Three-Dimension Transmissible Attention Network for Person Re-Identification

Huang, Yewen; Lian, Sicheng; Zhang, Suian; Hu, Haifeng; Chen, Dihu; Su, Tao

doi:10.1109/tcsvt.2020.2977427

Cited by 16 publications

(7 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Backbone Market-1501 Rank-1 mAP DaRe (CVPR18) [33] ResNet50 86.4 69.3 DaRe+RE (CVPR18) [33] ResNet50 88.5 74.2 PSE+ECN (CVPR18) [36] ResNet50 90.4 80.5 HA-CNN (CVPR18) [26] ResNet50 91.2 75.7 DuATM (CVPR18) [37] Inception-A 91.4 76.6 PCB+RPP (CVPR18) [38] ResNet50 93.8 81.6 MHN-PCB (ICCV19) [45] ResNet50 95.1 85.0 MGN (ACMMM18) [42] ResNet50 95.7 86.9 HPM (AAAI19) [66] ResNet50 94.2 82.7 AANet (CVPR19) [17] ResNet152 93.9 83.4 DCDS(ICCV19) [46] ResNet101 94.8 85.8 OSNet (ICCV19) [48] OSNeT 94.8 84.9 GCP (AAAI20) [47] ResNet50 94.8 88.0 SAN(AAAI20) [49] ResNet50 95.1 85.8 3DTANet (TCSVT20) [50] -95.3 86.9 HOReID (CVPR20) [51] ResNet50 94.2 84.9 RGA-CS(CVPR20) [ highest accuracy on mAP and only 0.3% improvement over our method. DukeMTMC-reID: It is another standard dataset that contains a sufficient number of images for deep learning.…”

Section: Methodsmentioning

confidence: 99%

“…Attention mechanism has also been adopted to address the re-ID task [17], [56], [55], [58], [45], [38], [50] hope to find out more discriminative features (color or texture) in the image through attention machanism, and ignore the other features (background) which are irrelevant to task. Sun et al [27] simply used the attention mechanism to locate the visible areas in a given image through a positioning method and learned their local features based on these visible areas.…”

Section: Attention Mechanism For Person Re-idmentioning

confidence: 99%

“…Recently, many researchers have introduced attention mechanisms into deep models to strengthen the learned features and suppress some disturbances of the features [17], [50], [55], [56]. The purpose of attention is to focus on important features and ignore irrelevant features.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Hybrid-attention guided network with multiple resolution features for person re-identification

Guo-qing

Yang

Zheng

et al. 2021

Information Sciences

View full text Add to dashboard Cite

Section: Methodsmentioning

confidence: 99%

Section: Attention Mechanism For Person Re-idmentioning

confidence: 99%

See 1 more Smart Citation

Hybrid-attention guided network with multiple resolution features for person re-identification

Guo-qing

Yang

Zheng

et al. 2021

Information Sciences

View full text Add to dashboard Cite

“…Currently, a majority of re-ID methods are based on supervised learning and have achieved impressive performance, which require a large amount of labeled data. With the developments of deep learning, methods based on convolutional neural network (CNN) [3], [22]- [25] have almost dominated this field. These methods mostly focus on how to extract more discriminative features to improve the recognition accuracy.…”

Section: Related Work a Supervised Person Re-identificationmentioning

confidence: 99%

“…Huang et al [30] proposes a method to spatially pay attention to the region of interest, and the discriminative information of the image will be magnified. Also, Huang et al [25] propose 3-Dimension Transmissible Attention (3DTA) that cooperatively utilizes channel attention and spatial attention, with a group loss to optimize the feature distances. In [31], relation network puts forward a global attention mechanism with relation-aware and has achieved superior performance.…”

Section: Related Work a Supervised Person Re-identificationmentioning

confidence: 99%

Noise Resistible Network for Unsupervised Domain Adaptation on Person Re-Identification

Zhang

Zeng

et al. 2021

IEEE Access

Self Cite

View full text Add to dashboard Cite

Unsupervised domain adaptation on person re-identification (re-ID), which adapts the model trained on source dataset to the target dataset, has drawn increasing attention over the past few years. It is more practical than the traditional supervised methods when applied in the real-world scenarios since they require a huge number of manual annotations in a specific domain, which is unrealistic and even under personal privacy concerns. Currently, pseudo label-based method is one of the most promising solutions in this area. However, in such methods, pseudo label noise is ignored and remains a huge challenge hindering further performance improvements. To solve this problem, this paper proposes a novel unsupervised domain adaptation re-ID framework named Noise Resistible Network (NRNet), which mainly consists of two dual-stream networks. For one thing, during pseudo label generation, NRNet utilizes one dual-stream network, denoted as clustering network, to generate discriminative features in the unseen domain for further clustering, reducing the pseudo label noise. For another, to avoid the problem of close loop noise amplification in conventional methods, the other dual-stream network named temporally average network is constructed outside the clustering loop to learn how to identify the images of the same person. In addition, two dual-stream networks are designed with a guiding mechanism, which allows the shallow network to learn more representative feature embedding from the deep network. Extensive experimental results on two widely-used benchmark datasets, i.e., Market-1501 and DukeMTMC-reID demonstrate that our proposed NRNet outperforms the state-of-the-art methods.INDEX TERMS Unsupervised domain adaptation, person re-identification, noise resistible network.

show abstract

Two‐level salient feature complementary network for person re‐identification

Liu

et al. 2022

Int J of Intelligent Sys

View full text Add to dashboard Cite

Given a query image from a camera, person re‐identification (Re‐ID) can retrieve the images of the same identity from a gallery, the images of which are captured by the other cameras. Therefore, person Re‐ID has been widely used in the field of video surveillance. However, person Re‐ID still suffers from a series of challenges, such as illumination changes, pose variations, and occlusions. Although the person Re‐ID methods based on attention mechanism give an effective and feasible solution for the above challenges, attention mechanism may make a network focus too much on the most salient discriminative features and ignore other potential discriminative features. To solve this problem, we propose a two‐level salient feature complementary network (TSFC‐Net) to extract the most salient discriminative features and the secondary salient discriminative features of pedestrian images for person Re‐ID. Specifically, TSFC‐Net first extracts the most salient discriminative features of pedestrian images by embedding the spatial and channel attention modules in the backbone network, and then extracts the secondary salient discriminative features of pedestrian images by a secondary salient feature mining module (SSFM). Since the final features of pedestrian images fuse the most salient discriminative features and the secondary salient discriminative features, TSFC‐Net can significantly improve the richness and discrimination capability of pedestrian representations. In addition, we conduct extensive experiments on the Market‐1501, DukeMTMC‐reID, and CUHK03 data sets, and the experimental results indicate that our TSFC‐Net has a better performance compared with most of the state‐of‐the‐art person Re‐ID methods.

show abstract

Three-Dimension Transmissible Attention Network for Person Re-Identification

Cited by 16 publications

References 39 publications

Hybrid-attention guided network with multiple resolution features for person re-identification

Hybrid-attention guided network with multiple resolution features for person re-identification

Noise Resistible Network for Unsupervised Domain Adaptation on Person Re-Identification

Two‐level salient feature complementary network for person re‐identification

Contact Info

Product

Resources

About