Skip to main content
. 2022 Feb 13;22(4):1435. doi: 10.3390/s22041435

Figure 2.

Figure 2

The comparison of the architectures of (a) HRGAN and (b) SRGAN. When the models are trained with the CIFAR-10 dataset, H and W become 32, signifying that the maximum resolution of generated images in SRGAN is 32 × 32 while HRGAN can produce 64 × 64 and 128 × 128 resolutions of images.