benquick123 / c-vton Goto Github PK

Could you provide the version used in the dataset(mpv / viton),
cause the densepose_parse you used seems different to detectron or detectorn2
(sorry for using different input image, but it is clear that they were generated by distinctive method)

Hello sir,i liked your project idea and i want to run it on my pc but i am encountering lot of error while doing to do so.sir can please guide me or mention some steps in order to run the program in my local device.

For a custom inference do I need to generate the densepose results on my own?

Hi.
Although I tried to run the test script, the script throws errors in ./dataloader/MPVDataset.py.
There are no person images and clothing images in ./data/mpv/xxxxxx.
For instance, the clothing image does not exist when desired clothing path is ./data/mpv/BJ721D039/BJ721D039-Q11@12=cloth_front.jpg.
Please teach me how to solve this problem.
Thanks.

@ajdal @benquick123 What is the dataset structure for the viton or mpv datasets? I know they should both go into a folder called data.

But there are many files and subfolders inside of viton or mpv, I believe, so I want replicate the file structure with my own data that I'll use instead of the viton or mpv datasets. Since those are no longer public, I can't figure out how they were structured and it was challenging to parse through the code to figure that part out.

Alternatively, if you're aware of somewhere I can access that data or a subset of it, that would also be appreciated.

Hi,

Thank you for your great work! I want to clarify what VITON-HD refers to in your project. My understanding is that VITON-HD is a full resolution version of the original VITON, which you get from the authors of the work directly. Is my understanding correct?
My confusion comes from the fact there is a dataset (more like a model?) called VITON-HD. I thought VITON-HD in your project refers to this VITON-HD but it seems not.

I really appreciate your help.

Hi, I've been reading about your project for about a week now, and now I want to see it in action!
I cloned the repo, installed the requerments as specified in the 'Testing' section, downloaded all the relevant datasets to the folders as specified in the 'For MPV' sub-section.
now, Im tring to run the 'test_mpv.sh' but unfortunately Im incountering the error that attached to this issue.
I think this error implies that the images can't be found. that is reasonable since Im not specifing any specific images...
I would very much appreciate it if you can give an example of the command needs to be written in the terminal in order to run the 'test_mpv.sh' with specific images from the MPV dataset. I looked a little bit in the config.py and tried to figure which argument is relevant, but without any luck.
thank you very much for the time and attention.
--the error log is attached to the issue--
my_problem.txt

Hey folks,

Thanks for the great work!

Would you mind updating the license file?

Cheers,
Nebojsa

Hi,
I try to evaluate FID on the cleaned VITON dataset (416 images) with the generated test results that this repo provides.
I use pytorch-fid to calculate FID. However, the result is worse than reported in the paper (37.05 and 19.54). Maybe my method is wrong, could you tell me how to reproduce the fid results?

Is it possible to take my photo of clothes and my photo of a person and combine them? Or is it only possible to see test data here?

I have done the testing with pretrained model on vton dataset. It looks to me that the testing needs all the other information along with model, cloth and edge, such as densepose, masks etc. Shouldnt testing only need model image, cloth and its edge?

Hello there!

Thank you for this project, I'm looking forward to experimenting a little bit with it.
I have a question regarding the license of the dependencies. Considering that this repo is friendly to commercial use, are all the dependencies free to use for commercial projects as well? I ask specifically about the datasets and similar, which are on Sharepoint links with no license info.

Thank you so much for clarifying!

Can you kindly mention the two models that you have used to create the two folders named 'image_body_parse' and 'image_parse_with_hands' to obtain the labels for the images in the preprocessed-viton dataset.
Which models have you used to obtain the labels as shown in these two images:

How to run with my person image and my cloth

I want to know where can I get the real model images and clothes of init MPV data. I discover the link in the internet is not available.

Hi, thank you for your great work! May I know why the FID score presented on your paper (Table 1) is so much different from the one reported in their original paper? For example, the FID score reported for PF-AFN is 10.09 on VTON dataset while it's reported as 27.23 in your paper.

Can you please provide the preprocessing methods of image_body_parse and image_parse_with_hands.

i get this error when i set arg --segmention into body or cloth

I appreciate all of your hard work. There are gaps in the dataset you have provided. Would you kindly submit the entire dataset or describe how the inferencing model's dataset is organized?