Dataset Preparation.

1. Download MARS dataset from here.

2. Download Market-1501 dataset from here.

3. Download CUHK03 dataset from here. You need to extract the images into folods. You can also download the new protocol version CUHK03-NP. If you use this dataset in your work, please cite their paper:

   @inproceedings{zhong2017re,
   title={Re-ranking Person Re-identification with k-reciprocal Encoding},
   author={Zhong, Zhun and Zheng, Liang and Cao, Donglin and Li, Shaozi},
   booktitle={CVPR},
   year={2017}
   }

   @inproceedings{li2014deepreid, title={DeepReID: Deep Filter Pairing Neural Network for Person Re-identification},
   author={Li, Wei and Zhao, Rui and Xiao, Tong and Wang, Xiaogang},
   booktitle={CVPR},
   year={2014} }

• All masks can be download from Baidu Yun OR Google Drive. If you use the masks in your work, please cite our paper:

Chunfeng Song, Yan Huang, Wanli Ouyang, and Liang Wang. Mask-guided Contrastive Attention Model for Person Re-Identification. In CVPR, 2018.

Make sure that all the datasets are saveing in the following structure:

MARS:

./data ./data/mars
./data/mars/bbox_train ./data/mars/bbox_test
./data/mars/bbox_train_seg ./data/mars/bbox_test_seg

Market-1501:

./data ./data/market-1501 ./data/market-1501/bounding_box_train
./data/market-1501/bounding_box_test
./data/market-1501/query
./data/market-1501/bounding_box_train_seg
./data/market-1501/bounding_box_test_seg
./data/market-1501/query_seg

CUHK03:

./data ./data/cuhk03
./data/cuhk03/labeled
./data/cuhk03/cuhk03_labeled_seg
./data/cuhk03/detected ./data/cuhk03/cuhk03_detected_seg

CUHK03-NP:

./data ./data/cuhk03-np
./data/cuhk03-np/labeled
./data/cuhk03-np/labeled/bounding_box_train
./data/cuhk03-np/labeled/bounding_box_test ./data/cuhk03-np/labeled/query ./data/cuhk03-np/labeled/bounding_box_train_seg
./data/cuhk03-np/labeled/bounding_box_test_seg ./data/cuhk03-np/labeled/query_seg

./data/cuhk03-np/detected
./data/cuhk03-np/detected/bounding_box_train
./data/cuhk03-np/detected/bounding_box_test
./data/cuhk03-np/detected/query
./data/cuhk03-np/detected/bounding_box_train_seg
./data/cuhk03-np/detected/bounding_box_test_seg
./data/cuhk03-np/detected/query_seg

Now, you can run the code to generate training set with running "python gen_market1501_and_cuhk03_train_set.py".