US 11,709,915 B2
Classifying images utilizing generative-discriminative feature representations
Pramuditha Perera, Baltimore, MD (US); Vlad Morariu, Potomac, MD (US); Rajiv Jain, Vienna, VA (US); Varun Manjunatha, College Park, MD (US); and Curtis Wigington, San Jose, CA (US)
Assigned to Adobe Inc., San Jose, CA (US)
Filed by Adobe Inc., San Jose, CA (US)
Filed on Aug. 26, 2020, as Appl. No. 17/3,149.
Prior Publication US 2022/0067449 A1, Mar. 3, 2022
Int. Cl. G06F 18/21 (2023.01); G06F 18/214 (2023.01); G06F 18/241 (2023.01); G06F 11/32 (2006.01); G06N 3/045 (2023.01)
CPC G06F 18/2185 (2023.01) [G06F 11/327 (2013.01); G06F 18/214 (2023.01); G06F 18/241 (2023.01); G06N 3/045 (2023.01)] 20 Claims
OG exemplary drawing
 
1. A non-transitory computer-readable medium storing which, when executed by at least one processing device, cause the at least one processing device to perform operations comprising:
encoding an input image into a latent space to generate a first encoding;
generating a reconstructed image by decoding the first encoding to regenerate the input image;
generating an input combination of the reconstructed image and the input image by concatenating the reconstructed image and the input image; and
determining, by processing the input combination utilizing a classification model, whether the input image depicts content corresponding to a known classification or unknown content corresponding to an open-set category based on a disparity between the reconstructed image and the input image.