首页 » 文章 » 文章详细信息
Advances in Multimedia Volume 2018 ,2018-05-03
Impostor Resilient Multimodal Metric Learning for Person Reidentification
Research Article
Muhamamd Adnan Syed 1 Zhenjun Han 1 Zhaoju Li 1 Jianbin Jiao 1
Show affiliations
DOI:10.1155/2018/3202495
Received 2018-01-15, accepted for publication 2018-03-22, Published 2018-03-22
PDF
摘要

In person reidentification distance metric learning suffers a great challenge from impostor persons. Mostly, distance metrics are learned by maximizing the similarity between positive pair against impostors that lie on different transform modals. In addition, these impostors are obtained from Gallery view for query sample only, while the Gallery sample is totally ignored. In real world, a given pair of query and Gallery experience different changes in pose, viewpoint, and lighting. Thus, impostors only from Gallery view can not optimally maximize their similarity. Therefore, to resolve these issues we have proposed an impostor resilient multimodal metric (IRM3). IRM3 is learned for each modal transform in the image space and uses impostors from both Probe and Gallery views to effectively restrict large number of impostors. Learned IRM3 is then evaluated on three benchmark datasets, VIPeR, CUHK01, and CUHK03, and shows significant improvement in performance compared to many previous approaches.

授权许可

Copyright © 2018 Muhamamd Adnan Syed et al. 2018
This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

图表

Three Modals M1, M2, and M3 in Image Space. Query and Gallery lie in modal M1, while one impostor for query lies in modal M2 and the other in modal M3. Metric DM1 is learned using the impostor from modal M2, and DM2 is learned using the impostor from modal M3. Then, the obtained retrieval results of DM1 and DM2 are shown in Ranklist 1 and Ranklist 2, respectively. Correct Match is in green rectangle.

Impostor resilient multimodal metric learning (IRM3) for person reidentification.

Two queries are shown, Query 1 and Query 2, and their retrieval results using XQDA [11] and our IRM3. Correct match is shown in green rectangle, while blue rectangle shows impostors.

Performance at rank@1 when centers mk are selected randomly, and when centers are selected with our approach provided in Section 2.2.

通讯作者

Zhenjun Han.University of Chinese Academy of Sciences, Beijing, China, ucas.ac.cn.hanzhj@ucas.ac.cn

推荐引用方式

Muhamamd Adnan Syed,Zhenjun Han,Zhaoju Li,Jianbin Jiao. Impostor Resilient Multimodal Metric Learning for Person Reidentification. Advances in Multimedia ,Vol.2018(2018)

您觉得这篇文章对您有帮助吗?
分享和收藏
0

是否收藏?

参考文献
[1] M. A. Syed, J. Jiao. Multi-kernel metric learning for person re-identification. :784-788. DOI: 10.5244/C.23.23.
[2] M. Dikmen, E. Akbas, T. S. Huang, N. Ahuja. et al.(2011). Pedestrian recognition with a learned metric. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.6495(4):501-512. DOI: 10.5244/C.23.23.
[3] X. Zhu, X.-Y. Jing, F. Wu, W. Zheng. et al.Distance learning by treating negative samples differently and exploiting impostors with symmetric triplet constraint for person re-identification. . DOI: 10.5244/C.23.23.
[4] W.-S. Zheng, S. Gong, T. Xiang. Associating groups of people. . DOI: 10.5244/C.23.23.
[5] C. Jose, F. Fleuret. (2016). Scalable metric learning via weighted approximate rank component analysis. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.9909:875-890. DOI: 10.5244/C.23.23.
[6] R. Zhao, W. Ouyang, X. Wang. Unsupervised salience learning for person re-identification. :3586-3593. DOI: 10.5244/C.23.23.
[7] W. Li, X. Wang. Locally aligned feature transforms across views. :3594-3601. DOI: 10.5244/C.23.23.
[8] W. Liao, M. Y. Yang, N. Zhan, B. Rosenhahn. et al.Triplet-Based Deep Similarity Learning for Person Re-Identification. :385-393. DOI: 10.5244/C.23.23.
[9] F. Xiong, M. Gou, O. Camps, M. Sznaier. et al.(2014). Person re-identification using kernel-based metric learning methods. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.8695(7):1-16. DOI: 10.5244/C.23.23.
[10] Z. Mingyong, Z. Wu, C. Tian, Z. Lei. et al.Efficient person re-identification by hybrid spatiogram and covariance descriptor. :48-56. DOI: 10.5244/C.23.23.
[11] Y. Yang, J. Yang, J. Yan, S. Liao. et al.(2014). Salient color names for person re-identification. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.8689(1):536-551. DOI: 10.5244/C.23.23.
[12] W. Li, X. Zhu, S. Gong. Person Re-Identification by Deep Joint Learning of Multi-Loss Classification. :2194-2200. DOI: 10.5244/C.23.23.
[13] M. Hirzer, P. M. Roth, H. Bischof. Person re-identification by efficient impostor-based metric learning. :203-208. DOI: 10.5244/C.23.23.
[14] R. Zhao, W. Oyang, X. Wang. (2017). Person Re-Identification by Saliency Learning. IEEE Transactions on Pattern Analysis and Machine Intelligence.39(2):356-370. DOI: 10.5244/C.23.23.
[15] J. Roth, X. Liu. (2015). On the exploration of joint attribute learning for person re-identification. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.9003:673-688. DOI: 10.5244/C.23.23.
[16] S. Bąk, S. Zaidenberg, B. Boulay, F. Brémond. et al.Improving person re-identification by viewpoint cues. :175-180. DOI: 10.5244/C.23.23.
[17] S. Liao, Y. Hu, X. Zhu, S. Z. Li. et al.Person re-identification by Local Maximal Occurrence representation and metric learning. :2197-2206. DOI: 10.5244/C.23.23.
[18] R. Rama Varior, G. Wang, J. Lu, T. Liu. et al.(2016). Learning invariant color features for person reidentification. IEEE Transactions on Image Processing.25(7):3395-3410. DOI: 10.5244/C.23.23.
[19] F. Wang, W. Zuo, L. Lin, D. Zhang. et al.Joint learning of single-image and cross-image representations for person re-identification. :1288-1296. DOI: 10.5244/C.23.23.
[20] M. Gou, O. Camps, M. Sznaier. moM: Mean of Moments Feature for Person Re-identification. :1294-1303. DOI: 10.5244/C.23.23.
[21] C.-H. Kuo, S. Khamis, V. Shet. Person re-identification using semantic color names and RankBoost. :281-287. DOI: 10.5244/C.23.23.
[22] T. Matsukawa, T. Okabe, E. Suzuki, Y. Sato. et al.Hierarchical Gaussian descriptor for person re-identification. :1363-1372. DOI: 10.5244/C.23.23.
[23] D. Cheng, Y. Gong, S. Zhou, J. Wang. et al.Person re-identification by multi-channel parts-based CNN with improved triplet loss function. :1335-1344. DOI: 10.5244/C.23.23.
[24] D. Chen, Z. Yuan, B. Chen, N. Zheng. et al.Similarity learning with spatial constraints for person re-identification. :1268-1277. DOI: 10.5244/C.23.23.
[25] E. Kodirov, T. Xiang, Z. Fu, S. Gong. et al.(2016). Person re-identification by unsupervised 1 graph learning. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.9905:178-195. DOI: 10.5244/C.23.23.
[26] Y. Hu, S. Liao, Z. Lei, D. Yi. et al.Exploring structural information and fusing multiple features for person re-identification. :794-799. DOI: 10.5244/C.23.23.
[27] K. Q. Weinberger, L. K. Saul. (2009). Distance metric learning for large margin nearest neighbor classification. Journal of Machine Learning Research.10:207-244. DOI: 10.5244/C.23.23.
[28] A. Bhuiyan, A. Perina, V. Murino. (2015). Person re-identification by discriminatively selecting parts and features. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.8927:147-161. DOI: 10.5244/C.23.23.
[29] Y. Ying, P. Li. (2012). Distance metric learning with eigenvalue optimization. Journal of Machine Learning Research.13:1-26. DOI: 10.5244/C.23.23.
[30] S. Khamis, C.-H. Kuo, V. K. Singh, V. D. Shet. et al.(2015). Joint learning for attribute-consistent person re-identification. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.8927:134-146. DOI: 10.5244/C.23.23.
[31] M. Kostinger, M. Hirzer, P. Wohlhart, P. M. Roth. et al.Large scale metric learning from equivalence constraints. :2288-2295. DOI: 10.5244/C.23.23.
[32] M. Farenzena, L. Bazzani, A. Perina, V. Murino. et al.Person re-identification by symmetry-driven accumulation of local features. :2360-2367. DOI: 10.5244/C.23.23.
[33] H. Yu, A. Wu, W. Zheng. Cross-View Asymmetric Metric Learning for Unsupervised Person Re-Identification. :994-1002. DOI: 10.5244/C.23.23.
[34] T. Xiao, H. Li, W. Ouyang, X. Wang. et al.Learning deep feature representations with Domain Guided Dropout for person re-identification. :1249-1258. DOI: 10.5244/C.23.23.
[35] Y. Zhang, B. Li, H. Lu, A. Irie. et al.Sample-specific SVM learning for person re-identification. :1278-1287. DOI: 10.5244/C.23.23.
[36] D. Li, X. Chen, Z. Zhang, K. Huang. et al.Learning Deep Context-Aware Features over Body and Latent Parts for Person Re-identification. :7398-7407. DOI: 10.5244/C.23.23.
[37] L. Zhang, T. Xiang, S. Gong. Learning a discriminative null space for person re-identification. :1239-1248. DOI: 10.5244/C.23.23.
[38] R. R. Varior, B. Shuai, J. Lu, D. Xu. et al.(2016). A siamese long short-term memory architecture for human re-identification. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.9911:135-153. DOI: 10.5244/C.23.23.
[39] S. Bai, X. Bai, Q. Tian. Scalable Person Re-identification on Supervised Smoothed Manifold. :3356-3365. DOI: 10.5244/C.23.23.
[40] R. R. Varior, M. Haloi, G. Wang. (2016). Gated siamese convolutional neural network architecture for human re-identification. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.9912:791-808. DOI: 10.5244/C.23.23.
[41] H. Shi, Y. Yang, X. Zhu, S. Liao. et al.(2016). Embedding deep metric for person Re-identification: A study against large variations. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics): Preface.9905:732-748. DOI: 10.5244/C.23.23.
[42] G. Lisanti, I. Masi, A. Del Bimbo. Matching people across camera views using kernel canonical correlation analysis. . DOI: 10.5244/C.23.23.
[43] Y.-C. Chen, W.-S. Zheng, J.-H. Lai, P. C. Yuen. et al.(2017). An asymmetric distance model for cross-view feature mapping in person reidentification. IEEE Transactions on Circuits and Systems for Video Technology.27(8):1661-1675. DOI: 10.5244/C.23.23.
[44] D. Chen, Z. Yuan, G. Hua, N. Zheng. et al.Similarity learning on an explicit polynomial kernel feature map for person re-identification. :1565-1573. DOI: 10.5244/C.23.23.
[45] D. R. Hardoon, S. Szedmak, J. Shawe-Taylor. (2004). Canonical correlation analysis: an overview with application to learning methods. Neural Computation.16(12):2639-2664. DOI: 10.5244/C.23.23.
[46] W. Li, R. Zhao, T. Xiao, X. Wang. et al.DeepReID: Deep filter pairing neural network for person re-identification. :152-159. DOI: 10.5244/C.23.23.
[47] L. Zhao, X. Li, Y. Zhuang, J. Wang. et al.Deeply-Learned Part-Aligned Representations for Person Re-identification. :3239-3248. DOI: 10.5244/C.23.23.
[48] W. Chen, X. Chen, J. Zhang, K. Huang. et al.Beyond Triplet Loss: A Deep Quadruplet Network for Person Re-identification. :1320-1329. DOI: 10.5244/C.23.23.
[49] X. Qian, Y. Fu, Y. Jiang, T. Xiang. et al.Multi-scale Deep Learning Architectures for Person Re-identification. :5409-5418. DOI: 10.5244/C.23.23.
[50] H. Liu, J. Feng, M. Qi, J. Jiang. et al.(2017). End-to-end comparative attention networks for person re-identification. IEEE Transactions on Image Processing.26(7):3492-3506. DOI: 10.5244/C.23.23.
文献评价指标
浏览 128次
下载全文 11次
评分次数 0次
用户评分 0.0分
分享 0次