Algorithm Comparison and Evaluation of GAN Models Based on Image Transferring from Desert to Green Field

Table 1

List of selected methods, including model name, publication year, and the type of training data, whether multimodal or not and corresponding insights.

Method	Publication	Data	Multimodal	Insights
CycleGAN	2017	Unpaired	No	Cyclic loss
GcGAN	2019	Unpaired	No	One-sided UI2I; geometric transformation preservation
CUT	2020	Unpaired	No	One-sided UI2I; contrastive learning
FastCUT	2020	Unpaired	No	One-sided UI2I; contrastive learning
DCLGAN	2021	Unpaired	No	Two-sided UI2I; contrastive learning
SimDCL	2021	Unpaired	No	Two-sided UI2I; contrastive learning
LSeSim	2021	Unpaired	No	One-sided UI2I; self-similarity