The purpose of person re-identification (re-id) is to identify images of the same person on different cameras. Although unsupervised models have better generalization than supervised models, unsupervised clustering will be more susceptible to noise interference. To address this problem, this paper proposes a model reverse backbone net (RBNet) that can reduce noise interference, using RBNet to learn the keypoints of the human body output from the pose detection model, adjusting the local spatial information and generating masks, and augmenting the specified location attention with the generated masks. The experimental results show that comparing baseline's cross-domain experimental results from Market-1501 to DukeMTMC-reID, mAP is enhanced by 7.0% and Rank-1 by 6.4%. Strengthening the attention to different local information can effectively improve the model accuracy.
聚类跨域实验结果如表3所示。反向骨干块的姿态引导模型生成的掩码可以自由地选择位置,因此,本文采取无监督领域自适应方法实现跨域检索时,输出到了不同的ResNet 50。在数据集Market-1501和Duke上对比两种UDA方法,分别是添加了聚类算法的ResNet 50方法(如fast-net)和MMT方法。基于fastnet的UDA方法很多,有GLT [26](Group-aware label transfer for domain adaptive person re-identification )等。
ChiatP T, SharmiliR, KimH Y. AANet: attribute attention network for person re-identifications[C]∥32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 7127-7136.
[2]
LinY T, ZhengL, ZhengZ D, et al. Improving person re-identification by attribute and identity learning[J]. Pattern Recognition, 2019, 95: 151-161.
[3]
SunY F, ZhengL, YangY, et al. Beyond part models: person retrieval with refined part pooling (and a strong convolutional baseline)[C]∥15th European Conference on Computer Vision, Munich, Germany, 2018: 501-518.
[4]
MiaoJ X, WuY, YangY. Identifying visible parts via pose estimation for occluded person re-identification[J]. IEEE Transactions on Neural Networks and Learning Systems, 2022, 33(9): 4624-4634.
[5]
CaiH L, WangZ G, ChengJ X. Multi-scale body-part mask guided attention for person re-identification[C]∥32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 1555-1564.
[6]
QiaoS Y, ChenL C, AlanY. Detectors: detecting objects with recursive feature pyramid and switchable atrous convolution[C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 10208-10219.
[7]
HouS Q, YinK N, LiangJ, et al. Gradient-supervised person re-identification based on dense feature pyramid network[J]. Complex & Intelligent Systems, 2022, 8(6I): 5329-5342.
[8]
ZhangZ Z, LanG L, ZengW J, et al. Relation-aware global attention for person re-identification[C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition, Seattle, USA, 2020: 3183-3192.
[9]
ZhouK Y, YangY X, CavallaroA, et al. Omni-scale feature learning for person re-identification[C]∥IEEE/CVF International Conference on Computer Vision, Seoul, South Korea, 2019: 3701-3711.
[10]
DivyanshG, NetraP, ThomasT, et al. Towards explainable person re-identification[C]∥IEEE Symposium Series on Computational Intelligence, Orlando, USA, 2021: 9660071.
[11]
NikiM, GianL F, ChristianM. Deep pyramidal pooling with attention for person re-identification[J]. IEEE Transactions on Image Processing, 2020, 29: 7306-7316.
[12]
YangF X, ZhongZ, LuoZ M, et al. Joint noise-tolerant learning and meta camera shift adaptation for unsupervised person re-identification[C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 4853-4862.
[13]
AnttiT, HarriV. Mean teachers are better role models: weight-averaged consistency targets improve semi-supervised deep learning results[C]∥31st Annual Conference on Neural Information Processing Systems, Long Beach, USA, 2017: 1195-1204.
[14]
GeY X, ChenD P, LiH S. Mutual mean-teaching: pseudo label refinery for unsupervised domain adaptation on person re-identification[C]∥8th International Conference on Learning Representations, Addis Ababa, Ethiopia, 2020: 200101526.
[15]
HeK M, ZhangX Y, RenS Q, et al. Deep residual learning for image recognition[C]∥ IEEE Conference on Computer Vision and Pattern Recognition, Seattle, USA, 2016: 770-778.
[16]
SunK, XiaoB, LiuD, et al. Deep high-resolution representation learning for human pose estimation[C]∥32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 5686-5696.
[17]
MartinE, HansP K, JorgS, et al. A density-based algorithm for discovering clusters in large spatial databases with noise[C]∥2nd International Conference on Knowledge Discovery and Data Mining, Portland, USA, 1996: 226-231.
[18]
ZhengL, ShenL Y, TianL, et al. Scalable person re-identification: a benchmark[C]∥IEEE International Conference on Computer Vision, Santiago, USA, 2015: 1116-1124.
[19]
ErgysR, FrancescoS, RogerZ, et al. Performance measures and a data set for multi-target, multi-camera tracking[C]∥14th European Conference on Computer Vision, Amsterdam, Netherlands, 2016: 17-35.
[20]
WeiL H, ZhangS L, GaoW, et al. Person transfer GAN to bridge domain gap for person re-identification[C]∥31st IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, USA, 2018: 79-88.
[21]
ParkH J, HamB. Relation network for person re-identification[C]∥34th AAAI Conference on Artificial Intelligence, New York, USA, 2020: 11839-11847.
[22]
HeL X, LiaoX Y, LiuW, et al. Fastreid: a pytorch toolbox for general instance re-identification[J]. Proceedings of the 31st ACM International Conference on Multimedia, 2020, 10: 9662-9667.
[23]
TanH C, LiuX P, BianY H, et al. Incomplete descriptor mining with elastic loss for person re-identification[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2022, 32(1): 160-171.
[24]
LiY L, HeJ F, ZhangT Z, et al. Diverse part discovery: occluded person re-identification with part-aware transformer[C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 2897-2906.
[25]
LiH J, WuG J, ZhengW S. Combined depth space based architecture search for person re-identification[C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 6725-6734.
[26]
ZhengK C, LiuW, HeL X, et al. Group-aware label transfer for domain adaptive person re-identification[C]∥IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 5306-5315.
[27]
WangD K, ZhangS L. Unsupervised person re-identification via multi-label classification[J]. International Journal of Computer Vision, 2022, 130(12): 2924-2939.
[28]
ZouY, YangX D, YuZ D, et al. Joint disentangling and adaptation for cross-domain person re-identification[C]∥16th European Conference on Computer Vision, Glasgow, UK, 2020: 87-104.
[29]
JinX, LanC L, ZengW J, et al. Global distance-distributions separation for unsupervised person re-identification[C]∥16th European Conference on Computer Vision, Glasgow, UK, 2020: 735-751.
[30]
ZhongZ, ZhengL, LuoZ M, et al. Learning to adapt invariance in memory for person re-identification[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(8): 2723-2738.
[31]
ChenG Y, GuT P, LuJ W, et al. Person re-identification via attention pyramid[J]. IEEE Transactions on Image Processing, 2021, 30: 7663-7676.