columbiadvmm/autogan

autogan's Issues

Inference on StyleGAN-generated images

I'm trying to use AutoGAN to detect StyleGAN-generated facial images.

• Do you have any examples of how to run the detector in any random image?
• Are any of the models in the Google Drive pre-trained with facial images?

Thanks!

Syntax error in code/run_test.py; missing close square bracket

The file code/run_test.py has a syntax error and is missing a close square bracket:

elif args.dataset == 'AutoGAN':
    datasets = ['horse_auto', 'zebra_auto', 'summer_auto', 'winter_auto', 'apple_auto', 'orange_auto', 'facades_auto', 'cityscapes_auto', 'satellite_auto', 'ukiyoe_auto', 'vangogh_auto', 'cezanne_auto', 'monet_auto', 'photo_auto',
else:

How is this line supposed to end? For now I've just ended it after the final comma.

There is no autogan folder..

Pretrained models do not contain checkpoint file

The pretrained model obtained from

https://drive.google.com/open?id=1IZZ5QCLbzAmrPDHh7EdOigPCdz46zPc6

does not have any file that matches the path {}{}/checkpoint_{}.pth. Each of the folders has the format

/tmp/autogan_model/monet_autogan                                                                                                                                                                                                                        
/tmp/autogan_model/monet_autogan/latest_net_G.pth                                                                                                                                                                                                       
/tmp/autogan_model/monet_autogan/latest_net_D_A.pth                                                                                                                                                                                                     
/tmp/autogan_model/monet_autogan/test_opt.txt                                                                                                                                                                                                           
/tmp/autogan_model/monet_autogan/train_opt.txt                  

Are there pretrained checkpoints for running the detector?

About the place of pretrained CycleGAN

Thanks for your work.
I noticed that there are 18 pretrained model to make the dataset.
And the script to download the models have failed. So I want to know if I should put them to the "autogan/scripts" and then run the "run_test.py"?
I got the error when I try to "python run_test.py --dataset Cyclegan" :
NameErroe: "parser" is not defined

Hello, can you release your trained GAN-Detection model parameters?

Hello, can you release your trained GAN-Detection model parameters?

No such file or directory: './datasets/horse_train.pt'

Hi, I got the error when I try to train the classifier
$ python run_training.py --dataset=CycleGAN --feature=fft --gpu-id=0

Test Parameter: --feature=fft
python GAN_Detection_Train.py --training-set horse --model=resnet --test-set=transposed_conv --data_augment --batch-size=16 --test-batch-size=16 --feature=fft --gpu-id 0 --model-dir ./model_resnet/ --log-dir ./resnet_log/ --enable-logging=False --epochs 20
Wait for process end

/home/lab/miniconda3/envs/AutoGAN/lib/python3.8/site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and will be removed in 0.15, please use 'weights' instead.
warnings.warn(
/home/lab/miniconda3/envs/AutoGAN/lib/python3.8/site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or None for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing weights=ResNet34_Weights.IMAGENET1K_V1. You can also use weights=ResNet34_Weights.DEFAULT to get the most up-to-date weights.
warnings.warn(msg)

Caching data ./datasets/horse_train.pt

Traceback (most recent call last):
File "GAN_Detection_Train.py", line 434, in
train_loader, test_loaders = create_loaders()
File "GAN_Detection_Train.py", line 261, in create_loaders
GANDataset(train=True,
File "GAN_Detection_Train.py", line 169, in init
super(GANDataset, self).init(train=train, *arg, **kw)
File "/home/lab/Downloads/AutoGAN/code/cycleGAN_dataset.py", line 68, in init
self.cache_data(data_file, name, check_cached)
File "/home/lab/Downloads/AutoGAN/code/cycleGAN_dataset.py", line 103, in cache_data
with open(data_file, 'wb') as f:
FileNotFoundError: [Errno 2] No such file or directory: './datasets/horse_train.pt'

Thanks for your work,
Robert

conda environment fails

Hello,

When installing the conda env it fails (for e.g. intel package is not found, conflicts etc).

Do you have another yml file?
Maybe a docker image - or some enviroment details you have - like linux version, python version, conda version.

Adrian

Hello, can you give me the original picture of of these two face images

Hello, Thanks for your work.I want to use your sample picture to do some experiments, can you give me a copy of the original picture of of these two face images?
I want to study the difference between real and fake faces。

Q

Excuse me, will the code of your paper be published on github?
I am looking forward to it, thank you.