Decoder

[xbtzpx]

My device environment:
python=3.9.18
NVIDIA GeForce RTX 3060
CUDA Version：12.2
torch=1.11.0+cu113
steganogan=0.1.2

What I did:
—————————— code
I ran the train.py file under research
Encoders use Dense structures
The data set of afhq used is divided into 5153 cat pictures and 9892 dog pictures in training set, 500 cats and 500 dogs in verification set, and the picture size is 256*256
(PS: About the data set. I combined the pictures of cats and dogs in the data set to report an error without classification, so I divided the training set into cats and dogs according to the verification set.
—————————— decode
I created a.py file,Enter the encoded image:

from steganogan import SteganoGAN
def main():
steganogan = SteganoGAN.load(architecture='dense',verbose=True)
steganogan.encode('1/input.png', 'output.png', 'This is a super secret message!')
steganogan.decode('output.png')
if name == 'main':
main()

The result will be an error:
raise ValueError('Failed to find message.')
ValueError: Failed to find message.
Process finished with exit code 1

I need your help:
I would like to ask you if my training process is correct?
I have reviewed the README documentation, but I am not quite sure what the training process is