1st Place Solution to VisDA-2020: Bias Elimination for Domain Adaptive Pedestrian Re-identification

This paper presents our proposed methods for domain adaptive pedestrian re-identification (Re-ID) task in Visual Domain Adaptation Challenge (VisDA-2020). Considering the large gap between the source domain and target domain, we focused on solving two biases that influenced the performance on domain adaptive pedestrian Re-ID and proposed a two-stage training procedure. At the first stage, a baseline model is trained with images transferred from source domain to target domain and from single camera to multiple camera styles. Then we introduced a domain adaptation framework to train the model on source data and target data simultaneously. Different pseudo label generation strategies are adopted to continuously improve the discriminative ability of the model. Finally, with multiple models ensembled and additional post processing approaches adopted, our methods achieve 76.56% mAP and 84.25% rank-1 on the test set. Codes are available at https://github.com/vimar-gu/Bias-Eliminate-DA-ReID.

[1]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[2]  Yi Yang,et al.  Image-Image Domain Adaptation with Preserved Self-Similarity and Domain-Dissimilarity for Person Re-identification , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[3]  Cheng Wang,et al.  Unsupervised Domain Adaptive Re-Identification: Theory and Practice , 2018, Pattern Recognit..

[4]  Yi Yang,et al.  Camera Style Adaptation for Person Re-identification , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[5]  Liang Zheng,et al.  Dissecting Person Re-Identification From the Viewpoint of Viewpoint , 2018, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

[6]  Wei Jiang,et al.  Bag of Tricks and a Strong Baseline for Deep Person Re-Identification , 2019, 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW).

[7]  Yi Yang,et al.  Random Erasing Data Augmentation , 2017, AAAI.

[8]  Xiaoou Tang,et al.  Two at Once: Enhancing Learning and Generalization Capacities via IBN-Net , 2018, ECCV.

[9]  Chunhua Shen,et al.  Self-Training With Progressive Augmentation for Unsupervised Cross-Domain Person Re-Identification , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[10]  Wei Jiang,et al.  A Strong Baseline and Batch Normalization Neck for Deep Person Re-Identification , 2019, IEEE Transactions on Multimedia.

[11]  Yang Zhao,et al.  Deep High-Resolution Representation Learning for Visual Recognition , 2019, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[12]  Shaogang Gong,et al.  Instance-Guided Context Rendering for Cross-Domain Person Re-Identification , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[13]  Yi Yang,et al.  Generalizing a Person Retrieval Model Hetero- and Homogeneously , 2018, ECCV.

[14]  Yunchao Wei,et al.  Self-Similarity Grouping: A Simple Unsupervised Cross Domain Adaptation Approach for Person Re-Identification , 2018, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[15]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[16]  Qi Tian,et al.  Beyond Part Models: Person Retrieval with Refined Part Pooling , 2017, ECCV.

[17]  Liang Zheng,et al.  Re-ranking Person Re-identification with k-Reciprocal Encoding , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).