search

IricsDo / Data-Hidding-Using-GAN

This is my graduation essay. You can do everything you want
3 stars 0 forks source link

A question about robustness #3

Open ybL-123 opened 4 months ago

ybL-123 commented 4 months ago

Hello! Your project is excellent. I am very interested in it. Could you please tell me if there is any research related to robustness in your project? My own other research does not seem to have achieved results in terms of robustness. Thank you.

IricsDo commented 4 months ago

Sorry for response too late. Currently, I don't know. I abandoned this project a long time ago and haven't updated it at all.

ybL-123 commented 4 months ago

Dear DoDuy,

I hope this message finds you well.

I am writing to seek your assistance regarding some issues I have encountered while working with the encryption and decryption of images as described in your paper/project.

Firstly, I am experiencing difficulties with decrypting the encrypted images. Despite following the steps outlined in your work, I am unable to resolve the issue. 1

Additionally, when converting the image to text, I encountered encoding error. In an attempt to fix this, I modified the encoding to UTF-8. However, this led to errors during the text-to-image decoding process. 2

the requirements i had install is:

Pillow==8.2.0 cryptography==3.4.7 imageio==2.9.0

torch==1.0.0

torchvision==0.2.1

tqdm==4.59.0 reedsolo==0.3 opencv-python==4.5.1.48 opencv-contrib-python==4.5.1.48 steganogan 0.1.2 gunicorn==20.0.4 streamlit 1.36.0 and my device OS is windows 11 CPU : i5 12500-h GPU : 3050 ti laptop

I would greatly appreciate your guidance on how to address these issues. Your assistance would be invaluable in helping me move forward with my work.

Thank you very much for your time and support.

Best regards,

IricsDo commented 4 months ago

The first error occurs when you use the incorrect PyTorch library. You must install the library following the requirement file I pushed in this repository. I know the library is too old but if you want to get it running, please follow my instructions. If not, you must manually update to the latest version. #torch==1.0.0 #torchvision==0.2.1 uncomment this line

And you can follow my question 3 years ago in here: https://discuss.pytorch.org/t/some-difference-between-pytorch-version-1-0-0-and-pytorch-version-1-7-1/122210

IricsDo commented 4 months ago

The second error, you can see your image shape is 1356×2040 = 2766240 pixels (1 pixel ~ 1 bytes). The result you see in the terminal after decode is 2755197 bytes. That means something is wrong when we encode <-> decode the images (it must be the same size).

For now, the reason is the model we selected. This model can not handle the larger information (I mean the big image size or the long text). So, you have two ways to try again :

IricsDo commented 4 months ago

Please let me know if it can help you solve the issues.

Thank you