nvlabs / dg-net-pp Goto Github PK

If my training is interrupted by a power outage, how do I pick up where I left off?

/home/ubuntu/DG-Net-PP/utils.py:239: YAMLLoadWarning: calling yaml.load() without Loader=... is deprecated, as the default Loader is unsafe. Please read https://msg.pyyaml.org/load for full details. return yaml.load(stream) Traceback (most recent call last): File "train.py", line 59, in <module> config['new_size'], config['crop_image_height'], config['crop_image_width'], config['num_workers'], False) File "/home/ubuntu/DG-Net-PP/utils.py", line 219, in get_data_loader_folder dataset = ReIDFolder(input_folder, transform=transform) File "/home/ubuntu/DG-Net-PP/reIDfolder.py", line 14, in __init__ super(ReIDFolder, self).__init__(root, transform) File "/home/ubuntu/anaconda3/envs/dgnpp/lib/python3.7/site-packages/torchvision/datasets/folder.py", line 178, in __init__ target_transform=target_transform) File "/home/ubuntu/anaconda3/envs/dgnpp/lib/python3.7/site-packages/torchvision/datasets/folder.py", line 79, in __init__ "Supported extensions are: " + ",".join(extensions))) RuntimeError: Found 0 files in subfolders of: ./datasets/Market/pytorch/train_all Supported extensions are: .jpg,.jpeg,.png,.ppm,.bmp,.pgm,.tif

While I have only .jpg files, its bassically the same Market data I downloaded from the git-page link.

Hi, Thank for your great work.

What does prepare market.py do???
Is it a must do requirement to run the code?

What does the teacher model do ? Is the teacher model necessary for dgnet++? it was present in dgnet paper for soft labeling, but it does not appear to be used in dgnet++ paper in adaptation or self-training.

If I need a teacher model, would it then suffice to get a model pretrained on the target domain? and must it only be a resnet50 or would any other type of model work?

i have a huge confusion...on these 2 terms
Can you give an explanation on what do you mean by "Market2Duke" and "Duke2Market"....?

Firstly, thanks for your contribution. Currently, all google drive download links were down. Can you please help update the new link? Thank you!