US 12,136,189 B2
Media enhancement using discriminative and generative models with feedback
Akhilesh Kumar, San Jose, CA (US); Zhe Lin, Bellevue, WA (US); and Baldo Faieta, San Francisco, CA (US)
Assigned to ADOBE INC., San Jose, CA (US)
Filed by ADOBE INC., San Jose, CA (US)
Filed on Feb. 10, 2021, as Appl. No. 17/172,744.
Prior Publication US 2022/0253990 A1, Aug. 11, 2022
Int. Cl. G06T 7/00 (2017.01); G06F 18/214 (2023.01); G06F 18/2411 (2023.01); G06N 3/04 (2023.01); G06T 5/20 (2006.01)
CPC G06T 5/20 (2013.01) [G06F 18/214 (2023.01); G06F 18/2411 (2023.01); G06N 3/04 (2013.01); G06T 7/0002 (2013.01)] 19 Claims
OG exemplary drawing
 
1. A method comprising:
receiving a media object comprising an image or a video;
performing an enhancement on the media object using an enhancement network, wherein the enhancement network comprises a generative neural network of a generative adversarial network (GAN); and
generating an enhancement score using a discriminator network of the GAN, wherein the enhancement score quantifies the enhancement performed by the enhancement network, and wherein the discriminator network is trained to produce an output based on a sigmoid activation function that approximates a linear function over a majority of a training set.