US 12,136,235 B2
Human model recovery using deep learning techniques
Meng Zheng, Cambridge, MA (US); Srikrishna Karanam, Bangalore (IN); and Ziyan Wu, Lexington, MA (US)
Assigned to Shanghai United Imaging Intelligence Co., Ltd., Shanghai (CN)
Filed by Shanghai United Imaging Intelligence Co., Ltd., Shanghai (CN)
Filed on Dec. 22, 2021, as Appl. No. 17/559,364.
Prior Publication US 2023/0196617 A1, Jun. 22, 2023
Int. Cl. G06T 7/73 (2017.01); G06N 3/045 (2023.01); G06T 7/50 (2017.01); G06V 40/10 (2022.01)
CPC G06T 7/75 (2017.01) [G06N 3/045 (2023.01); G06T 7/50 (2017.01); G06V 40/10 (2022.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01)] 16 Claims
OG exemplary drawing
 
1. An apparatus, comprising:
one or more processors configured to:
determine, using a first neural network, body keypoints of a person based on at least a first image of the person;
determine, using a second neural network, at least a first plurality of parameters associated with a pose of the person based on the body keypoints of the person determined by the first neural network;
determine, using a third neural network, a second plurality of parameters associated with a shape of the person, wherein the one or more processors are configured to determine a normal map based on a second image of the person and further determine the second plurality of parameters based on the normal map; and
generate, based on at least the first plurality of parameters and the second plurality of parameters, a three-dimensional (3D) human model that represents at least the pose and the shape of the person.