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.
| 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.
The model was trained on the horse2zebra dataset link.
To set the parameters edit the config.py file and then run the code as given in the main.ipynb notebook file.