training error

[antithing]

Hi, and thanks for making this code available. i am trying to train it, and am getting the following error:

===> Environment init ===> Loading datasets --Valid Samples: 0 Traceback (most recent call last): File "core/train.py", line 357, in <module> main() File "core/train.py", line 337, in main train_loader = get_dataset(args) File "core/train.py", line 67, in get_dataset train_set = MatDatasetOffline(args, train_transform, normalize) File "X:\dev\ML\DEEP_MATTE\pytorch-deep-image-matting-master\core\data.py", line 99, in __init__ assert(self.cnt > 0) AssertionError

running with this cmd:

python core/train.py --crop_h=320,480,640 --crop_w=320,480,640 --size_h=320 --size_w=320 --alphaDir=X:/dev/ML/DEEP_MATTE/data/train/gt --fgDir=X:/dev/ML/DEEP_MATTE/data/train/fg --bgDir=X:/dev/ML/DEEP_MATTE/data/train/bg --imgDir=X:/dev/ML/DEEP_MATTE/data/train/image --saveDir=model/stage0_norm --batchSize=1 --nEpochs=25 --step=-1 --lr=0.00001 --wl_weight=0.5 --threads=4 --printFreq=10 --ckptSaveFreq=1 --cuda --stage=0 --pretrain=model/vgg_state_dict.pth --testFreq=1 --testImgDir=X:/dev/ML/DEEP_MATTE/data/test/image --testTrimapDir=X:/dev/ML/DEEP_MATTE/data/test/trimap --testAlphaDir=X:/dev/ML/DEEP_MATTE/data/test/gt --testResDir=result/tmp --crop_or_resize=whole --max_size=1600

what am i doing wrong here?

Thanks!