Details

Autor(en) / Beteiligte

• Fang, Zhenghui
• Wang, Anna
• Bu, Chunguang
• Liu, Chen

Titel

3D Human Pose Estimation Using RGBD Camera

Ist Teil von

• 2021 IEEE International Conference on Computer Science, Electronic Information Engineering and Intelligent Control Technology (CEI), 2021, p.582-587

Ort / Verlag

IEEE

Erscheinungsjahr

2021

Quelle

IEEE Electronic Library Online

Beschreibungen/Notizen

• This work addresses a method to map the estimated 2D pose from a color image to 3D pose and obtain more accurate 3D keypoints. In this paper, the RGBD camera is used to simultaneously acquire color images and depth images. AlphaPose is used to estimate the 2D keypoints of the target, and the 2D keypoints are mapped to the corresponding depth maps to obtain the corresponding 3D keypoint. We adopt the parametric human body model, SMPL, as the 3D prior of the human body. The SMPL model and the data of point cloud are aligned by optimizing the global orientation and translation of the SMPL as well as optimizing the error of the SMPL vertices and the point cloud of the human body. The experimental results show that our method can effectively improve the accuracy of the existing 3D human pose and position estimation methods.