search

NicoRahm / CGvsPhoto

Computer Graphics vs Real Photographic Images : A Deep-learning approach
MIT License
25 stars 14 forks source link

hi, I try to run CgvsPhoto, but I got an error #2

Closed NemoCpp closed 6 years ago

NemoCpp commented 6 years ago

Restore weight from previous session ? (y/N) : N train ... Traceback (most recent call last): File "test_pipeline.py", line 18, in nb_validation_batch = 40, File "/home/ubuntu/anaconda2/envs/pyannote/lib/python3.5/site-packages/CGvsPhoto/model.py", line 632, in train show_filters = show_filters) File "/home/ubuntu/anaconda2/envs/pyannote/lib/python3.5/site-packages/CGvsPhoto/model.py", line 496, in validation_testing random_rotate = True) File "/home/ubuntu/anaconda2/envs/pyannote/lib/python3.5/site-packages/CGvsPhoto/image_loader.py", line 405, in get_batch_validation data = self.get_next_validation(crop, self.rand_crop, random_flip_flop,random_rotate, verbose=False) File "/home/ubuntu/anaconda2/envs/pyannote/lib/python3.5/site-packages/CGvsPhoto/image_loader.py", line 389, in get_next_validation image = image.reshape(self.size, self.size, self.nb_channels) ValueError: cannot reshape array of size 30000 into shape (100,100,1)

NicoRahm commented 6 years ago

Hi, thank you for reporting this issue. Did you install the package directly with pip? Because I think it was not up-to-date. If you try to update the package now it should work better. If not please let me know what is the dimension and type of images you are using (I only tried with JPEG so far) and if to what value you set the parameter 'only_green' in the Model definition.