This work conceptualizes a deep learning model - "S-(Pro)-CycleGAN" to solve the SAR-Optical Image Translation problem . This model is a Supervised CycleGAN with a U-Net architecture using techniques like equalized learning rate and layer normalization introduced in the ProGAN paper to smoothen the training process, created in an attempt to build on the state-of-the-art.
Kindly refer to the paper uploaded to understand the model proposed.