US 12,272,172 B2
Method and device for human parsing
Won Jun Hwang, Seoul (KR); and Kun Liang Liu, Tianjin (CN)
Assigned to AJOU UNIVERSITY INDUSTRY-ACADEMIC COOPERATION FOUNDATION, Gyeonggi-do (KR)
Filed by AJOU UNIVERSITY INDUSTRY-ACADEMIC COOPERATION FOUNDATION, Gyeonggi-do (KR)
Filed on Jun. 13, 2022, as Appl. No. 17/839,443.
Claims priority of application No. 10-2021-0076804 (KR), filed on Jun. 14, 2021.
Prior Publication US 2022/0398865 A1, Dec. 15, 2022
Int. Cl. G06V 40/16 (2022.01); G06T 7/246 (2017.01); G06V 10/77 (2022.01)
CPC G06V 40/166 (2022.01) [G06T 7/246 (2017.01); G06V 10/7715 (2022.01); G06T 2207/30196 (2013.01)] 18 Claims
OG exemplary drawing
 
1. A method for human parsing, the method comprising:
receiving, by a controller, at least one piece of image data and ground truth values for human parsing;
generating, by the controller, a height distribution map and a width distribution map for the image data;
acquiring, by the controller, attention maps and scaled feature maps for each of a height and a width of the image data using the distribution maps;
concatenating, by the controller, the scaled feature maps and calculating a distribution loss rates for each of the height and the width;
acquiring, by the controller, an improved feature map on the basis of the calculated distribution loss rates; and
performing, by the controller, human parsing on an object included in the image data using the improved feature map.