《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (11): 3573-3579.DOI: 10.11772/j.issn.1001-9081.2021122124
所属专题: 第二十一届中国虚拟现实大会
收稿日期:
2021-12-17
修回日期:
2022-02-28
接受日期:
2022-03-07
发布日期:
2022-05-17
出版日期:
2022-11-10
通讯作者:
廉永健
作者简介:
耿艳兵(1980—),女,河南漯河人,讲师,博士,CCF会员,主要研究方向:图像处理、模式识别、人工智能基金资助:
Received:
2021-12-17
Revised:
2022-02-28
Accepted:
2022-03-07
Online:
2022-05-17
Published:
2022-11-10
Contact:
Yongjian LIAN
About author:
GENG Yanbing, born in 1980, Ph. D., lecturer. Her research interests include image processing, pattern recognition, artificial intelligence.Supported by:
摘要:
现有基于生成对抗网络(GAN)的超分辨率(SR)重建方法用于跨分辨率行人重识别(ReID)时,重建图像在纹理结构内容的恢复和特征一致性保持方面均存在不足。针对上述问题,提出基于多粒度信息生成网络的跨分辨率行人ReID方法。首先,在生成器的多层网络上均引入自注意力机制,聚焦多粒度稳定的结构关联区域,重点恢复低分辨率(LR)行人图像的纹理结构信息;同时,在生成器后增加一个识别器,在训练过程中最小化生成图像与真实图像在不同粒度特征上的损失,提升生成图像与真实图像在特征上的一致性。然后,联合自注意力生成器和识别器,与判别器交替优化,在内容和特征上改进生成图像。最后,联合改进的GAN和行人ReID网络交替训练优化网络的模型参数,直至模型收敛。在多个跨分辨率行人数据集上的实验结果表明,所提算法的累计匹配曲线(CMC)在其首选识别率(rank?1)上的准确率较现有同类算法平均提升10个百分点,在提升SR图像内容一致性和特征表达一致性方面均表现更优。
中图分类号:
耿艳兵, 廉永健. 基于多粒度特征生成对抗网络的跨分辨率行人重识别[J]. 计算机应用, 2022, 42(11): 3573-3579.
Yanbing GENG, Yongjian LIAN. Cross‑resolution person re‑identification by generative adversarial network based on multi‑granularity features[J]. Journal of Computer Applications, 2022, 42(11): 3573-3579.
方法 | MLR‑CUHK03 | |||
---|---|---|---|---|
SSIM | PSNR/dB | LPIPS | rank‑1/% | |
GAN | 0.30 | 13.9 | 0.42 | 47.8 |
GAN+自注意力 | 0.61 | 19.2 | 0.19 | 65.3 |
GAN+识别网络(高层损失) | 0.41 | 18.1 | 0.23 | 76.9 |
GAN+识别网络(中低层损失) | 0.57 | 20.9 | 0.16 | 78.1 |
GAN+识别网络 (高层&中低层损失) | 0.61 | 21.3 | 0.13 | 80.5 |
GAN+自注意力+识别网络 (高层损失) | 0.69 | 22.5 | 0.12 | 85.5 |
GAN+自注意力+识别网络 (中低层损失) | 0.71 | 22.8 | 0.08 | 85.3 |
GAN+自注意力+识别网络 (高层&中低层损失) | 0.89 | 23.5 | 0.06 | 87.1 |
表1 多粒度信息融合生成对抗网络在MLR?CUHK03数据集上的消融实验结果
Tab. 1 Ablation experimental results of multi?granularity feature fusion based generative adversarial network on MLR?CUHK03 dataset
方法 | MLR‑CUHK03 | |||
---|---|---|---|---|
SSIM | PSNR/dB | LPIPS | rank‑1/% | |
GAN | 0.30 | 13.9 | 0.42 | 47.8 |
GAN+自注意力 | 0.61 | 19.2 | 0.19 | 65.3 |
GAN+识别网络(高层损失) | 0.41 | 18.1 | 0.23 | 76.9 |
GAN+识别网络(中低层损失) | 0.57 | 20.9 | 0.16 | 78.1 |
GAN+识别网络 (高层&中低层损失) | 0.61 | 21.3 | 0.13 | 80.5 |
GAN+自注意力+识别网络 (高层损失) | 0.69 | 22.5 | 0.12 | 85.5 |
GAN+自注意力+识别网络 (中低层损失) | 0.71 | 22.8 | 0.08 | 85.3 |
GAN+自注意力+识别网络 (高层&中低层损失) | 0.89 | 23.5 | 0.06 | 87.1 |
图4 多粒度信息融合生成对抗网络的不同组件在MLR?CUHK03数据集上的可视化结果
Fig. 4 Visual results of different components in multi?granularity feature fusion based GAN on MLR?CUHK03 dataset
方法 | MLR‑Market1501 | MLR‑CUHK03 | MLR‑VIPeR | MLR‑DukeMTMC‑reID | CAVIAR | |||||
---|---|---|---|---|---|---|---|---|---|---|
rank‑1 | rank‑5 | rank‑1 | rank‑5 | rank‑1 | rank‑5 | rank‑1 | rank‑5 | rank‑1 | rank‑5 | |
CamStyle | 74.5 | 88.6 | 69.1 | 89.6 | 34.4 | 56.8 | 64.0 | 78.1 | 32.1 | 72.3 |
FD‑GAN | 79.6 | 91.6 | 73.4 | 92.8 | 39.1 | 62.1 | 67.5 | 82.0 | 33.5 | 71.4 |
SLD2L | ― | ― | ― | ― | 20.3 | 44.0 | ― | ― | 18.4 | 44.8 |
SING | 74.4 | 87.8 | 67.7 | 90.7 | 33.5 | 57.0 | 65.2 | 80.1 | 33.5 | 72.7 |
CSR‑GAN | 76.4 | 88.5 | 71.3 | 92.1 | 37.2 | 62.3 | 67.6 | 81.4 | 34.7 | 72.5 |
JUDEA | ― | ― | 26.2 | 58.0 | 26.0 | 55.1 | ― | ― | 22.0 | 60.1 |
SDF | ― | ― | 22.2 | 48.0 | 9.3 | 38.1 | ― | ― | 14.3 | 37.5 |
RAIN | ― | ― | 78.9 | 97.3 | 42.5 | 68.3 | ― | ― | 42.0 | 77.3 |
CAD | 83.7 | 92.7 | 82.1 | 97.4 | 43.1 | 68.2 | 75.6 | 86.7 | 42.8 | 76.2 |
INTACT | 88.1 | 95.0 | 86.4 | 97.4 | 46.2 | 73.1 | 81.2 | 90.1 | 44.0 | 81.8 |
本文方法 | 88.9 | 96.3 | 87.1 | 97.5 | 46.9 | 73.6 | 82.1 | 91.8 | 44.9 | 83.2 |
表2 本文方法与现有同类方法在不同数据集上的CMC rank?1和rank?5准确率 ( %)
Tab. 2 Rank?1 and rank?5 accuracies on CMC of proposed method and existing similar methods on different datasets
方法 | MLR‑Market1501 | MLR‑CUHK03 | MLR‑VIPeR | MLR‑DukeMTMC‑reID | CAVIAR | |||||
---|---|---|---|---|---|---|---|---|---|---|
rank‑1 | rank‑5 | rank‑1 | rank‑5 | rank‑1 | rank‑5 | rank‑1 | rank‑5 | rank‑1 | rank‑5 | |
CamStyle | 74.5 | 88.6 | 69.1 | 89.6 | 34.4 | 56.8 | 64.0 | 78.1 | 32.1 | 72.3 |
FD‑GAN | 79.6 | 91.6 | 73.4 | 92.8 | 39.1 | 62.1 | 67.5 | 82.0 | 33.5 | 71.4 |
SLD2L | ― | ― | ― | ― | 20.3 | 44.0 | ― | ― | 18.4 | 44.8 |
SING | 74.4 | 87.8 | 67.7 | 90.7 | 33.5 | 57.0 | 65.2 | 80.1 | 33.5 | 72.7 |
CSR‑GAN | 76.4 | 88.5 | 71.3 | 92.1 | 37.2 | 62.3 | 67.6 | 81.4 | 34.7 | 72.5 |
JUDEA | ― | ― | 26.2 | 58.0 | 26.0 | 55.1 | ― | ― | 22.0 | 60.1 |
SDF | ― | ― | 22.2 | 48.0 | 9.3 | 38.1 | ― | ― | 14.3 | 37.5 |
RAIN | ― | ― | 78.9 | 97.3 | 42.5 | 68.3 | ― | ― | 42.0 | 77.3 |
CAD | 83.7 | 92.7 | 82.1 | 97.4 | 43.1 | 68.2 | 75.6 | 86.7 | 42.8 | 76.2 |
INTACT | 88.1 | 95.0 | 86.4 | 97.4 | 46.2 | 73.1 | 81.2 | 90.1 | 44.0 | 81.8 |
本文方法 | 88.9 | 96.3 | 87.1 | 97.5 | 46.9 | 73.6 | 82.1 | 91.8 | 44.9 | 83.2 |
1 | 魏文钰,杨文忠,马国祥,等. 基于深度学习的行人再识别技术研究综述[J]. 计算机应用, 2020, 40(9):2479-2492. 10.11772/j.issn.1001-9081.2020010038 |
WEI W Y, YANG W Z, MA G X, et al. Survey of person re‑identification technology based on deep learning[J]. Journal of Computer Applications, 2020, 40(9):2479-2492. 10.11772/j.issn.1001-9081.2020010038 | |
2 | CHANG X B, HOSPEDALES T M, XIANG T. Multi‑level factorisation net for person re‑identification[C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 2109-2118. 10.1109/cvpr.2018.00225 |
3 | CHEN D P, XU D, LI H S, et al. Group consistent similarity learning via deep CRF for person re‑identification[C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 8649-8658. 10.1109/cvpr.2018.00902 |
4 | FARENZENA M, BAZZANI L, PERINA A, et al. Person re‑identification by symmetry‑driven accumulation of local features[C]// Proceedings of the 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2010:2360-2367. 10.1109/cvpr.2010.5539926 |
5 | CHEN Y C, LI Y J, DU X F, et al. Learning resolution‑invariant deep representations for person re‑identification[C]// Proceedings of the 33rd AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2019:8215-8222. 10.1609/aaai.v33i01.33018215 |
6 | LI X, ZHENG W S, WANG X J, et al. Multi‑scale learning for low‑resolution person re‑identification[C]// Proceedings of the 2015 IEEE International Conference on Computer Vision. Piscataway: IEEE, 2015: 3765‑3773. 10.1109/iccv.2015.429 |
7 | JING X Y, ZHU X K, WU F, et al. Super‑resolution person re‑identification with semi‑coupled low‑rank discriminant dictionary learning[C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015: 695-704. 10.1109/cvpr.2015.7298669 |
8 | 胡雪影,郭海儒,朱蓉. 基于混合深度卷积网络的图像超分辨率重建[J]. 计算机应用, 2020, 40(7): 2069-2076. 10.11772/j.issn.1001-9081.2019122149 |
HU X Y, GUO H R, ZHU R. Image super‑resolution reconstruction based on hybrid deep convolutional network[J]. Journal of Computer Applications, 2020, 40(7):2069-2076. 10.11772/j.issn.1001-9081.2019122149 | |
9 | WANG S L, ZHANG L, LIANG Y, et al. Semi‑coupled dictionary learning with applications to image super‑resolution and photo‑sketch synthesis[C]// Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2012: 2216-2223. 10.1109/cvpr.2012.6247930 |
10 | DONG C, LOY C C, HE K M, et al. Image super‑resolution using deep convolutional networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2):295-307. 10.1109/tpami.2015.2439281 |
11 | LEDIG C, THEIS L, HUSZÁR F, et al. Photo‑realistic single image super‑resolution using a generative adversarial network[C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017: 105-114. 10.1109/cvpr.2017.19 |
12 | 欧阳宁,梁婷,林乐平. 基于自注意力网络的图像超分辨率重建[J]. 计算机应用, 2019, 39(8):2391-2395. 10.11772/j.issn.1001-9081.2019010158 |
OUYANG N, LIANG T, LIN L P. Self‑attention network based image super‑resolution[J]. Journal of Computer Applications, 2019, 39(8):2391-2395. 10.11772/j.issn.1001-9081.2019010158 | |
13 | 焦云清,王世新,周艺,等.基于神经网络的遥感影像超高分辨率目标识别[J].系统仿真学报,2007,19(14):3223-3225. 10.3969/j.issn.1004-731X.2007.14.025 |
JIAO Y Q, WANG S X, ZHOU Y, et al. Super‑resolution target identification from remotely sensed imagery using Hopfield Neural Network[J]. Journal of System Simulation, 2007, 19(14):3223-3225. 10.3969/j.issn.1004-731X.2007.14.025 | |
14 | 潘志庚,郑星,张明敏. 多分辨率模型生成中颜色和纹理属性的处理[J]. 系统仿真学报, 2002, 14(11):1506-1508, 1530. 10.3969/j.issn.1004-731X.2002.11.027 |
PAN Z G, ZHENG X, ZHANG M M. Processing of color and texture attributes in multi‑resolution modeling[J]. Journal of System Simulation, 2002, 14(11):1506-1508, 1530. 10.3969/j.issn.1004-731X.2002.11.027 | |
15 | 陈佛计,朱枫,吴清潇,等. 生成对抗网络及其在图像生成中的应用研究综述[J]. 计算机学报, 2021, 44(2):347-369. 10.11897/SP.J.1016.2021.00347 |
CHEN F J, ZHU F, WU Q X, et al. A survey about image generation with generative adversarial nets[J]. Chinese Journal of Computers, 2021, 44(2):347-369. 10.11897/SP.J.1016.2021.00347 | |
16 | HAN K, HUANG Y, SONG C F, et al. Adaptive super‑resolution for person re‑identification with low‑resolution images[J]. Pattern Recognition, 2021, 114: No.107682. 10.1016/j.patcog.2020.107682 |
17 | JIAO J N, ZHENG W S, WU A C, et al. Deep low‑resolution person re‑identification[C]// Proceedings of the 32nd AAAI Conference on Artificial Intelligence. Palo Alto, CA: AAAI Press, 2018:6967-6974. 10.1609/aaai.v32i1.12284 |
18 | WANG Z, YE M, YANG F, et al. Cascaded SR‑GAN for scale‑ adaptive low resolution person re‑identification[C]// Proceedings of the 27th International Joint Conference on Artificial Intelligence. California: ijcai.org, 2018: 3891-3897. 10.24963/ijcai.2018/541 |
19 | LI Y J, CHEN Y C, LIN Y Y, et al. Recover and identify: a generative dual model for cross‑resolution person re‑identification[C]// Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. Piscataway: IEEE, 2019: 8089-8098. 10.1109/iccv.2019.00818 |
20 | CHENG Z Y, DONG Q, GONG S G, et al. Inter‑task association critic for cross‑resolution person re‑identification[C]// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 2602-2612. 10.1109/cvpr42600.2020.00268 |
21 | 丁明航,邓然然,邵恒. 基于注意力生成对抗网络的图像超分辨率重建方法[J]. 计算机系统应用, 2020, 29(2):209-211. |
DING M H, DENG R R, SHAO H. Image super‑resolution reconstruction method based on attentive generative adversarial network[J]. Computer Systems and Applications, 2020, 29(2):209-211. | |
22 | 许一宁,何小海,张津,等. 基于多层次分辨率递进生成对抗网络的文本生成图像方法[J]. 计算机应用, 2020, 40(12):3612-3617. 10.11772/j.issn.1001-9081.2020040575 |
XU Y N, HE X H, ZHANG J, et al. Text‑to‑image synthesis method based on multi‑level progressive resolution generative adversarial networks[J]. Journal of Computer Applications, 2020, 40(12):3612-3617. 10.11772/j.issn.1001-9081.2020040575 | |
23 | 王雪松,晁杰,程玉虎. 基于自注意力生成对抗网络的图像超分辨率重建[J]. 控制与决策, 2021, 36(6):1324-1332. 10.13195/j.kzyjc.2019.1290 |
WANG X S, CHAO J, CHENG Y H. Image super‑resolution reconstruction based on self‑attention GAN[J]. Control and Decision, 2021, 36(6):1324-1332. 10.13195/j.kzyjc.2019.1290 | |
24 | 杨婉香,严严,陈思,等. 基于多尺度生成对抗网络的遮挡行人重识别方法[J]. 软件学报, 2020, 31(7):1943-1958. |
YANG W X, YAN Y, CHEN S, et al. Multi‑scale generative adversarial network for person re‑identification under occlusion[J]. Journal of Software, 2020, 31(7):1943-1958. | |
25 | DENG J, DONG W, SOCHER R, et al. ImageNet: a large‑scale hierarchical image database[C]// Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2009:248-255. 10.1109/cvpr.2009.5206848 |
26 | GULRAJANI I, AHMED F, ARJOVSKY M, et al. Improved training of Wasserstein GANs[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2017: 5769-5779. |
27 | LI W, ZHU X T, GONG S G. Person re‑identification by deep joint learning of multi‑loss classification[C]// Proceedings of the 26th International Joint Conference on Artificial Intelligence. California: ijcai.org, 2017: 2194-2200. 10.24963/ijcai.2017/305 |
28 | YUAN Y, CHEN W Y, YANG Y, et al. In defense of the triplet loss again: learning robust person re‑identification with fast approximated triplet loss and label distillation[C]// Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. Piscataway: IEEE, 2020: 1454-1463. 10.1109/cvprw50498.2020.00185 |
29 | ZHONG Z, ZHENG L, ZHENG Z D, et al. Camera style adaptation for person re‑identification[C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 5157-5166. 10.1109/cvpr.2018.00541 |
30 | GE Y X, LI Z W, ZHAO H Y, et al. FD‑GAN: pose‑guided feature distilling GAN for robust person re‑identification[C]// Proceedings of the 32nd International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2018: 1230-1241. |
31 | WANG Z, HU R M, YU Y, et al. Scale‑adaptive low‑resolution person re‑identification via learning a discriminating surface[C]// Proceedings of the 25th International Joint Conference on Artificial Intelligence. California: ijcai.org, 2016: 2669-2675. |
32 | SUN Y F, ZHENG L, YANG Y, et al. Beyond part models: person retrieval with refined part pooling (and a strong convolutional baseline)[C]// Proceedings of the 2018 European Conference on Computer Vision, LNCS 11208. Cham: Springer, 2018: 501-518. |
[1] | 贾洁茹, 杨建超, 张硕蕊, 闫涛, 陈斌. 基于自蒸馏视觉Transformer的无监督行人重识别[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2893-2902. |
[2] | 秦璟, 秦志光, 李发礼, 彭悦恒. 基于概率稀疏自注意力神经网络的重性抑郁疾患诊断[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2970-2974. |
[3] | 李力铤, 华蓓, 贺若舟, 徐况. 基于解耦注意力机制的多变量时序预测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2732-2738. |
[4] | 王翠, 邓淼磊, 张德贤, 李磊, 杨晓艳. 基于图像的端到端行人搜索算法综述[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2544-2550. |
[5] | 刘丽, 侯海金, 王安红, 张涛. 基于多尺度注意力的生成式信息隐藏算法[J]. 《计算机应用》唯一官方网站, 2024, 44(7): 2102-2109. |
[6] | 徐泽鑫, 杨磊, 李康顺. 较短的长序列时间序列预测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(6): 1824-1831. |
[7] | 刘越, 刘芳, 武奥运, 柴秋月, 王天笑. 基于自注意力机制与图卷积的3D目标检测网络[J]. 《计算机应用》唯一官方网站, 2024, 44(6): 1972-1977. |
[8] | 王昊冉, 于丹, 杨玉丽, 马垚, 陈永乐. 面向工控系统未知攻击的域迁移入侵检测方法[J]. 《计算机应用》唯一官方网站, 2024, 44(4): 1158-1165. |
[9] | 黄荣, 宋俊杰, 周树波, 刘浩. 基于自监督视觉Transformer的图像美学质量评价方法[J]. 《计算机应用》唯一官方网站, 2024, 44(4): 1269-1276. |
[10] | 郑毅, 廖存燚, 张天倩, 王骥, 刘守印. 面向城区的基于图去噪的小区级RSRP估计方法[J]. 《计算机应用》唯一官方网站, 2024, 44(3): 855-862. |
[11] | 罗歆然, 李天瑞, 贾真. 基于自注意力机制与词汇增强的中文医学命名实体识别[J]. 《计算机应用》唯一官方网站, 2024, 44(2): 385-392. |
[12] | 黄子麒, 胡建鹏. 实体类别增强的汽车领域嵌套命名实体识别[J]. 《计算机应用》唯一官方网站, 2024, 44(2): 377-384. |
[13] | 仇丽青, 苏小盼. 个性化多层兴趣提取点击率预测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3411-3418. |
[14] | 杨兴耀, 沈洪涛, 张祖莲, 于炯, 陈嘉颖, 王东晓. 基于层级过滤器和时间卷积增强自注意力网络的序列推荐[J]. 《计算机应用》唯一官方网站, 2024, 44(10): 3090-3096. |
[15] | 李言博, 何庆, 陆顺意. 融合语义和句法信息的方面情感三元组抽取[J]. 《计算机应用》唯一官方网站, 2024, 44(10): 3275-3280. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||