US 12,299,927 B2
Apparatus and methods for three-dimensional pose estimation
Shandong Wang, Beijing (CN); Yangyuxuan Kang, Beijing (CN); Anbang Yao, Beijing (CN); Ming Lu, Beijing (CN); and Yurong Chen, Beijing (CN)
Assigned to Intel Corporation, Santa Clara, CA (US)
Appl. No. 18/000,389
Filed by Intel Corporation, Santa Clara, CA (US)
PCT Filed Jun. 26, 2020, PCT No. PCT/CN2020/098306
§ 371(c)(1), (2) Date Nov. 30, 2022,
PCT Pub. No. WO2021/258386, PCT Pub. Date Dec. 30, 2021.
Prior Publication US 2023/0298204 A1, Sep. 21, 2023
Int. Cl. G06T 17/00 (2006.01); G06T 7/73 (2017.01); G06T 19/20 (2011.01)
CPC G06T 7/74 (2017.01) [G06T 17/00 (2013.01); G06T 19/20 (2013.01); G06T 2207/10016 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30196 (2013.01); G06T 2207/30244 (2013.01)] 20 Claims
OG exemplary drawing
 
1. An apparatus comprising:
memory;
machine-readable instructions; and
at least one processor circuit to be programmed by the machine-readable instructions to:
synchronize a first image generated by a first image capture device and a second image generated by a second image capture device based on time, the first image including a subject and the second image including the subject;
execute a first neural network model to generate first two-dimensional position data, the first two-dimensional position data including predicted first positions of keypoints of the subject based on the first image, the keypoints corresponding to joints of the subject;
execute the first neural network model to generate second two-dimensional position data, the second two-dimensional position data including predicted second positions of the keypoints of the subject based on the second image;
execute a second neural network model to predict three-dimensional coordinates of respective joints of the subject based on the first two-dimensional position data and the second two-dimensional position data, the second neural network model to apply a depth offset between a first joint of the subject relative to a second joint of the subject to predict a three-dimensional coordinate for the first joint; and
generate a three-dimensional graphical model based on the three-dimensional coordinates of the respective joints, the three-dimensional graphical model representing a pose of the subject in the first image and the second image.