zheng222 / ppon Goto Github PK

Hi,
The results look good.
How do I use your code on my images?
Do I need to train?
Please provide simple instructions.
BZW: What about JPG artifacts?
Thanks

Hello!!
I think the idea of your paper is cool!!
So, I downloaded the checkpoint and tested the set14, b100, and cg images, but the LPIPS results of the paper are a little different.
I wonder if the result of the paper is different from the ckpt and Google Drive.

Hello, i'm trying to run your code from latest commit
But there is an attributeError

Namespace(cuda=True, isHR=True, models='ckpt/PPON_G.pth', only_y=True, output_folder='result/Set5/', test_hr_folder='Test_Datasets/Set5/', test_lr_folder='Test_Datasets/Set5_LR/', upscale_factor=4, which_model_G='ppon')
Traceback (most recent call last):
File "test_PPON.py", line 50, in
model = networks.define_G(opt)
File "/home/fw/Upscale/PPON/models/networks.py", line 76, in define_G
if args.which_model == 'content':
AttributeError: 'Namespace' object has no attribute 'which_model'

Can you help me fix that? Thanks!

can you upload training code？
thank you！

Not an issue, but out of curiosity, why do the images have to be renamed “ 00xxxx.png ” and converted to npy? Is there a performance benefit?

Thank you for your sharing. But I don't understand the variable '--test_every', would you explain it? Thank you!

Hi, Please add the requirements.txt for reproducing the results.

I just wanted to make sure that there is no loading of the PPON_D.pth model for training?

Also I am assuming that the Final Stage1 Model should be loaded into Stage 2, and stage 2 for Stage 3. This was not mentioned.

Thanks for the help.

Hello, could you please provide your training model? I want to test it on my dataset!!please!