Camera style transformation with preserved self-similarity and domain-dissimilarity in unsupervised person re-identification

Zhu, Zhiqin; Luo, Yaqin; Chen, Sixin; Qi, Guanqiu; Mazur, Neal; Zhong, Chengyan; Li, Qiwang

doi:10.1016/j.jvcir.2021.103303

Cited by 19 publications

(8 citation statements)

References 57 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the future, we will explore the not only learning of part-aggregated features by mining the contextual part relations within each modality, but also more efficient attention network architectures, to solve RGB-IR Re-ID. The current methods for domain adaptation person re-identification [53][54][55][56] are mostly for the RGB-RGB datasets, and there is little research on RGB-IR datasets. Hence, we will pay attention to the RGB-IR domain adaptation and try to solve this problem.…”

Section: Discussionmentioning

confidence: 99%

MFCNet: Mining Features Context Network for RGB–IR Person Re-Identification

Mei

et al. 2021

Future Internet

View full text Add to dashboard Cite

RGB–IR cross modality person re-identification (RGB–IR Re-ID) is an important task for video surveillance in poorly illuminated or dark environments. In addition to the common challenge of Re-ID, the large cross-modality variations between RGB and IR images must be considered. The existing RGB–IR Re-ID methods use different network structures to learn the global shared features associated with multi-modalities. However, most global shared feature learning methods are sensitive to background clutter, and contextual feature relationships are not considered among the mined features. To solve these problems, this paper proposes a dual-path attention network architecture MFCNet. SGA (Spatial-Global Attention) module embedded in MFCNet includes spatial attention and global attention branches to mine discriminative features. First, the SGA module proposed in this paper focuses on the key parts of the input image to obtain robust features. Next, the module mines the contextual relationships among features to obtain discriminative features and improve network performance. Finally, extensive experiments demonstrate that the performance of the network architecture proposed in this paper is better than that of state-of-the-art methods under various settings. In the all-search mode of the SYSU and RegDB data sets, the rank-1 accuracy reaches 51.64% and 69.76%, respectively.

show abstract

Section: Discussionmentioning

confidence: 99%

MFCNet: Mining Features Context Network for RGB–IR Person Re-Identification

Mei

et al. 2021

Future Internet

View full text Add to dashboard Cite

show abstract

“…Looking into some typical deep networks, such as AlexNet [ 20 ], GoogLeNet [ 26 ], and ResNet [ 27 , 28 , 29 ], they are composed of repeating modules and structured by the branching and merging of various layers. Within each module, the pooling layer plays a vital role in achieving image transformation invariance, compact representation, yet effective expression in subsequent layers.…”

Section: Related Workmentioning

confidence: 99%

Discriminative Shape Feature Pooling in Deep Neural Networks

Dixit

2022

J. Imaging

Self Cite

View full text Add to dashboard Cite

Although deep learning approaches are able to generate generic image features from massive labeled data, discriminative handcrafted features still have advantages in providing explicit domain knowledge and reflecting intuitive visual understanding. Much of the existing research focuses on integrating both handcrafted features and deep networks to leverage the benefits. However, the issues of parameter quality have not been effectively solved in existing applications of handcrafted features in deep networks. In this research, we propose a method that enriches deep network features by utilizing the injected discriminative shape features (generic edge tokens and curve partitioning points) to adjust the network’s internal parameter update process. Thus, the modified neural networks are trained under the guidance of specific domain knowledge, and they are able to generate image representations that incorporate the benefits from both handcrafted and deep learned features. The comparative experiments were performed on several benchmark datasets. The experimental results confirmed our method works well on both large and small training datasets. Additionally, compared with existing models using either handcrafted features or deep network representations, our method not only improves the corresponding performance, but also reduces the computational costs.

show abstract

“…Some studies [2] [3] focus on unsupervised learning (USL) methods to address the problems of lack of labeled data, with learning features from unlabeled pedestrian images. Moreover, some methods [4][5] [6] focus on unsupervised domain adaptation (UDA) methods, which need both labeled source-domain data and unlabeled targetdomain data to train the model. It is costly and difficult to gain a large number of labeled pedestrian images, which limits the application of person Re-ID in practical scenarios.…”

Section: Introductionmentioning

confidence: 99%

Global-local transformer for unsupervised person re-identification

Xie,

Chen

2023

Fourth International Conference on Artificial Intelligence and Electromechanical Automation (AIEA 2023)

View full text Add to dashboard Cite

Unsupervised person re-identification (Re-ID) focuses on studying the issues of person Re-ID in the open world where there are no available annotated labels for pedestrian images. The original transformer only learns global features. It ignores local features and fails to consider high-level local information on images. Therefore, we introduce a transformer-based multi-branch unsupervised person Re-ID framework, named GLTReID, with learning more robust features of pedestrian images. Specifically, we split the original transformer into one global branch and two local branches, with learning global and local features in parallel. The global branch learns global features of pedestrian images, and two local branches learn local features of different parts of pedestrian images. Numerous experiments are conducted on several person Re-ID datasets, including CUHK03, Market-1501 and MSMT17. Numerous experiments denote that our method is better than most unsupervised learning methods and unsupervised domain adaptation methods in person Re-ID.

show abstract

Camera style transformation with preserved self-similarity and domain-dissimilarity in unsupervised person re-identification

Cited by 19 publications

References 57 publications

MFCNet: Mining Features Context Network for RGB–IR Person Re-Identification

MFCNet: Mining Features Context Network for RGB–IR Person Re-Identification

Discriminative Shape Feature Pooling in Deep Neural Networks

Global-local transformer for unsupervised person re-identification

Contact Info

Product

Resources

About