search

HansSunY / DiffAM

[CVPR 2024] Official repository of paper "DiffAM: Diffusion-based Adversarial Makeup Transfer for Facial Privacy Protection".
32 stars 4 forks source link

edit_one_image function #2

Open zerojin0603 opened 3 months ago

zerojin0603 commented 3 months ago

Hello,

Thank you for sharing the code. I have a question regarding the edit_one_image function for makeup transfer.

Based on the instructions provided, it seems that running the command: python main.py --edit_one_image_MT --config celeba.yml --exp ./runs/test --n_iter 1 --t_0 60 --n_inv_step 20 --n_train_step 6 --n_test_step 6 --img_path {IMG_PATH} --model_path {MODEL_PATH} should generate an adversarial makeup image for the file specified in {IMG_PATH}. However, after reviewing & running the implementation, I couldn’t find where the adversarial optimization is carried out within the edit_one_image function.

Am I misunderstanding the purpose of the edit_one_image function, or is it not intended to generate an adversarial makeup image? If it’s not, could you please advise on how I might apply your method to my custom dataset?

Thank you for your assistance.

fangxuehouwuming commented 1 month ago

I tried to use the edit_one_imageto remove makeup, but it didn't work. Have you tried it?

zerojin0603 commented 1 month ago

Hello, I haven't tried edit_one_image_MR. Only tried edit_one_image_MT and it didn't work so had to do it through main.py

HansSunY commented 1 month ago

Sorry for the mistakes in the source code. They have already been fixed.

-----原始邮件----- 发件人:"Lq Wang" @.> 发送时间:2024-10-13 18:03:01 (星期日) 收件人: HansSunY/DiffAM @.> 抄送: Subscribed @.***> 主题: Re: [HansSunY/DiffAM] edit_one_image function (Issue #2)

I tried to use the edit_one_image to remove makeup, but it didn't work. Have you tried it?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>