vinthony / ghost-free-shadow-removal

Traceback (most recent call last):
File "train_sr.py", line 52, in
shadow_free_image,predicted_mask=build_aggasatt_joint(input,channel,vgg_19_path=vgg_19_path)
File "/public/wuwenxuan/projects/ghost-free-shadow-removal-master/networks.py", line 143, in build_aggasatt_joint
vgg19_features=build_vgg19(input[:,:,:,0:3]*255.0,vgg_19_path)
File "/public/wuwenxuan/projects/ghost-free-shadow-removal-master/networks.py", line 78, in build_vgg19
vgg_path=scipy.io.loadmat(vgg_path)
File "/public/wuwenxuan/anaconda3/envs/shadow/lib/python3.6/site-packages/scipy/io/matlab/mio.py", line 222, in loadmat
with _open_file_context(file_name, appendmat) as f:
File "/public/wuwenxuan/anaconda3/envs/shadow/lib/python3.6/contextlib.py", line 81, in enter
return next(self.gen)
File "/public/wuwenxuan/anaconda3/envs/shadow/lib/python3.6/site-packages/scipy/io/matlab/mio.py", line 17, in _open_file_context
f, opened = _open_file(file_like, appendmat, mode)
File "/public/wuwenxuan/anaconda3/envs/shadow/lib/python3.6/site-packages/scipy/io/matlab/mio.py", line 39, in _open_file
return open(file_like, mode), True
TypeError: expected str, bytes or os.PathLike object, not dict

Hi, thanks for your attention to our work.

This error often occurs when the path of VGG19 is not correct.

Please config it in the right place:

ghost-free-shadow-removal/train_sr.py

Line 34 in 0707b0a

vgg_19_path = scipy.io.loadmat('./Models/imagenet-vgg-verydeep-19.mat')

Or you can try to run the demo from the command line via the introduction here:
https://github.com/vinthony/ghost-free-shadow-removal#3-demo-from-command-line-thanks-aliericcantona

error in train_sr.py when testing