AGW-mindspore

This is a mindspore implementation of the AGW baseline purposed in Deep Learning for Person Re-identification: A Survey and Outlook. arXiv

Environment

• Python 3.7
• Mindspore 1.5.0 (Ascend)

Usage

• prepare Imagenet pretrained resnet50 checkpoints
• organize datasets as below

├──"data_path" in agw_config.yaml
   ├──market1501
      ├──Market-1501
         ├──bounding_box_train
         ├──query
         ├──bounding_box_test
   ├──dukemtmc-reid
      ├──DukeMTMC-reID
         ├──bounding_box_train
         ├──query
         ├──bounding_box_test
   ├──cuhk03
      ├──cuhk03_release
         ├──cuhk-03.mat
         ├──cuhk03_new_protocol_config_labeled.mat
         ├──cuhk03_new_protocol_config_detected.mat
   ├──msmt17
      ├──MSMT17_V1
         ├──train
         ├──test
         ├──list_val.txt
         ├──list_train.txt
         ├──list_query.txt
         ├──list_query.txt

• Train

export DEVICE_ID=0
python train.py --source=DATASET_NAME > $LOG_DIR 2>&1 

• Evaluate

python eval.py --target=DATASET_NAME --checkpoint_path=TRAINED_AGW_CHECKPOINTS > $LOG_DIR 2>&1 

where checkpoints are saved in "output_path" in config and is set to ./output in default.