What‘s input of the model when training?

[ztwe]

        FLAGS['path_data'] = '/dataset/LPGAN'
        FLAGS['path_result_root'] = '/dataset/LPGAN-Result/%03d-DGX-LPGAN'

What does LPGAN point to? How to modify it according to self environment?

Are there any other places that we should modify to make the code running?

[vuzya]

@nothinglo please, help to understand how can we actually use it? The code is mostly understandable, but I really can't deal with the part related to input(image reading, etc.)

[JustinhoCHN]

I am guessing this path too, maybe it's FiveK dataset path. Am I right? @nothinglo

[nothinglo]

Yes. it points to the datasets.

[XBOOS]

Hi nothinglo,
Thanks for your great work. I have some questions for the code data part also.
Do I have to use both the input and the corresponding retouched label?
Why is that in the code , the training sets contains input+label + input_label three parts.
`
input_img = cv2.imread(FLAGS['folder_input'] + file_name_input + FLAGS['data_input_ext'], -1)

        if os.path.exists(FLAGS['folder_label'] + file_name_label + FLAGS['data_input_ext']):   
            label_img = cv2.imread(FLAGS['folder_label'] + file_name_label + FLAGS['data_input_ext'], -1)    
        else:
            label_img = cv2.imread(FLAGS['folder_label_HDR'] + file_name_label + FLAGS['data_input_ext'], -1)
         input_label_img = cv2.imread(FLAGS['folder_label'] + file_name_input + FLAGS['data_input_ext'], -1)
        label_input_img = cv2.imread(FLAGS['folder_input'] + file_name_label + FLAGS['data_input_ext'], -1)

`

What if I dont have the corresponding pairs? (Also in the do_testing function I have to provide input and label)
Thank you !
Looking forward to your reply.

[nothinglo]

If you do not have input_label then you can just modify the code. Sorry for your inconvenience.

[XBOOS]

Thank you nothinglo! I've already modified the code. Excellent work !

[xiaozhi2015]

When train HDR models, "input_img" means source img, "label_img" means target img, what role did "input_label_img" play?
When train UL(unpaired learning) models, "input_img" means source img, "label_img" means target img,
what role did "input_label_img" play in this case?
@nothinglo @XBOOS

[JustinhoCHN]

I used the inputZeroed as input. @xiaozhi2015