Skip to content

PyTorch Implementation of unpaired img-to-img translation via SDE.

Notifications You must be signed in to change notification settings

Ishan-phys/Unpaired-via-SDE

Folders and files

NameName
Last commit message
Last commit date

Latest commit

0db16e3 · Jan 10, 2024

History

4 Commits
Aug 18, 2022
Aug 18, 2022
Aug 18, 2022
Aug 18, 2022
Jan 10, 2024
Aug 18, 2022
Aug 18, 2022
Aug 18, 2022
Aug 18, 2022
Aug 18, 2022
Aug 18, 2022
Aug 18, 2022
Aug 18, 2022

Repository files navigation

Unpaired-via-SDE

We implement a Score-based Generative Model to perform unpaired image-to-image translations from an image domain A to an image domain B and vice versa. Domain A consists of satellite images. Domain B consists of the corresponding maps style images.

Results:

1st column: Input Images / 2nd column: Domain Tranlated(CycleGANs) / 3rd column: Domain Translated(Ours)

From the above image, we observe that our model needs further improvements to perform the unpaired image-to-image translation task. This is the subject of our future work.

Dataset

The model was trained on the horse2zebra dataset link.

Training

To set the parameters edit the config.py file and then run the code as given in the main.ipynb notebook file.