Parameterized 2D Contour Person Model for 3D Body Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 2D models of the human body fail to accurately represent body shape and pose, and do not account for the influence of clothing on human shape, making it difficult to estimate 3D body shape from 2D images and recognize clothing types.
Innovation Solution
A 2D Contour Person (CP) model is developed, which captures natural shape and pose variations by factoring deformations into shape, pose, and camera viewpoint, and a Dressed Contour Person (DCP) model explicitly models clothing deformation using eigen-clothing learned through principal component analysis, allowing estimation of underlying body shape and clothing type.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 3D models of the human body are used to accurately model human shapes and poses, then modeling precision is improved, but device complexity and computational intensity increase
Solution Approach 1:
The patent segments the human body model into multiple articulated parts (head, torso, limbs, etc.) that can move independently. This segmentation allows the complex 3D model to be broken down into manageable components while maintaining overall accuracy. The articulated structure enables pose estimation through hierarchical transformations of these segmented parts.
Solution Approach 2:
The patent creates a simplified 2D contour-based copy of the 3D human body model that retains essential shape and pose information. This 2D contour model serves as a computationally efficient representation that captures the key geometric properties needed for estimation tasks without requiring full 3D complexity.
2Device complexity
If 2D models of the human body are used for representational and computational simplicity, then device complexity is reduced, but body shape representation accuracy deteriorates
Solution Approach 1:
The patent introduces dynamic articulation parameters to the 2D contour model, allowing it to represent pose variations through hierarchical transformations. The model includes movable joints and articulated parts that can dynamically adjust to match different body poses, transforming a static 2D model into a dynamic representation system.
Solution Approach 2:
The patent enhances the 2D contour model by incorporating shape parameters that describe body morphology variations. These parameters allow the model to adapt to different body types and shapes while maintaining computational simplicity, effectively adding expressive power to the 2D representation.
3Ease of operation
If existing 2D articulated person models focus on estimating human pose, then pose estimation capability is improved, but body shape recognition capability deteriorates
Solution Approach 1:
The patent creates a unified 2D articulated contour model that serves multiple functions simultaneously: pose estimation, body shape recognition, and clothing deformation analysis. The model's articulated structure and shape parameters enable it to perform these diverse tasks within a single framework, making it universally applicable to various human analysis problems.
4Device complexity
If clothing deformation is not modeled in 2D human body models, then model simplicity is maintained, but accuracy in estimating underlying body shape deteriorates
Solution Approach 1:
The patent extracts and separately models the clothing deformation component from the overall observed contour. By isolating the clothing layer and modeling its deformation independently, the system can distinguish between body shape and clothing effects, enabling accurate estimation of the underlying body shape even when clothing is present.
Data Source
Figure 1~2
Figure 3(a)~4
Figure 5~7(d)
AI summary
A novel "contour person" (CP) model of the human body is proposed that has the expressive power of a detailed 3D model and the computational benefits of a simple 2D part-based model. The CP model is learned from a 3D model of the human body that captures natural shape and pose variations; the projected contours of this model, along with their segmentation into parts forms the training set. The CP model factors deformations of the body into three components: shape variation, viewpoint change and pose variation. The CP model can be "dressed" with a low-dimensional clothing model, referred to as "dressed contour person" (DCP) model. The clothing is represented as a deformation from the underlying CP representation. This deformation is learned from training examples using principal component analysis to produce so-called eigen-clothing. The coefficients of the eigen-clothing can be used to recognize different categories of clothing on dressed people. The parameters of the estimated 2D body can be used to discriminatively predict 3D body shape using a learned mapping approach. The prediction framework can be used to estimate/predict the 3D shape of a person from a cluttered video sequence and/or from several snapshots taken with a digital camera or a cell phone.