Virtual Human Projection for Accurate 3D Action Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating human actions in a real target space from two-dimensional images lack accuracy, necessitating improved techniques for precise action estimation.
Innovation Solution
A projection system that estimates a three-dimensional pose of a real human and projects a virtual human onto a virtual target space, allowing action estimation based on the relationship between the virtual human and the virtual space configuration, thereby enhancing accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If action estimation is performed directly from two-dimensional images, then the process is simple, but the estimation accuracy is insufficient
Solution Approach 1:
The patent creates a virtual copy of the real human by projecting the detected real human onto a virtual target space, generating a virtual human that replicates the real human's pose and characteristics. This virtual copy enables more accurate action estimation by analyzing the virtual human's relationship with virtual objects, resolving the accuracy limitation of direct 2D image analysis while maintaining a manageable system through virtual representation rather than complex physical measurement systems
Solution Approach 2:
The patent transitions from two-dimensional image space to three-dimensional virtual space by projecting the real human detected in 2D images onto a virtual target space with 3D coordinates. This dimensional transformation allows for more accurate pose estimation and action analysis by utilizing depth information and spatial relationships that are not available in 2D images alone, thereby improving measurement precision without requiring complex 3D sensing hardware
2Measurement precision
If a virtual human is projected onto virtual target space with three-dimensional pose, then action estimation accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent introduces a virtual target space as an intermediary between the real world and the analysis system. The real human is first detected in 2D images, then projected onto the virtual target space to create a virtual human with 3D pose. This virtual representation serves as a mediator that simplifies subsequent action estimation by enabling straightforward analysis of spatial relationships between the virtual human and virtual objects, avoiding the need for complex direct 3D measurement systems
Solution Approach 2:
The patent creates a virtual copy of the real human with three-dimensional pose information by projecting the detected real human onto the virtual target space. This virtual copy preserves the essential characteristics and pose of the real human while existing in a simplified virtual environment where action estimation can be performed more accurately and efficiently, reducing processing complexity compared to direct real-world analysis
Data Source
AI summary
A projection system stores virtual space configuration information indicating a configuration of objects defined in a virtual target space representing a real target space. The projection system detects a real human shown in an image taken by a real camera installed in the real target space, and estimates a three dimensional pose of the real human based on the image. Then, the projection system projects a virtual human having the three dimensional pose and representing the real human onto the virtual target space. In addition, the projection system performs an action estimation process that estimates an action of the real human in the real target space by estimating an action of the virtual human in the virtual target space based on a relationship between the virtual human having the three dimensional pose and the configuration of the objects in the virtual target space.