search

aamixsh / VStegNET

Code for our research paper VStegNET
11 stars 3 forks source link

data loader problem #1

Open alighofrani95 opened 4 years ago

alighofrani95 commented 4 years ago

I'm so interested in your great paper VStegNET and try to reimplement from your repo after downloading UCF-101 and put it in the directory as explained in the readme file, I faced with the error below

"""  [] Reading checkpoints...  [] Failed to find a checkpoint  [!] Load failed... Traceback (most recent call last):   File "stego_net.py", line 627, in     m.train()   File "stego_net.py", line 354, in train     for covers, secrets, c_name, s_name, epoch, batch in generator():   File "stego_net.py", line 347, in generator     frs = sorted(os.listdir(c_path), key=lambda x: int(x.split('.')[0]))[:t self.frames_per_batch]   File "stego_net.py", line 347, in     frs = sorted(os.listdir(c_path), key=lambda x: int(x.split('.')[0]))[:t self.frames_per_batch] ValueError: invalid literal for int() with base 10: '' """

Do you use the default name of videos or rename them before loading them?

indeed if it's possible please add a requirement file to help reimplement the project with the same results...

thanks a lot

aamixsh commented 4 years ago

Please extract frames from videos, and rename every object (videos and frames of videos) with numbers. That's how it is sorted frame-wise.

alighofrani95 commented 4 years ago

thank you for your fast replying based on your guide, I extract frames from videos and now I faced with loading file error.

Could you please share the dataset loader and requirements file

aamixsh commented 4 years ago

It's all there in these set of codes. For frame extraction you can use ffmpeg. I am a bit busy so can not put up the requirement file instantly, will do that in some time.

2459122312 commented 1 year ago

thank you for your fast replying based on your guide, I extract frames from videos and now I faced with loading file error.

Could you please share the dataset loader and requirements file

Hello,have you solved this problem?

2459122312 commented 1 year ago

It's all there in these set of codes. For frame extraction you can use ffmpeg. I am a bit busy so can not put up the requirement file instantly, will do that in some time.

Could you please share the dataset loader and requirements file?Thank you .

aamixsh commented 1 year ago

Hi,

I'm really sorry about this but I don't have the requirements file. This code was written quickly in a prototype fashion without documentation and I regret it now. It was written at the start of 2019 so I think tensorflow version 1.10 or something should work. Although I don't remember using very niche functions there so most of the code should work in more recent versions too. I'm afraid if there's an error now, it will have to be fixed manually.

My best recommendation would be to just look at the model architecture in the paper and reimplement it in a current version of tensorflow or pytorch. This code can be used as template for that, as I assure you it worked when I wrote it.

I will try to find some time to fix some obvious issues here and maybe get a requirements file that works (but assume that won't work).

All the best!

2459122312 commented 1 year ago

HI, recently I have tryed my best to reimplement your results according to your readme file, but I failed. I'm so interested in your great paper VStegNET . Could you please share the dataset loader and requirements file?Thank you