融合空频信息的多粒度师生网络无监督行人重识别

doi:10.19678/j.issn.1000-3428.0070010

计算机工程 ›› 2026, Vol. 52 ›› Issue (1): 217-227. doi: 10.19678/j.issn.1000-3428.0070010

• 计算机视觉与图形图像处理 • 上一篇下一篇

融合空频信息的多粒度师生网络无监督行人重识别

陈玉敏¹^,², 车进¹^,²^,*(), 杨莹莹¹^,²

1. 宁夏大学电子与电气工程学院, 宁夏银川 750021
2. 宁夏大学宁夏沙漠信息智能感知重点实验室, 宁夏银川 750021

收稿日期:2024-06-16 修回日期:2024-08-06 出版日期:2026-01-15 发布日期:2024-09-23
通讯作者: 车进
作者简介:
陈玉敏, 女, 硕士研究生, 主研方向为行人重识别
车进(通信作者), 教授、博士
杨莹莹, 硕士研究生
基金资助:
国家自然科学基金(62366042)

Unsupervised Person Re-Identification for Multigrained Teacher-Student Networks Incorporating Spatial Frequency Information

CHEN Yumin¹^,², CHE Jin¹^,²^,*(), YANG Yingying¹^,²

1. School of Electronic and Electrical Engineering, Ningxia University, Yinchuan 750021, Ningxia, China
2. Ningxia Key Laboratory of Intelligent Sensing for Desert Information, Ningxia University, Yinchuan 750021, Ningxia, China

Received:2024-06-16 Revised:2024-08-06 Online:2026-01-15 Published:2024-09-23
Contact: CHE Jin

摘要/Abstract

摘要：

无监督行人重识别旨在挖掘无标注数据的判别性表示用于行人检索任务。基于伪标签进行训练的无监督行人重识别方法目前已经取得了瞩目的进展。然而在训练过程中引入的噪声和信息利用不完全问题限制了该任务的进一步发展。提出一种融合浅层空频信息的多粒度师生网络。首先, 同时考虑全局和局部特征并将其集成到聚类对比学习中, 丰富特征表示, 利用训练好的教师模型指导学生模型快速收敛, 减少噪声伪标签的干扰; 其次, 设计一个新颖的空频信息交互模块, 利用网络加深过程中丢失的浅层空间域、频域有用信息; 此外, 在学生网络的训练过程中采用一种重利用策略, 将以往方法中直接丢弃的部分未聚类实例作为难样本重新利用。在Market1501、DukeMTMC-reID和MSMT17 3个大型数据集上的均值平均精度(mAP)结果分别达到87.5%、74.8%和41.9%, 证明了该方法的优越性。

关键词: 无监督行人重识别, 伪标签噪声, 多粒度特征, 师生网络, 空频信息

Abstract:

Unsupervised person Re-Identification (Re-ID) aims to mine discriminative representations from unlabeled data for person retrieval. Currently, unsupervised person Re-ID methods based on pseudo-labels have achieved remarkable progress. However, the noise introduced during the training process and incomplete utilization of information limit its further development. This paper proposes a multigrained teacher-student network that integrates shallow spatial and frequency information. First, it simultaneously considers global and local features and integrates them into clustering-based contrastive learning, enriching feature representation. A well-trained teacher model is used to guide the student model to converge quickly, thereby reducing the interference of noisy pseudo-labels. Second, a novel spatial frequency interaction module that utilizes useful information in the shallow spatial and frequency domains that is lost during the network deepening process is proposed. Additionally, a recycling strategy is adopted in the training process of the student network, in which some unclustered instances that are directly discarded in the previous methods are recycled as hard samples. The mean Average Precision (mAP) results for three large datasets, Market1501, DukeMTMC-reID, and MSMT17, reach 87.5%, 74.8%, and 41.9%, respectively, proving the superiority of the proposed method.

Key words: unsupervised person Re-Identification (Re-ID), pseudo-label noise, multi-grained features, teacher-student network, spatial frequency information

陈玉敏, 车进, 杨莹莹. 融合空频信息的多粒度师生网络无监督行人重识别[J]. 计算机工程, 2026, 52(1): 217-227.

CHEN Yumin, CHE Jin, YANG Yingying. Unsupervised Person Re-Identification for Multigrained Teacher-Student Networks Incorporating Spatial Frequency Information[J]. Computer Engineering, 2026, 52(1): 217-227.

https://www.ecice06.com/CN/Y2026/V52/I1/217

图/表 15

图1 基于空频信息交互的多粒度特征师生网络

Fig.1 Multigrained feature teacher-student network based on spatial frequency information interaction

图2 ResNet50双分支网络结构

Fig.2 ResNet50 dual-branch network structure

图3 空频信息交互模块

Fig.3 Spatial frequency information interaction module

图4 行人局部相似现象

Fig.4 Person local similarity phenomenon

图5 离群值重利用策略

Fig.5 Outlier recycling strategy

图6 eps参数分析

Fig.6 Analysis of eps parameters

图7 SFIM嵌入方法

Fig.7 SFIM embedding method

图8 Top-5检索结果可视化

Fig.8 Visualisation of Top-5 retrieval results

参考文献 35

1	YE M , SHEN J B , LIN G J , et al. Deep learning for person re-identification: a survey and outlook. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 44 (6): 2872- 2893.
2	郝阿香, 贾郭军. 结合注意力与批特征擦除的行人重识别模型. 计算机工程, 2022, 48 (7): 270-276, 306. doi: 10.19678/j.issn.1000-3428.0062020
	HAO A X , JIA G J . Person re-identification model combining attention and batch feature erasure. Computer Engineering, 2022, 48 (7): 270-276, 306. doi: 10.19678/j.issn.1000-3428.0062020
3	朱锦雷, 李艳凤, 陈后金, 等. 近邻优化跨域无监督行人重识别算法. 中国图象图形学报, 2023, 28 (11): 3471- 3484.
	ZHU J L , LI Y F , CHEN H J , et al. Cross-domain unsupervised Re-ID algorithm based on neighbor adversarial and consistency loss. Journal of Image and Graphics, 2023, 28 (11): 3471- 3484.
4	贾洁茹, 张硕蕊, 钱宇华, 等. 基于伪标签正则化损失的无监督行人重识别. 电子学报, 2024, 52 (5): 1743- 1758.
	JIA J R , ZHANG S R , QIAN Y H , et al. Unsupervised person re-identification with pseudo label regularization loss. Acta Electronica Sinica, 2024, 52 (5): 1743- 1758.
5	LAN L , TENG X , ZHANG J , et al. Learning to purification for unsupervised person re-identification. IEEE Transactions Image Processing, 2023, 32, 3338- 3353. doi: 10.1109/TIP.2023.3278860
6	GE Y X, CHEN D P, LI H S. Mutual mean-teaching: pseudo label refinery for unsupervised domain adaptation on person re-identification[EB/OL]. [2024-05-10]. https://arxiv.org/abs/2001.01526v2.
7	ZHOU Q Q , ZHONG B N , LAN X Y , et al. Fine-grained spatial alignment model for person re-identification with focal triplet loss. IEEE Transactions on Image Processing, 2020, 29, 7578- 7589. doi: 10.1109/TIP.2020.3004267
8	DAI Y , LIU J , BAI Y , et al. Dual-refinement: joint label and feature refinement for unsupervised domain adaptive person re-identification. IEEE Transactions on Image Processing, 2021, 30, 7815- 7829. doi: 10.1109/TIP.2021.3104169
9	CHO Y, KIM W J, HONG S, et al. Part-based pseudo label refinement for unsupervised person re-identification[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans, USA: IEEE Press, 2022: 7308-7318.
10	GE Y X, CHEN D P, ZHU F, et al. Self-paced contrastive learning with hybrid memory for domain adaptive object Re-ID[C]//Proceedings of the Advances in Neural Information Processing Systems. Cambridge, USA: MIT Press, 2020: 11309-11321.
11	DAI Z Z, WANG G Y, YUAN W H, et al. Cluster contrast for Unsupervised person re-identification[C]//Proceedings of the ACCV'22. Berlin, Germany: Springer, 2023: 319-337.
12	LI P N, WU K Y, HUANG W L, et al. Camera-aware label refinement for unsupervised person re-identification[EB/OL]. [2024-05-10]. https://arxiv.org/abs/2403.16450.
13	ZHANG G Q , ZHANG H W , LIN W S , et al. Camera contrast learning for unsupervised person re-identification. IEEE Transactions on Circuits and Systems for Video Technology, 2023, 33 (8): 4096- 4107. doi: 10.1109/TCSVT.2023.3240001
14	毕方明, 王为奎, 陈龙. 基于空间密度的群以噪声发现聚类算法研究. 南京大学学报(自然科学), 2012, 48 (4): 491- 498.
	BI F M , WANG W K , LONG C . DBSCAN: density-based spatial clustering of applications with noise. Journal of Nanjing University(Natural Sciences), 2012, 48 (4): 491- 498.
15	WU F Z , WU J S , KONG Y Y , et al. Multiscale low-frequency memory network for improved feature extraction in convolutional neural networks. Artificial Intelligence, 2024, 38 (6): 5967- 5975.
16	HUANG X, BELONGIE S. Arbitrary style transfer in real-time with adaptive instance normalization[C]//Proceedings of the IEEE International Conference on Computer Vision. Venice, USA: IEEE Press, 2017: 1501-1510.
17	XUAN S , ZHANG S . Intra-inter domain similarity for unsupervised person re-identification. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024, 46 (3): 1711- 1726. doi: 10.1109/TPAMI.2022.3163451
18	ZHENG L, SHEN L Y, TIAN L, et al. Scalable person re-identification: a benchmark[C]//Proceedings of the IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2015: 1116-1124.
19	ZHENG Z D, ZHENG L, YANG Y. Unlabeled samples generated by GAN improve the person re-identification baseline in vitro[C]//Proceedings of the IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2017: 3754-3762.
20	WEI L H, ZHANG S L, GAO W, et al. Person transfer GAN to bridge domain gap for person re-identification[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2018: 79-88.
21	ZHONG Z, ZHENG L, CAO D L, et al. Re-ranking person re-identification with k-reciprocal encoding[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2017: 1318-1327.
22	DENG J, DONG W, SOCHER R, et al. ImageNet: a large-scale hierarchical image database[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2009: 248-255.
23	BERMAN M, JÉGOU H, VEDALDI A, et al. MultiGrain: a unified image embedding for classes and instances[EB/OL]. [2024-05-10]. https://arxiv.org/abs/1902.05509.
24	FU D P, CHEN D D, YANG H, et al. Large-scale pre-training for person re-identification with noisy labels[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans, USA: IEEE Press, 2022: 2476-2486.
25	WANG W , ZHAO F , LIAO S , et al. Attentive WaveBlock: complementarity-enhanced mutual networks for unsupervised domain adaptation in person re-identification and beyond. IEEE Transactions on Image Processing, 2022, 31, 1532- 1544. doi: 10.1109/TIP.2022.3140614
26	HE T , SHEN L Q , GUO Y C , et al. SECRET: self-consistent pseudo label refinement for unsupervised domain adaptive person re-identification. Artificial Intelligence, 2022, 36 (1): 879- 887.
27	WANG M L , LAI B S , HUANG J Q , et al. Camera-aware proxies for unsupervised person re-identification. Artificial Intelligence, 2021, 35 (4): 2764- 2772.
28	CHEN H, LAGADEC B, BREMOND F. ICE: inter-instance contrastive encoding for unsupervised person re-identification[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2021: 14960-14969.
29	JI H , WANG L , ZHOU S P , et al. Transfer easy to hard: adversarial contrastive feature learning for unsupervised person re-identification. Pattern Recognition, 2024, 145, 109973. doi: 10.1016/j.patcog.2023.109973
30	ZHENG Y, TANG S X, TENG G L, et al. Online pseudo label generation by hierarchical cluster dynamics for adaptive person re-identification[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2021: 8371-8381.
31	JIN X, HE T Y, SHEN X, et al. Meta clustering learning for large-scale unsupervised person re-identification[C]//Proceedings of the 30th ACM International Conference on Multimedia. New York, USA: ACM Press, 2022: 2163-2172.
32	CHENG D , ZHOU J Y , WANG N N , et al. Hybrid dynamic contrast and probability distillation for unsupervised person re-identification. IEEE Transactions on Image Processing, 2022, 31, 3334- 3346. doi: 10.1109/TIP.2022.3169693
33	ZHANG X Y, LI D D, WANG Z G, et al. Implicit sample extension for unsupervised person re-identification[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2022: 7369-7378.
34	HAN X , YU X , LI G , et al. Rethinking sampling strategies for unsupervised person re-identification. IEEE Transactions on Image Processin, 2023, 32, 29- 42. doi: 10.1109/TIP.2022.3224325
35	YIN J , ZHANG X , MA Z , et al. A real-time memory updating strategy for unsupervised person re-identification. IEEE Transactions on Image Processin, 2023, 32, 2309- 2321. doi: 10.1109/TIP.2023.3266166

[1]	陈玉敏, 车进, 吴金蔓. 融合注意力机制的师生网络无监督行人重识别[J]. 计算机工程, 2025, 51(11): 304-316.
[2]	朱彦斌, 王润民, 陈华, 曹小菲, 朱祯琳, 丁亚军. 基于多粒度特征增强网络的交通文本检测方法[J]. 计算机工程, 2024, 50(11): 80-88.
[3]	薛振宇, 线岩团, 余正涛, 高盛祥, 普浏清. 融合词典与对抗迁移的越南语事件实体识别[J]. 计算机工程, 2022, 48(3): 107-114,145.
[4]	刘瑄, 池明旻. 基于多粒度特征蒸馏的遥感图像场景分类研究[J]. 计算机工程, 2021, 47(1): 224-229,238.

选择文件类型/文献管理软件名称

选择包含的内容

融合空频信息的多粒度师生网络无监督行人重识别

Unsupervised Person Re-Identification for Multigrained Teacher-Student Networks Incorporating Spatial Frequency Information

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 15

参考文献 35

相关文章 4

编辑推荐

Metrics

本文评价

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

融合空频信息的多粒度师生网络无监督行人重识别

Unsupervised Person Re-Identification for Multigrained Teacher-Student Networks Incorporating Spatial Frequency Information

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 15

参考文献 35

相关文章 4

编辑推荐

Metrics

本文评价