search

affromero / SMILE

SMILE: Semantically-guided Multi-attribute Image and Layout Editing, ICCV Workshops 2021.
https://affromero.github.io/SMILE/
MIT License
34 stars 1 forks source link

Training scripts #5

Open lijiaxinxin opened 1 year ago

lijiaxinxin commented 1 year ago

Hello, I appreciate your great work!However, I noticed that the training scripts for the models are not provided in the repository. Would it be possible for you to share the training scripts used to train the models? This would be very helpful for me and other users who are interested in experimenting with the models. Thank you in advance for your time and consideration.

Thanks, Best regards.

affromero commented 1 year ago

Hello, thank you for your interest.

That is a very good point.

If I remember correctly, the script used to train the semantic model is the one provided in the demo link, by modifying it with mode=train. You can also see more ways to use the training script here: https://github.com/affromero/SMILE/blob/master/demo.py#L479-L497

I deeply apologize on behalf of my past self, I used to write horrible code :sweat_smile:.

lijiaxinxin commented 1 year ago

Thank you very much for your reply! But I still have some questions about the dataset during traning. The code will read images from ./data/CelebA_HQ/celeba-256. I have downloaded the data from https://www.kaggle.com/datasets/lamsimon/celebahq and stored it in ./data/celeba_hq. May I ask if I need to deal with CelebA_HQ? Or could you please tell me how to download the corresponding dataset. image

affromero commented 1 year ago

Hello, As I do not remember exactly from where I downloaded CelebAHQ, the txt_file to read the labels might differ. All you need to make sure is that self.labels contains the labels for all the classes, as you will need a subset of them (eg., eyeglasses, hat, etc) for training. Hope that helps.