Mesh Generation Using Visibility Features for Partial Body Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional mesh generation techniques fail to accurately estimate three-dimensional pose and shape data for bodies depicted in images, especially when parts of the body are hidden or obscured, leading to erroneous outputs.
Innovation Solution
A machine learning model is employed to generate visibility features indicating which parts of the body are visible in an image, allowing for the generation of an intermediate mesh and subsequent morphable model parameters, which are used to create an output mesh that includes non-visible portions, thereby improving the accuracy of vertex and joint predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional mesh generation techniques are used to estimate three-dimensional pose and shape data from images, then the process can be completed, but the accuracy of the generated mesh deteriorates when parts of the body are hidden or obscured
Solution Approach 1:
The patent segments the body into visible and non-visible parts by generating visibility features that indicate whether each part is visible in the input image. This segmentation allows the system to process only visible parts for mesh generation while maintaining awareness of non-visible parts, thereby improving accuracy when dealing with partial-body images.
Solution Approach 2:
The patent introduces visibility features as an intermediary element between the input image and the mesh generation process. These visibility features serve as a mediator that conveys information about which body parts are visible, enabling the system to adjust its mesh generation strategy accordingly and improve accuracy for partial-body images.
2Adaptability or versatility
If the system generates mesh data for all body parts including non-visible portions, then complete body modeling is achieved, but the accuracy of vertex and joint predictions deteriorates due to inclusion of invisible parts
Solution Approach 1:
The patent applies local quality by differentiating the treatment of visible and non-visible body parts. The system generates visibility features that locally indicate the visibility status of each body part, allowing the mesh generation process to apply different strategies to different regions - using visible parts for accurate prediction and inferred parts for completion, thereby maintaining overall accuracy while achieving complete modeling.
Data Source
AI summary
Systems and methods for mesh generation are described. One aspect of the systems and methods includes receiving an image depicting a visible portion of a body; generating an intermediate mesh representing the body based on the image; generating visibility features indicating whether parts of the body are visible based on the image; generating parameters for a morphable model of the body based on the intermediate mesh and the visibility features; and generating an output mesh representing the body based on the parameters for the morphable model, wherein the output mesh includes a non-visible portion of the body that is not depicted by the image.