To Run The Code

1 Configure the environment according to the requirements in the configuration file.
To learn more detailed information, please refer to the requirement.txt file.

2 Enter the virtual environment and import the COCO dataset.

3 Download DEKR model and put it into main folder

4 Execute the following command for training.
python -m torch.distributed.run --nproc_per_node=(Your GPU Number!!!) --master_port 29579 main.py -c config/grouppose.py --coco_path <path/to/coco/> --output_dir <path/to/output>

5 Execute the following command for testing.
python -m torch.distributed.run --nproc_per_node=(Your GPU Number!!!) --master_port 29579 main.py -c config/grouppose.py --backbone resnet50 --coco_path <path/to/coco> --output_dir <path/to/output> --resume <checkpoint> --eval

(Note: the dataset path and output result path should be specified by the user.)