3D Human Skeleton Estimation Through Occlusion-Aware Coordinates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital twin technologies face accuracy issues in estimating skeleton information of a person due to occlusion in parts like the foot or waist, leading to decreased estimation precision.
Innovation Solution
A non-transitory computer-readable recording medium stores a specifying program that processes video data to generate two-dimensional skeleton information, calculates relative and absolute three-dimensional coordinates based on the waist position, and estimates the relationship between hidden and non-hidden parts, specifying absolute coordinates using similarity transformation parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If three-dimensional skeleton position is estimated using coordinates of foot part and waist part, then estimation can be performed, but estimation accuracy decreases when occlusion occurs in predetermined parts
Solution Approach 1:
The patent segments the skeleton estimation process into multiple independent parts: detecting visible body parts, estimating their three-dimensional positions, identifying occluded parts, and separately estimating positions of occluded parts based on relationships with visible parts. This segmentation allows each part to be processed independently, improving overall accuracy even when some parts are occluded.
Solution Approach 2:
The patent introduces an intermediary relationship model that connects visible and occluded body parts. By establishing spatial relationships between visible parts (mediators) and occluded parts, the system can infer positions of occluded parts through the intermediary visible parts, maintaining estimation accuracy despite occlusion.
2Extent of automation
If skeleton information is generated from two-dimensional video data, then three-dimensional position can be estimated, but accuracy decreases when body parts are hidden
Solution Approach 1:
The patent performs preliminary detection of visible body parts and their two-dimensional coordinates before attempting to estimate three-dimensional positions. By pre-identifying which parts are visible and establishing their coordinates first, the system creates a reliable foundation for subsequent three-dimensional estimation, improving overall precision.
Solution Approach 2:
The system uses feedback from the detection results of visible parts to guide the estimation of occluded parts. The spatial relationships and coordinate information from visible parts serve as feedback that constrains and refines the estimation of occluded parts, improving measurement precision through iterative refinement.
Data Source
AI summary
The non-transitory computer-readable recording medium has stored therein a specifying program that causes a computer to execute a process including generating relative coordinates of a three-dimensional skeleton position based on a position of a waist of the person based on the two-dimensional skeleton information generating absolute coordinates of a three-dimensional skeleton position of a plurality of predetermined parts of the person estimating a relationship between the relative coordinates and the absolute coordinates based on absolute coordinates of the three-dimensional skeleton position corresponding to a part of the person that is not hidden and relative coordinates of the three-dimensional skeleton position corresponding to the part of the person that is not hidden and specifying absolute coordinates of a skeleton position of the person including a part other than the part that is not hidden or a part other than the predetermined parts based on the relationship that has been estimated.