	US 12,277,676 B2
	Image processing method and apparatus based on machine learning
Seung Yong Lee, Pohang-si (KR); Sung Hyun Cho, Pohang-si (KR); and Hyeong Seok Son, Pohang-si (KR)
Assigned to POSTECH RESEARCH AND BUSINESS DEVELOPMENT FOUNDATION, Pohand-si (KR)
Appl. No. 17/770,993
Filed by POSTECH RESEARCH AND BUSINESS DEVELOPMENT FOUNDATION, Pohang-si (KR)
PCT Filed Nov. 11, 2020, PCT No. PCT/KR2020/015722 § 371(c)(1), (2) Date Apr. 21, 2022, PCT Pub. No. WO2021/137415, PCT Pub. Date Jul. 8, 2021.
Claims priority of application No. 10-2019-0178401 (KR), filed on Dec. 30, 2019.
Prior Publication US 2022/0366539 A1, Nov. 17, 2022
Int. Cl. G06K 9/00 (2022.01); G06N 3/045 (2023.01); G06N 3/0455 (2023.01); G06N 3/0464 (2023.01); G06N 3/0475 (2023.01); G06N 3/08 (2023.01); G06N 3/088 (2023.01); G06T 3/4046 (2024.01); G06T 5/00 (2006.01); G06T 5/50 (2006.01); G06T 5/60 (2024.01); G06T 5/90 (2024.01); G06V 10/44 (2022.01); G06V 10/778 (2022.01); G06V 10/82 (2022.01); G06V 10/98 (2022.01); G06N 3/09 (2023.01)

CPC G06T 5/00 (2013.01) [G06N 3/045 (2023.01); G06N 3/0455 (2023.01); G06N 3/0464 (2023.01); G06N 3/0475 (2023.01); G06N 3/08 (2013.01); G06N 3/088 (2013.01); G06T 3/4046 (2013.01); G06T 5/50 (2013.01); G06T 5/60 (2024.01); G06T 5/90 (2024.01); G06V 10/454 (2022.01); G06V 10/7796 (2022.01); G06V 10/82 (2022.01); G06V 10/993 (2022.01); G06N 3/09 (2023.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01)]

12 Claims

1. A method of image processing based on machine learning, the method comprising:

generating a first correction image by a first convolution neural network receiving an input image;

generating an intermediate image by processing the input image to change a color of at least one pixel of the input image;

generating a first loss function based on the first correction image and the intermediate image;

generating a second loss function based on output from a second convolution neural network receiving the first correction image and an original image;

generating a second correction image by a third convolution neural network receiving the first correction image;

generating a third loss function based on the second correction image and the input image; and

performing machine learning of the first convolution neural network and the second convolution neural network based on the first loss function, the second loss function, and the third loss function.