learning for Generalizable Person Re-Identification [article]

Shengcai Liao, Ling Shao
2022 arXiv   pre-print
Together with an adapted competitive baseline, we improve the state of the art in generalizable person re-identification significantly, by 25.1% in Rank-1 on MSMT17 when trained on RandPerson.  ...  Recent studies show that, both explicit deep feature matching as well as large-scale and diverse training data can significantly improve the generalization of person re-identification.  ...  Acknowledgements Special thanks to Yanan Wang who helped creating Fig. 1 , and Anna Hennig who helped proofreading the paper.  ... 
arXiv:2104.01546v4 fatcat:3ssacfxynzgnnhkjweooklg544

Generalizable Person Re-identification via Self-Supervised Batch Norm Test-Time Adaption

Ke Han, Chenyang Si, Yan Huang, Liang Wang, Tieniu Tan
2022 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
In this paper, we investigate the generalization problem of person re-identification (re-id), whose major challenge is the distribution shift on an unseen domain.  ...  Specifically, BNTA quickly explores the domain-aware information within unlabeled target data before inference, and accordingly modulates the feature distribution normalized by BN to adapt to the target  ...  Related Work Cross-Domain Person Re-Identification.  ... 
doi:10.1609/aaai.v36i1.19963 fatcat:gyjosu3y5jbe5jdrxboghdh54m

Generalizable Person Re-Identification via Self-Supervised Batch Norm Test-Time Adaption [article]

Ke Han, Chenyang Si, Yan Huang, Liang Wang, Tieniu Tan
2022 arXiv   pre-print
In this paper, we investigate the generalization problem of person re-identification (re-id), whose major challenge is the distribution shift on an unseen domain.  ...  Specifically, BNTA quickly explores the domain-aware information within unlabeled target data before inference, and accordingly modulates the feature distribution normalized by BN to adapt to the target  ...  Related Work Cross-Domain Person Re-Identification.  ... 
arXiv:2203.00672v2 fatcat:af42p7g4n5d3fiu4ft4ebebsxu

Learn by Guessing: Multi-Step Pseudo-Label Refinement for Person Re-Identification [article]

Tiago de C. G. Pereira, Teofilo E. de Campos
2021 arXiv   pre-print
Unsupervised Domain Adaptation (UDA) methods for person Re-Identification (Re-ID) rely on target domain samples to model the marginal distribution of the data.  ...  Our refinement method includes a cluster selection strategy and a camera-based normalization method which reduces the within-domain variations caused by the use of multiple cameras in person Re-ID.  ...  Dr. de Campos received support from the National Council for Scientific and Technological Development -CNPq -through grant PQ 314154/2018-3. We have also received some support from FAPDF.  ... 
arXiv:2101.01215v1 fatcat:eot7sr2pxbcrlndo564wtrwl6y

Learning Domain Invariant Representations for Generalizable Person Re-Identification [article]

Yi-Fan Zhang, Zhang Zhang, Da Li, Zhen Jia, Liang Wang, Tieniu Tan
2022 arXiv   pre-print
According to the causal analysis, we propose a novel Domain Invariant Representation Learning for generalizable person Re-Identification (DIR-ReID) framework.  ...  Generalizable person Re-Identification (ReID) has attracted growing attention in recent computer vision community.  ...  Introduction Person Re-IDentification (ReID) aims at matching person images of the same identity across multiple camera views.  ... 
arXiv:2103.15890v3 fatcat:iaecmswnvzhdlcgzw6acnyr4hu
« Previous Showing results 1 — 15 out of 310 results