ASCC-CCReID

ASCC: Semantic Cross-branch Collaboration with Chebyshev’s Theorem-guided Graph Re-ranking for Cloth-Changing Person Re-Identification Our core code of the model will be updated afer the paper is accepted!

Getting Started

Environment

• Python == 3.8
• PyTorch == 1.12.1

Prepare Data

Please download cloth-changing person re-identification datasets and place them in any path DATASET_ROOT:

DATASET_ROOT
	└─ LTCC-reID or PRCC or Celeb-reID
		├── train
		├── query
		└── gallery

Training

# LTCC
python main.py --gpu_devices 0 --dataset ltcc --dataset_root DATASET_ROOT --dataset_filename LTCC-reID --save_dir SAVE_DIR --save_checkpoint

# PRCC
python main.py --gpu_devices 0 --dataset prcc --dataset_root DATASET_ROOT --dataset_filename PRCC --save_dir SAVE_DIR --save_checkpoint

# Celeb-reID
python main.py --gpu_devices 0 --dataset celeb --dataset_root DATASET_ROOT --dataset_filename Celeb-reID --num_instances 4 --save_dir SAVE_DIR --save_checkpoint

--dataset_root : replace DATASET_ROOT with your dataset root path

--save_dir: replace SAVE_DIR with the path to save log file and checkpoints

To facilitate reproduction and comparison, we release the trained model weights and complete training logs for the datasets. This includes our proposed method's SOTA performance models. All resources can be accessed via: Download Link

Evaluation

python main.py --gpu_devices 0 --dataset DATASET --dataset_root DATASET_ROOT --dataset_filename DATASET_FILENAME --resume RESUME_PATH --save_dir SAVE_DIR --evaluate

--dataset: replace DATASET with the dataset name

--dataset_filename: replace DATASET_FILENAME with the folder name of the dataset

--resume: replace RESUME_PATH with the path of the saved checkpoint

The above three arguments are set corresponding to Training.

The code is based on CSSC-CCReID

@inproceedings{wang2025content,
  title={Content and salient semantics collaboration for cloth-changing person re-identification},
  author={Wang, Qizao and Qian, Xuelin and Li, Bin and Chen, Lifeng and Fu, Yanwei and Xue, Xiangyang},
  booktitle={ICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={1-5},
  year={2025},
  organization={IEEE}
}