Code of paper "Parallel Augmentation and Dual Enhancement for Occluded Person Re-identification"
A simple but effective method for both Occluded Person Re-identification and Normal Person Re-identification (with few occlusions)
Paper link: Parallel Augmentation and Dual Enhancement for Occluded Person Re-identification. PDF
The codes are based on the TransReID (ICCV 2021), the basic preparation and environment installation, please refer to TransReID.
We visualized some of the ranking list results (rank-10) on the Partial ReID dataset. The above is the result of the baseline, and below is our method.
We will evaluate the model every few epochs.
Note: Since the Partial-REID and Occluded-ReID datasets have few samples in the test set and are easy to overfit, we adopted the "early stopping" strategy and manually selected the results for better accuracy.
# Training on Occluded-Duke
python train.py --config_file configs/OCC_Duke/vit_transreid_stride.yml MODEL.DEVICE_ID "('2')"
# Training on Partial-REID
python train.py --config_file configs/Partial_ReID/vit_transreid_stride.yml MODEL.DEVICE_ID "('2')"
# Training on Occluded-ReID
python train.py --config_file configs/OCC_ReID/vit_transreid_stride.yml MODEL.DEVICE_ID "('2')"
# Training on Market-1501
python train.py --config_file configs/Market/vit_transreid_stride.yml MODEL.DEVICE_ID "('2')"
# Training on DukeMTMC-reID
python train.py --config_file configs/DukeMTMC/vit_transreid_stride.yml MODEL.DEVICE_ID "('2')"The pre-trained models will come soon...
We demonstrate the training and test data imbalance problem of occluded ReID by displaying samples in the training set and test set. Note: Pick one image for each ID as a representative.
Training data (Market 1501): only a few IDs are obscured
Testing data (Partial-REID): query (left, 100% occluded), gallery (right, ~ 100% non-occluded)
Testing data (Occluded-REID): query (left, 100% occluded), gallery (right, ~ 100% non-occluded)
Please contact Zi Wang (email address: ziwang1121@foxmail.com). Feel free to drop me an email if you have any questions.