US 12,266,079 B2
Semantically accurate super-resolution generative adversarial networks
Nagita Mehr Seresht, West Pennant Hills (AU); Tristan Frizza, Barangaroo (AU); and Michael Bewley, Barangaroo (AU)
Assigned to NEARMAP AUSTRALIA PTY LTD., Barangaroo (AU)
Filed by Nearmap Australia Pty Ltd., Barangaroo (AU)
Filed on Apr. 11, 2022, as Appl. No. 17/717,824.
Claims priority of provisional application 63/173,536, filed on Apr. 12, 2021.
Prior Publication US 2022/0335572 A1, Oct. 20, 2022
Int. Cl. G06T 3/4053 (2024.01); G06N 3/04 (2023.01); G06T 3/4046 (2024.01); G06T 7/11 (2017.01)
CPC G06T 3/4053 (2013.01) [G06N 3/04 (2013.01); G06T 3/4046 (2013.01); G06T 7/11 (2017.01); G06T 2207/10032 (2013.01); G06T 2207/20084 (2013.01)] 13 Claims
OG exemplary drawing
 
1. An image processing system, comprising:
circuitry configured to
generate up-sampled images corresponding to input images based on the input images, image realism prediction values, and a corresponding semantic feature mask,
generate the image realism prediction values based on at least one of the input images and the up-sampled images,
generate the semantic feature mask based on at least one of the input images and the up-sampled images, and
detect at least a portion of the images requiring up-sampling when the portion of the images has a lower resolution than a resolution of surrounding images.