US 12,340,484 B2
Techniques to use a neural network to expand an image
Guilin Liu, San Jose, CA (US); Andrew Tao, Los Altos, CA (US); Bryan Christopher Catanzaro, Los Altos Hills, CA (US); Ting-Chun Wang, Santa Clara, CA (US); Zhiding Yu, Santa Clara, CA (US); Shiqiu Liu, Santa Clara, CA (US); Fitsum Reda, Santa Clara, CA (US); Karan Sapra, Santa Clara, CA (US); and Brandon Rowlett, Cedar Park, TX (US)
Assigned to NVIDIA Corporation, Santa Clara, CA (US)
Filed by NVIDIA Corporation, Santa Clara, CA (US)
Filed on Mar. 9, 2020, as Appl. No. 16/813,589.
Prior Publication US 2021/0279841 A1, Sep. 9, 2021
Int. Cl. G06T 3/00 (2024.01); G06N 3/08 (2023.01); G06T 3/4038 (2024.01); G06T 3/4046 (2024.01); G06T 7/40 (2017.01); G06V 10/44 (2022.01); G06V 10/54 (2022.01); G06V 10/776 (2022.01); G06V 10/82 (2022.01)
CPC G06T 3/4038 (2013.01) [G06N 3/08 (2013.01); G06T 3/4046 (2013.01); G06T 7/40 (2013.01); G06V 10/454 (2022.01); G06V 10/54 (2022.01); G06V 10/776 (2022.01); G06V 10/82 (2022.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01)] 4 Claims
OG exemplary drawing
 
1. A processor comprising:
one or more circuits to use one or more neural networks to generate a second image based, at least in part, on one or more feature maps corresponding to a first image, wherein the first image is smaller than the second image; wherein:
the one or more feature maps are generated by the one or more neural networks from the first image;
one or more shifted feature maps are generated from the one or more feature maps;
one or more weights are computed based, at least in part, on features shared between the one or more shifted feature maps and the one or more feature maps;
one or more combined feature maps are generated based, at least in part, on combining the one or more shifted feature maps and the one or more feature maps according to the one or more weights; and
the second image is generated by aggregating and upsampling the one or more combined feature maps.