基于孪生网络和重排序的行人重识别

doi:10.11772/j.issn.1001-9081.2018041223

计算机应用 ›› 2018, Vol. 38 ›› Issue (11): 3161-3166.DOI: 10.11772/j.issn.1001-9081.2018041223

• 第七届中国数据挖掘会议(CCDM 2018) • 上一篇下一篇

基于孪生网络和重排序的行人重识别

陈首兵¹, 王洪元¹, 金翠¹, 张玮²

1. 常州大学信息科学与工程学院, 江苏常州 213164;
2. 常州信息职业技术学院电子与电气工程学院, 江苏常州 213164

收稿日期:2018-03-23 修回日期:2018-06-03 出版日期:2018-11-10 发布日期:2018-11-10
通讯作者: 王洪元
作者简介:陈首兵(1991-),男,江苏盐城人,硕士研究生,主要研究方向:计算机视觉;王洪元(1960-),男,江苏常熟人,教授,博士,CCF会员,主要研究方向:计算机视觉;金翠(1994-),女,江苏常州人,硕士研究生,主要研究方向:计算机视觉;张玮(1980-),女,山东淄博人,讲师,硕士,主要研究方向:机器视觉。
基金资助:
国家自然科学基金资助项目（61572085）。

Person re-identification based on siamese network and reranking

CHEN Shoubing¹, WANG Hongyuan¹, JIN Cui¹, ZHANG Wei²

1. College of Information Science and Engineering, Changzhou University, Changzhou Jiangsu 213164, China;
2. Institute of Electronic and Electrical, Changzhou College of Information Technology, Changzhou Jiangsu 213164, China

Received:2018-03-23 Revised:2018-06-03 Online:2018-11-10 Published:2018-11-10
Supported by:
This work is partially supported by the National Natural Science Foundation of China (61572085).

摘要/Abstract

摘要： 针对非重叠多摄像头下的行人重识别（Re-ID）易受到光照、姿势及遮挡等影响和实验过程中存在图像错误匹配的情况，提出一种基于孪生网络和重排序的行人重识别方法。首先，给定一对行人训练图像，孪生网络可以同时学习一个具有辨别力的卷积神经网络（CNN）特征和相似性度量，并预测两个输入图像的行人身份以及判断它们是否属于同一个行人；然后，通过k互近邻方法来降低图像错误匹配的情况；最后，将欧氏距离和杰卡德距离加权来对排序表进行重排序。在数据集Market1501和CUHK03上进行多次实验，实验结果显示在Market1501上Single Query情况下在图库中第一次就成功匹配的概率（Rank1）达到83.44%，平均精度均值（mAP）为68.75%，在CUHK03上single-shot情况下Rank1达到85.56%，mAP为88.32%，明显高于传统的基于特征表示和度量学习的方法。

关键词: 行人重识别, 孪生网络, k互近邻, 杰卡德距离, 重排序

Abstract: Person Re-Identification (Re-ID) under non-overlapping multi-camera is easily affected by illumination, posture, and occlusion, and there are image mismatches in the experimental process. A Re-ID method based on siamese network and reranking was proposed. Firstly, a pair of pedestrian training images were given, a discriminative Convolutional Neural Network (CNN) feature and similarity measure could be simultaneously learned by the siamese network to predict the pedestrian identity of the two input images and determine whether they belonged to the same pedestrian. Then, the k-reciprocal neighbor method was used to reduce the image mismatches. Finally, Euclidean distance and Jaccard distance were weighted to rerank the sorted list. Several experiments were performed on the datasets Market1501 and CUHK03. The experimental results show that the Rank1 (the probability of matching successfully for the first time) reaches 83.44% and mAP (mean Average Precision) is 68.75% under Single Query on Market1501. In the case of single-shot on CUHK03, the Rank1 reaches 85.56% and mAP is 88.32%, which are significantly higher than those of the traditional methods based on feature representation and metric learning.

Key words: Person Re-Identification (Re-ID), siamese network, k-reciprocal nearest neighbors, Jaccard distance, reranking

中图分类号:

TP391.41

陈首兵, 王洪元, 金翠, 张玮. 基于孪生网络和重排序的行人重识别[J]. 计算机应用, 2018, 38(11): 3161-3166.

CHEN Shoubing, WANG Hongyuan, JIN Cui, ZHANG Wei. Person re-identification based on siamese network and reranking[J]. Journal of Computer Applications, 2018, 38(11): 3161-3166.

参考文献

[1] ZHENG L, YANG Y, HAUPTMANN A G. Person re-identification:past, present and future[EB/OL].[2018-03-20]. http://opus.lib.uts.edu.au/bitstream/10453/118051/1/1610.02984.pdf.
[2] LI W, ZHAO R, XIAO T, et al. DeepReID:deep filter pairing neural network for person re-identification[C]//CVPR 2014:Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2014:152-159.
[3] ZHENG L, SHEN L, TIAN L, et al. Scalable person re-identification:a benchmark[C]//ICCV 2015:Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway, NJ:IEEE, 2015:1116-1124.
[4] JEGOU H, HARZALLAH H, SCHMID C. A contextual dissimilarity measure for accurate and efficient image search[C]//Proceedings of the 2007 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2011:1-8.
[5] QIN D, GAMMETER S, BOSSARD L, et al. Hello neighbor:accurate object retrieval with k-reciprocal nearest neighbors.[C]//Proceedings of the 2011 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2011:777-784.
[6] BROMLEY J, GUYON I, LeCUN Y, et al. Signature verification using a "Siamese" time delay neural network[C]//Proceedings of the 6th International Conference on Neural Information Processing Systems. San Francisco, CA:Morgan Kaufmann Publishers, 1993:737-744.
[7] YI D, LEI Z, LIAO S, et al. Deep metric learning for person re-identification[C]//Proceedings of the 201422nd International Conference on Pattern Recognition. Washington, DC:IEEE Computer Society, 2014:34-39.
[8] 张华. 基于空间颜色特征的行人重识别方法[J]. 华中科技大学学报(自然科学版), 2013, 41(S2):209-212.(ZHANG H. Spatial color feature based pedestrian re-identification[J]. Journal of Huazhong University of Science and Technology(Natural Science Edition), 2013, 41(S2):209-212.)
[9] AHMED E, JONES M, MARKS T K. An improved deep learning architecture for person re-identification[C]//CVPR 2015:2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2015:3908-3916.
[10] WU L, SHEN C, HENGEL A V D. PersonNet:person re-identification with deep convolutional neural networks[J/OL]. arXiv Preprint, 2017, 2017:arXiv:1601.07255(2017-01-27)[2017-06-20]. http://cn.arxiv.org/abs/1601.07255.
[11] VARIOR R R, HALOI M, WANG G. Gated siamese convolutional neural network architecture for human re-identification[C]//Proceedings of the 2016 European Conference on Computer Vision. Berlin:Springer, 2016:791-808.
[12] GARCIA J, MARTINEL N, MICHELONI C, et al. Person re-identification ranking optimisation by discriminant context information analysis[C]//Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway, NJ:IEEE, 2015:1305-1313.
[13] YE M, CHEN J, LENG Q, et al. Coupled-view based ranking optimization for person re-identification[C]//Proceedings of the 2015 International Conference on Multimedia Modeling. Berlin:Springer, 2015:105-117.
[14] YE M, LIANG C, YU Y, et al. Person reidentification via ranking aggregation of similarity pulling and dissimilarity pushing[J]. IEEE Transactions on Multimedia, 2016, 18(12):2553-2566.
[15] XIAO T, LI H, OUYANG W, et al. Learning deep feature representations with domain guided dropout for person re-identification[C]//Proceedings of the 2016 IEEE Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2016:1249-1258.
[16] CHUM O, PHILBIN J, SIVIC J, et al. Total recall:automatic query expansion with a generative feature model for object retrieval[C]//Proceedings of the 2007 IEEE International Conference on Computer Vision. Piscataway, NJ:IEEE, 2007:1-8.
[17] HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2016:770-778.
[18] ZHENG L, BIE Z, SUN Y, et al. Mars:a video benchmark for large-scale person re-identification[C]//Proceedings of the 2016 European Conference on Computer Vision. Berlin:Springer, 2016:868-884.
[19] SRIVASTAVA N, HINTON G, KRIZHEVSKY A, et al. Dropout:a simple way to prevent neural networks from overfitting[J]. Journal of Machine Learning Research, 2014, 15(1):1929-1958.
[20] FELZENSZWALB P F, GIRSHICK R B, MCALLESTER D, et al. Object detection with discriminatively trained part-based models[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2014, 47(2):6-7.
[21] SU C, ZHANG S, XING J, et al. Deep attributes driven multi-camera person re-identification[C]//Proceedings of the 2016 European Conference on Computer Vision. Berlin:Springer, 2016:475-491.
[22] USTINOVA E, GANIN Y, LEMPITSKY V. Multi-region bilinear convolutional neural networks for person re-identification[C]//Proceedings of the 2017 Advanced Video and Signal Based Surveillance. Washington, DC:IEEE Computer Society, 2017:1-6.
[23] ZHANG L, XIANG T, GONG S. Learning a discriminative null space for person re-identification[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2016:1239-1248.
[24] ZHENG L, HUANG Y, LU H, et al. Pose invariant embedding for deep person re-identification[J/OL]. arXiv Preprint, 2017, 2017:arXiv:1701.07732[2017-01-26]. http://cn.arxiv.org/abs/1701.07732.
[25] BARBOSA I B, CRISTANI M, CAPUTO B, et al. Looking beyond appearances:synthetic training data for deep CNNs in re-identification[J]. Computer Vision and Image Understanding, 2017, 167:50-62.
[26] ZHENG L, ZHANG H, SUN S, et al. Person re-identification in the wild[C]//Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Washington, DC:IEEE Computer Society, 2017:3346-3355.
[27] LIAO S, HU Y, ZHU X, et al. Person re-identification by local maximal occurrence representation and metric learning[C]//Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ:IEEE, 2015:2197-2206.
[28] VARIOR R R, SHUAI B, LU J, et al. A siamese long short-term memory architecture for human re-identification[C]//Proceedings of the 2016 European Conference on Computer Vision. Berlin:Springer, 2016:135-153.

基于孪生网络和重排序的行人重识别

Person re-identification based on siamese network and reranking

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

[1]	姬张建, 任兴旺. 带旋转与尺度估计的全卷积孪生网络目标跟踪算法[J]. 计算机应用, 2021, 41(9): 2705-2711.
[2]	钟莎, 黄玉清. 基于孪生区域候选网络的无人机指定目标跟踪[J]. 计算机应用, 2021, 41(2): 523-529.
[3]	刘紫燕, 朱明成, 袁磊, 马珊珊, 陈霖周廷. 基于非局部关注和多重特征融合的视频行人重识别[J]. 计算机应用, 2021, 41(2): 530-536.
[4]	楼豪杰, 郑元林, 廖开阳, 雷浩, 李佳. 基于Siamese-YOLOv4的印刷品缺陷目标检测[J]. 《计算机应用》唯一官方网站, 2021, 41(11): 3206-3212.
[5]	韩建栋, 李晓宇. 基于多尺度特征融合的行人重识别方法[J]. 计算机应用, 2021, 41(10): 2991-2996.
[6]	李自强, 王正勇, 陈洪刚, 李林怡, 何小海. 基于外观和动作特征双预测模型的视频异常行为检测[J]. 计算机应用, 2021, 41(10): 2997-3003.
[7]	陈莉, 王洪元, 张云鹏, 曹亮, 殷雨昌. 联合均等采样随机擦除和全局时间特征池化的视频行人重识别方法[J]. 计算机应用, 2021, 41(1): 164-169.
[8]	邱耀儒, 孙为军, 黄永慧, 唐瑜祺, 张浩川, 吴俊鹏. 基于生成对抗网络联合时空模型的行人重识别方法[J]. 计算机应用, 2020, 40(9): 2493-2498.
[9]	熊昌镇, 李言. 动态的加权孪生网络跟踪算法[J]. 计算机应用, 2020, 40(8): 2214-2218.
[10]	杨锋, 许玉, 尹梦晓, 符嘉成, 黄冰, 梁芳烜. 基于深度学习的行人重识别综述[J]. 计算机应用, 2020, 40(5): 1243-1252.
[11]	刘紫燕, 万培佩. 基于注意力机制的行人重识别特征提取方法[J]. 计算机应用, 2020, 40(3): 672-676.
[12]	邓轩, 廖开阳, 郑元林, 袁晖, 雷浩, 陈兵. 基于深度多视图特征距离学习的行人重识别[J]. 计算机应用, 2019, 39(8): 2223-2229.
[13]	杨康, 宋慧慧, 张开华. 基于双重注意力孪生网络的实时视觉跟踪[J]. 计算机应用, 2019, 39(6): 1652-1656.
[14]	祁子梁, 曲寒冰, 赵传虎, 董良, 李博昭, 王长生. 基于孪生网络和双向最大边界排序损失的行人再识别[J]. 计算机应用, 2019, 39(4): 977-983.
[15]	邓杨, 谢宁, 杨阳. 基于孪生检测网络的实时视频追踪算法[J]. 计算机应用, 2019, 39(12): 3440-3444.