suetandtie / cycle-image-gan Goto Github PK

Hi,

when running with the command python main.py --cfg cfg/bird_cycle.yaml --gpu 0 the function crashes on PIL_att = Image.fromarray(np.uint8(one_map)).

This is the error:

 File "/Data/anaconda3/envs/cyclegan-py3/lib/python3.6/site-packages/PIL/Image.py", line 2772, in fromarray
    mode, rawmode = _fromarray_typemap[typekey]
KeyError: ((1, 1, 6), '|u1')

I guess is a problem of dimensions, one_map ha dimension (128, 128, 6) while img has (128, 128, 3).

Any ideas how to solve it?
Thanks

super(_open_file, self).__init__(open(name, mode))

FileNotFoundError: [Errno 2] No such file or directory: 'output/birds_STREAM_2019_06_07_18_55_55/Model/image_encoder100.pth'

Hi,

Your YAML file refer to pth file which I assume is the trained model. Where do we get that?

attn = attn_maps[i].cpu().view(1, -1, att_sze, att_sze)
IndexError: index 2 is out of bounds for dimension 0 with size 2

Please help how can I resolve this issue?

how to solve
“Model name 'bert-base-uncased' was not found in model name list (bert-base-uncased, bert-large-uncased, bert-base-cased, bert-large-cased, bert-base-multilingual-uncased, bert-base-multilingual-cased, bert-base-chinese). We assumed 'https://s3.amazonaws.com/models.huggingface.co/bert/bert-base-uncased-vocab.txt' was a path or url but couldn't find any file associated to this path or url.”

Hi guys,

I think depending on your version of scikit-image the function pyramid_expand might require the flag multichannel

Hence:

                one_map = \
                    skimage.transform.pyramid_expand(one_map, sigma=20,
                                                     upscale=vis_size // att_sze)

Should become:

                one_map = \
                    skimage.transform.pyramid_expand(one_map, sigma=20,
                                                     upscale=vis_size // att_sze,
                                                     multichannel=True)

Hope this can be of some help to anyone that encounter an issue related to expanding all the channels.

Best wishes,

I was having the problem of training 11788 images in google colab (I had GPU limitations). So I decided to run the same code using 10 species instead of 200 which resulted in 543 images after making necessary changes in datasets.

After running the pretrain STREAM.py, I got image encoder.pth and text encoder.pth.
Then the problem occured after running the second command, "python main.py --cfg cfg/bird_cycle.yaml --gpu 0"

Traceback (most recent call last):
File "/content/drive/MyDrive/BPI_V2 - Copy/cycle-image-gan-master/main.py", line 141, in
algo.train()
File "/content/drive/MyDrive/BPI_V2 - Copy/cycle-image-gan-master/trainer.py", line 701, in train
errD = discriminator_loss(netsD[i], imgs[i], fake_imgs[i],
File "/content/drive/MyDrive/BPI_V2 - Copy/cycle-image-gan-master/miscc/losses.py", line 153, in discriminator_loss
cond_real_logits = netD.COND_DNET(real_features, conditions)
File "/usr/local/lib/python3.9/dist-packages/torch/nn/modules/module.py", line 1194, in _call_impl
return forward_call(*input, **kwargs)
File "/content/drive/MyDrive/BPI_V2 - Copy/cycle-image-gan-master/model.py", line 829, in forward
h_c_code = torch.cat((h_code, c_code), 1)
RuntimeError: Sizes of tensors must match except in dimension 1. Expected size 18 but got size 4 for tensor number 1 in the list.

** I printed the size of h_code and c_code
h_code.shape = torch.Size([20, 512, 18, 18])
c_code.shape = torch.Size([20, 768, 4, 4])

Hi there thanks for the great work!
When I was trying to run
python pretrain_STREAM.py --cfg cfg/STREAM/bird.yaml --gpu 0
on colab, I got this error:

TypeError: Cannot handle this data type: (1, 1, 48), |u1

I guess that might be an environmental issue, would you mind share your environment requirement for this implementation?