This is an implimented project with the reference of original paper unpaired image to image translation using cycle conistance adversarial loss. you can also check my project report here
Collect the data set from kaggle
| class | Domain-A training data | Domain-B training Data |
|---|---|---|
| AerialToSatellite |  |
 |
| VangoghToPhoto |  |
 |
| SummerToWinter |  |
 |
The aim of this project is to convert image in domain A to domain B and then back to domain A.This translation is puerly in presence of unpaired dataset.Though the results generated by the pix2pix were good,dataset is not widley available(because model translates in presence of paired datasets)
| class | sample O/P-1 (domain x to y and back to domain x) | sample O/P-2 (domain x to y and back to domain x) |
|---|---|---|
| AerialToSatellite |  |
 |
| VangoghToPhoto |  |
 |
| SummerToWinter |  |
 |
CycleGAN is performing well in all cases but the resolution of generated images is little bit low(due to loss in translation),good research is still needed to improve the quality.