US 12,148,119 B2
Utilizing a generative neural network to interactively create and modify digital images based on natural language feedback
Ruiyi Zhang, San Jose, CA (US); Yufan Zhou, Buffalo, NY (US); Christopher Tensmeyer, Columbia, MD (US); Jiuxiang Gu, College Park, MD (US); Tong Yu, San Jose, CA (US); and Tong Sun, San Jose, CA (US)
Assigned to Adobe Inc., San Jose, CA (US)
Filed by Adobe Inc., San Jose, CA (US)
Filed on Jan. 14, 2022, as Appl. No. 17/576,091.
Prior Publication US 2023/0230198 A1, Jul. 20, 2023
Int. Cl. G06T 5/00 (2024.01); G06N 3/04 (2023.01); G06T 3/10 (2024.01); G06T 5/60 (2024.01); G06T 11/00 (2006.01); G06T 11/80 (2006.01); G10L 15/22 (2006.01); G10L 15/26 (2006.01)
CPC G06T 3/10 (2024.01) [G06N 3/04 (2013.01); G06T 11/00 (2013.01); G10L 15/22 (2013.01); G10L 15/26 (2013.01); G10L 2015/223 (2013.01)] 20 Claims
OG exemplary drawing
 
16. A computer-implemented method comprising:
receiving a natural language command indicating one or more targeted image modifications for a digital image previously generated utilizing a style vector;
generating, utilizing a text encoder, an additional textual feature vector from the natural language command;
generating a modified style vector by determining a set of elements of the style vector to modify based on the additional textual feature vector; and
generating, utilizing a generative neural network, a modified digital image with the one or more targeted image modifications according to the modified style vector.