Person Transfer GAN to Bridge Domain Gap for Person Re-identification

By Longhui Wei, Shiliang Zhang, Wen Gao, Qi Tian. 2017

1,975 citations recorded by the source corpus.

Although the performance of person Re-Identification (ReID) has been significantly boosted, many challenging issues in real scenarios have not been fully investigated, e.g., the complex scenes and lighting variations, viewpoint and pose changes, and the large number of identities in a camera network. To facilitate the research towards conquering those issues, this paper contributes a new dataset called MSMT171 with many important features, e.g., 1) the raw videos are taken by an 15-camera network deployed in both indoor and outdoor scenes, 2) the videos cover a long period of time and present complex lighting variations, and 3) it contains currently the largest number of annotated identities, i.e., 4,101 identities and 126,441 bounding boxes. We also observe that, domain gap commonly exists between datasets, which essentially causes severe performance drop when training and testing on different datasets. This results in that available training data cannot be effectively leveraged for new testing domains. To relieve the expensive costs of annotating new training samples, we propose a Person Transfer Generative Adversarial Network (PTGAN) to bridge the domain gap. Comprehensive experiments show that the domain gap could be substantially narrowed-down by the PTGAN.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Continual Representation Learning for Biometric Identification (2020, methodology): “Biometric identification [23,51], including face recognition [11, 37, 67] and person re-identification (re-id) [34, 66, 71], has achieved significant progress in the recent years due to the advances in modern learnable representations [8, 9, 11, 20, 34, 52, 60, 67] and emerging large datasets [17, 2”
Adaptive Transfer Network for Cross-Domain Person Re-Identification (2019, methodology): “Recently, a few cycle generative adversarial Networks (CycleGAN) [48, 1, 7] based UDA approaches [2, 45, 46, 33, 3] are proposed for person re-identiﬁcation, which focus on learning a generator network that transforms samples in the pixel space from one domain to another.”
Cluster Contrast for Unsupervised Person Re-Identification (2021, methodology): “Datasets We evaluate our proposed method on four largescale person re-identification (re-ID) benchmarks: Market1501 [53], DukeMTMC-reID [33], MSMT17 [44], PersonX [37], and one vehicle ReID dataset, VeRi-776 [27].”
Omni-Scale Feature Learning for Person Re-Identification (2019, methodology): “The ﬁrst four are considered as ‘big’ datasets even though their sizes (around 30K training images for the largest MSMT17) are fairly moderate; while VIPeR and GRID are generally too small to train without using those big datasets for pre-training.”
Generalising without Forgetting for Lifelong Person Re-Identification (2021, background): “In the past decade, the development of deep learning and the emergence of large-scale datasets have signiﬁcantly advanced person Re-ID (Song et al. 2019; Wei et al. 2018; Wu, Zhu, and Gong 2019).”
Horizontal Pyramid Matching for Person Re-identification (2018, background): “In the ﬁrst type, prior knowledge like poses or body landmarks are estimated and extracted to localize the discriminative regions (Su et al. 2017; Zheng et al. 2017; Wei et al. 2017).”
Single-Shot Freestyle Dance Reenactment (2020, background): “In still images, the problem of pose transfer is well studied [28, 37, 1, 45, 9, 57, 11, 10, 39, 30, 26], out of which [9, 39, 30] use a human parser, as we do.”
CLIP-ReID: Exploiting Vision-Language Model for Image Re-Identification without Concrete Text Labels (2022, methodology): “We evaluate our method on four person re-identification datasets, including MSMT17 (Wei et al. 2018), Market-1501 (Zheng et al.”
A Survey of Unsupervised Deep Domain Adaptation (2018, methodology): “ource situations (where there are very few target data points) [87], in situations with different label sets for each domain [195], for object detection [33, 84, 92], and for person re-identification [9, 46, 61, 218]. 7.2 Natural Language Processing Domain adaptation has been used in natural languag”
Mutual Mean-Teaching: Pseudo Label Refinery for Unsupervised Domain Adaptation on Person Re-identification (2020, methodology): “We utilize k-means clustering algorithm and the number M t of pseudo classes is set as 500, 700, 900 for Market-1501 and DukeMTMC-reID, and 500, 1000, 1500, 2000 (Zheng et al., 2015), DukeMTMC-reID (Ristani et al., 2016), and MSMT17 (Wei et al., 2018) datasets, where MMT-M t represents the result wi”
Applications of Generative Adversarial Networks (GANs): An Updated Review (2019, background): “Wei et al. [120] contributed a new dataset called MSMT17 with many important features, e.g., (1) the raw videos are taken by an 15-camera network deployed in both indoor and outdoor scenes, (2) the videos cover a long period of time and present complex lighting variations, and (3) it contains…”
FD-GAN: Pose-guided Feature Distilling GAN for Robust Person Re-identification (2018, methodology): “PTGAN [4] was proposed to bridge the domain gap between separate datasets.”
VERI-Wild: A Large Dataset and a New Method for Vehicle Re-Identification in the Wild (2019, methodology): “[22] proposed a PTGAN to transfer person images between different styles for reducing domain gap.”
Hi-CMD: Hierarchical Cross-Modality Disentanglement for Visible-Infrared Person Re-Identification (2019, background): “Most of the existing work is categorized in two ways as follows: pose transfer [18, 22, 6] and style transfer [32, 19, 5, 31].”
Semantics-Enhanced Adversarial Nets for Text-to-Image Synthesis (2019, background): “In several other computer vision tasks, such as unsupersived domain person Re-identification [3, 39, 46] and image-to-image style transfer (CycleGAN) [15], such guidance from groundtruth images has been demonstrated effective in enhancing structural coherence of synthesized images.”