Human shape recovery can in general be categorized in two classes: top-down vs. bottom-up. The former estimates poses by deforming a pre-defined 3D humanoid template non-rigidly frame by frame and avoids unreasonable poses by enforcing constraints on surface deformations. The latter starts with the detected 2D body parts or skeletons, then back-projects the points into 3D space either via depth data or triangulating multiple views. While the top-down scenario has been widely covered and applied in multiple view studios, e.g., Kinovis, the bottom-up approach is less studied and discussed. In this project, we aim at advancing the bottom-up approach. Specifically, one first recovers rough 3D skeletal poses with state-of-the-art bottom-up approach as initialization, and deform pre-defined human surfaces to fit to those poses. The generic human surface has to be adapted according to the size/scale of the skeleton, so as to yield correct anthropometric measurements (height, length of arms and circumference of waists, etc), as depeicted in the left figure. Skeleton-based animation such as linear-blend skinning, dual-quaternion-blend skinning is also worthwhile a look. Furthermore, besides relying on skeletal poses, we would also like to investigate the possibility of adjusting/refining human shapes according to the pressure measurements, as shown in the right figure.
|
![]() |
![]() |
Students.ProjectForm | |
---|---|
Title: | Human Shape Adaptation with with skeletons and pressure measurement |
Abstract: | |
Student: | |
Director: | Prof. Nassir Navab |
Supervisor: | Chun-Hao Paul Huang, Dr. Stefanie Demirci, Dr. Diana Mateus |
Type: | DA/MA/BA |
Area: | |
Status: | open |
Start: | |
Finish: | |
Thesis (optional): | |
Picture: |