US 11,861,762 B2
Generating synthesized digital images utilizing class-specific machine-learning models
Yuheng Li, Davis, CA (US); Yijun Li, Seattle, WA (US); Jingwan Lu, Santa Clara, CA (US); Elya Shechtman, Seattle, WA (US); and Krishna Kumar Singh, San Jose, CA (US)
Assigned to Adobe Inc., San Jose, CA (US)
Filed by Adobe Inc., San Jose, CA (US)
Filed on Aug. 12, 2021, as Appl. No. 17/400,474.
Prior Publication US 2023/0051749 A1, Feb. 16, 2023
Int. Cl. G06T 11/00 (2006.01)
CPC G06T 11/00 (2013.01) [G06T 2210/12 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A non-transitory computer readable storage medium comprising instructions that, when executed by at least one processor, cause a computing device to:
generate a synthesized digital image comprising one or more objects by utilizing an image synthesis neural network;
determine one or more classes associated with the one or more objects of the synthesized digital image;
select one or more class-specific generator neural networks based on the one or more classes associated with the one or more objects;
crop the synthesized digital image to one or more bounding boxes corresponding to the one or more objects;
crop a semantic label map associated with the synthesized digital image to one or more regions corresponding to the one or more objects;
generate, utilizing the one or more bounding boxes and the one or more regions, one or more synthesized objects by utilizing the one or more class-specific generator neural networks according to the one or more classes associated with the one or more objects; and
replace the one or more objects in the synthesized digital image with one or more synthesized objects.