Unsupervised Cross-Dataset Transfer Learning for Person Re-identification

By Peixi Peng, Tao A. Xiang, Yaowei Wang, Massimiliano Pontil, Shaogang Gong, Tiejun Huang, Yonghong Tian. Computer Vision and Pattern Recognition, 2016

388 citations recorded by the source corpus.

Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/4d6e98fb5fcb7b5983f615a45ac1d81d1b570ca0

How researchers see this paper

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

In Biomedical Engineering

1. Yi Yang
2. Liang Zheng
3. Shaogang Gong
4. Qi Tian
5. Mang Ye
6. Shiliang Zhang
7. Xiatian Zhu
8. Pong Chi Yuen

In Computer Vision and Pattern Recognition

1. Yi Yang
2. Liang Zheng
3. Dacheng Tao
4. Luc Van Gool
5. Xiaogang Wang
6. Qi Tian
7. Ming-Hsuan Yang
8. Ling Shao

Citations in the corpus

• Transferable Joint Attribute-Identity Deep Learning for Unsupervised Person Re-identification (2018, methodology): “(3) Source identity and attribute knowledge based transfer methods: kLFDA N [52] SADA+kLFDA [52] AdaRSVM [33] UDML [36].”
• Deep Representation Learning With Part Loss for Person Re-Identification (2017, background): “Existing approaches conquer this challenge by either seeking discriminative metrics [1], [2], [3], [4], [5], [6], [7], [8], [2], [9], [10], [11], or generating discriminative features [12], [13], [14], [15], [16], [17], [18], [19].”
• Image-Image Domain Adaptation with Preserved Self-Similarity and Domain-Dissimilarity for Person Re-identification (2017, methodology): “We also compare existing unsupervised methods, including the Clustering-based Asymmetric MEtric Learning (CAMEL) [49], the Progressive Unsupervised Learning (PUL) [6], and UMDL [35].”
• Deeply-Learned Part-Aligned Representations for Person Re-identification (2017, background): “obabilistic patch matching for handling pose misalignment [4,3,36], similarity learning for dealing with probe and gallery images with different resolutions [24,17], connection with transfer learning [34,38], reranking inspired by the connection with image search [65,13], partial person matching [66”
• Part-Aligned Bilinear Representations for Person Re-identification (2018, background): “Recently, attributes [33–35], transfer learning [36, 37], re-ranking [38, 39], partial person matching [40], and human-in-the-loop learning [41,42], have also been studied.”
• Deep Attributes Driven Multi-Camera Person Re-identification (2016, background): “represent the visual appearance of a person [1,2,3,4,5,6,7] and b) learning a discriminative distance metric hence the distance of features from the same person can be smaller [8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25].”
• Discovering Biased News Articles Leveraging Multiple Human Annotations (2020, background): “Other similar works that miss correct labels include tracking and matching individuals in images with transfer learning (Peng et al., 2016).”
• Person Re-identification: Past, Present and Future (2016, background): “In [162], [163], unsupervised transfer learning is proposed in which the target dataset is unlabeled.”
• Distilled Person Re-Identification: Towards a More Scalable System (2019, methodology): “We compared with unsupervised ReID methods including unsupervised features LOMO [31], BOW [66] and unsupervised learning models UMDL [42], PTGAN [53], PUL [14], CAMEL [62], SPGAN [13], TJAIDL [51] and HHL [71].”
• Group-aware Label Transfer for Domain Adaptive Person Re-identification (2021, neutral): “Methods DukeMTMC→Market1501 Market1501→DukeMTMC mAP top-1 top-5 top-10 mAP top-1 top-5 top-10 UMDL [32] (CVPR’16) 12.”
• Invariance Matters: Exemplar Memory for Domain Adaptive Person Re-Identification (2019, methodology): “We compare with two hand-crafted feature based methods without transfer learning: LOMO [15] and BOW [37], three unsupervised methods that use a labeled source data to initialize the model but ignore the labeled source data during learning feature for target domain: CAMEL [36], UMDL [20] and PUL [9],”
• Adaptive Transfer Network for Cross-Domain Person Re-Identification (2019, background): “This results in intensive disparities across pedestrian image galleries known as the challenge of domain gap in literatures [23, 16], hindering the application of exiting person re-identification systems.”
• Security and Privacy for 6G: A Survey on Prospective Technologies and Challenges (2021, background): “, person re-identification [223]) by exploiting the power of trained models that were often carried out on high-performance computing platforms and related large-scale datasets.”
• Self-Similarity Grouping: A Simple Unsupervised Cross Domain Adaptation Approach for Person Re-Identification (2018, methodology): “ Moreover, the performance under five shots setting is extremely close to fully supervised baseline method. Specif7 Methods mAP R1 R5 R10 LOMO [24] 8.0 27.2 41.6 49.1 Bow [44] 14.8 35.8 52.4 60.3 UMDL [29] 12.4 34.5 52.6 59.6 PTGAN [40] - 38.6 - 66.1 PUL [10] 20.5 45.5 60.7 66.7 SPGAN [8] 22.8 51.5 7”
• Cross-View Asymmetric Metric Learning for Unsupervised Person Re-Identification (2017, methodology): “Model SDALF CPS UDML GL GTS SDC CAMEL [8] [7] [24] [12] [29] [35]”
• Unsupervised Domain Adaptive Re-Identification: Theory and Practice (2018, methodology): “[23] adopt a multi-task dictionary learning scheme to learn a view-invariant representation.”
• A Bottom-Up Clustering Approach to Unsupervised Person Re-Identification (2019, methodology): “In recent years, some transfer learning methods (Hehe et al. 2018; Peng et al. 2016; Deng et al. 2018) are proposed upon the success of deep learning (Dong et al. 2018a; Li, Tang, and Mei 2018).”