numpy, torch, torchvision
six, h5py, Pillow, scipy,
scikit-learn, metric-learn,
faiss_gpu
Market-1501, MSMT17
CUDA_VISIBLE_DEVICES=0,1,2,3 python examples/RTMem_train_usl.py -b 256 -a resnet50 -d market1501 --iters 200 --eps 0.5 --num-instances 16
Thanks to ClusterContrast and SpCL for the opening source