assert os.path.isdir(root) AssertionError

[zhaoqiguo]

Hello, LuckyDC
When I run train.py as your readme mentioned. The AssertionError is raised both in softmax and softmax-triplet loss:
CUDA_VISIBLE_DEVICES=0 python3 train.py --cfg configs/softmax-triplet.yml
{'batch_size': 128,
'classification': True,
'color_jitter': False,
'data_root': '/home/chuanchen_luo/data/SYSU',
'dataset': 'sysu',
'drop_last_stride': True,
'dual_path': False,
'eval_interval': 50,
'fp16': True,
'image_size': (256, 128),
'k_size': 8,
'log_period': 10,
'lr': 0.00035,
'lr_step': [150],
'num_cam': 6,
'num_epoch': 200,
'num_id': 395,
'optimizer': 'adam',
'p_size': 16,
'padding': 10,
'prefix': 'sysu-nodual-adam-16x8-128',
'random_crop': True,
'random_erase': True,
'random_flip': True,
'sample_method': 'identity_uniform',
'triplet': True,
'wd': 0.0005}
Traceback (most recent call last):
File "train.py", line 134, in
train(cfg)
File "train.py", line 46, in train
num_workers=8)
File "/media/zbx01/0cdb9b67-bb97-4984-b104-f1bbe012e4fd/guozhaoqi/Gitcode/RGB-IR-ReID-Baseline-master/data/init.py", line 42, in get_train_loader
train_dataset = SYSUDataset(root, mode='train', transform=transform)
File "/media/zbx01/0cdb9b67-bb97-4984-b104-f1bbe012e4fd/guozhaoqi/Gitcode/RGB-IR-ReID-Baseline-master/data/dataset.py", line 15, in init
assert os.path.isdir(root)
AssertionError
My torch vision is 1.3.1, python vision is 3.7.4, cuda vision is 9.0
How could I fix it? Thx

[LuckyDC]

*.yml in ./configs directory define the setting of the experiment. In this case, you should append data_root entry to the file to override the default setting. That is, append data_root: YOUR_DATASET_PATH to configs/softmax-triplet.yml or configs/softmax.yml, where YOUR_DATASET_PATH denotes the path of the dataset in your computer