ModuleNotFoundError: No module named 'vgg_loss'

你好，我安装了kornia但是还是出现这个错误，你使用的kornia的版本是多少呢？这个源代码可以跑通么？

I met a error :
File "/root/miniconda3/envs/pytorch1.12/lib/python3.8/site-packages/PIL/Image.py", line 2142, in save
if format.upper() not in SAVE:
AttributeError: 'int' object has no attribute 'upper'
and i solved it by changing the version of torch from 0.12.1 to 0.5.0 .
I don't know how to explain it. I submitted this issue to help anyone who may encounter similar problems as me.

您好，我在使用您的代码的时候，使用您提供的噪声参数格式 help="Noise layers configuration. Use quotes when specifying configuration, e.g. 'cropout((0.55, 0.6), (0.55,0.66)) 噪声参数noise_config 一直没有值。

The command you gave for training has a spelling error: 'mian' should be changed to 'main'

麻烦问一下，args一直为空，配置参数的默认值一直读不到该怎么解决？

你好，方便提供options-and-config.pickle和ARWGAN.pyt文件吗::>_<::

We did not train a corresponding model according to the experimental settings in the paper (the effect was very poor). Excuse me for the impertinence, could you please explain how you trained the combined noise model (i.e., the combined noise model used during testing with Crop, Crpout, Dropout, Gaussian blur, and JPEG compression)? Could you provide training details?

noise参数怎么设置，就是-n后面的？

File "/root/ARWGAN-main/model/ARWGAN.py", line 7, in
from model.encoder_decoder import EncoderDecoder
ModuleNotFoundError: No module named 'model.encoder_decoder'

作者您好，我想咨询一下您。请问模型训练batchsize为多少合适呢、epoch=100是否合适、我尝试batchsize=32时，模型在epoch=7后loss在下降到0.55左右，后续模型训练就不再收敛的原因。
Uploading 问题描述.docx…

image文件相关保存的操作被注释掉了 ，将注释的操作取消注释后似乎无法正常输出图片，这是报错：
Traceback (most recent call last):
File "G:\ARWGAN-main\main.py", line 148, in
main()
File "G:\ARWGAN-main\main.py", line 144, in main
train(model, device, net_config, train_options, this_run_folder, tb_logger)
File "G:\ARWGAN-main\train.py", line 75, in train
utils.save_images(image.cpu()[:images_to_save, :, :, :],
File "G:\ARWGAN-main\utils.py", line 62, in save_images
torchvision.utils.save_image(stacked_images, filename, format= original_images.shape[0], normalize=False)
File "E:\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
return func(*args, **kwargs)
File "E:\lib\site-packages\torchvision\utils.py", line 151, in save_image
im.save(fp, format=format)
File "C:\Users\tudou\AppData\Roaming\Python\Python310\site-packages\PIL\Image.py", line 2335, in save
if format.upper() not in SAVE:
AttributeError: 'int' object has no attribute 'upper'
请问是否方便分享可跑通的代码呢？

我使用的数据集是COCO_val2014

Why is the watermark not robust after adding noise?

ef jpeg_compress_decompress(image, downsample_c=False, rounding=diff_round, factor=1.,device=True):

todo 是01区间，而不是-1至1区间

image=(image+1)/2改为

image*=255
b, c, h, w = image.size()