US 12,373,923 B2
Semantically-aware image extrapolation
Kuldeep Kulkarni, Karnataka (IN); Soumya Dash, Orissa (IN); Hrituraj Singh, Uttar Pradesh (IN); Bholeshwar Khurana, Rajasthan (IN); Aniruddha Mahapatra, West Bengal (IN); and Abhishek Bhatia, Punjab (IN)
Assigned to Adobe Inc., San Jose, CA (US)
Filed by Adobe Inc., San Jose, CA (US)
Filed on Jun. 7, 2024, as Appl. No. 18/737,344.
Application 18/737,344 is a continuation of application No. 17/521,503, filed on Nov. 8, 2021, granted, now 12,020,403.
Prior Publication US 2024/0331102 A1, Oct. 3, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G06T 5/50 (2006.01); G06T 7/181 (2017.01); G06V 20/70 (2022.01); G06N 3/045 (2023.01); G06T 7/11 (2017.01); G06V 10/26 (2022.01); G06V 10/82 (2022.01)
CPC G06T 5/50 (2013.01) [G06T 7/181 (2017.01); G06V 20/70 (2022.01); G06N 3/045 (2023.01); G06T 7/11 (2017.01); G06V 10/26 (2022.01); G06V 10/82 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
generating, using a peripheral object generation network, an extrapolated semantic label map for an extrapolated image corresponding to an input image;
generating, using a generator network and the extrapolated semantic label map, a panoptic label map of a plurality of predicted objects in the extrapolated image;
synthesizing the extrapolated image based on the extrapolated semantic label map and the panoptic label map, the extrapolated image comprising an outpainted region including at least one of the plurality of predicted objects; and
rendering the extrapolated image.