Vehicle Occupant Physique Estimation Using 2D Skeleton Coordinates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing devices for estimating the physique and posture of vehicle occupants using three-dimensional coordinates are costly and prone to errors due to incorrect distance measurements when body parts are hidden or positioned diagonally, and they do not accurately acquire the body shape from the front view.
Innovation Solution
A device that uses a camera to capture images inside a vehicle, acquires skeleton point coordinates in a two-dimensional coordinate system, estimates the front coordinate of the occupant using machine learning with distance and angle information, and determines the physique based on this estimation, reducing errors and costs by processing in a two-dimensional coordinate system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If three-dimensional coordinate processing is used to estimate occupant physique, then measurement capability is improved, but device cost increases
Solution Approach 1:
The patent creates a two-dimensional front coordinate representation (copy) of the occupant's body from the captured image, which suffices for physique estimation without requiring expensive three-dimensional coordinate measurement systems. This copying approach maintains measurement capability while reducing device cost.
2Measurement precision
If three-dimensional coordinate measurement is used, then distance measurement capability is improved, but measurement accuracy deteriorates when body parts are hidden
Solution Approach 1:
The patent extracts only the necessary two-dimensional coordinate information from the captured image that is required for physique estimation, eliminating the need for three-dimensional distance measurements that fail when body parts are hidden. By taking out only the essential front-view coordinate data, the system achieves reliable physique estimation without the reliability issues of 3D measurement.
3Measurement precision
If three-dimensional coordinate system is used for posture estimation, then spatial positioning is improved, but processing complexity increases
Solution Approach 1:
The patent uses a simple two-dimensional coordinate system instead of a complex three-dimensional system, achieving sufficient position estimation accuracy for physique and posture analysis at lower processing complexity. This approach treats the coordinate system as a simplified model that is adequate for the specific application without requiring expensive or complex 3D processing infrastructure.
Data Source
AI summary
A physique estimation device includes: a camera capturing an image of an inside of a vehicle; a skeleton point coordinate acquisition unit acquiring positions of skeleton points of a body image of an occupant included in a captured image of the inside as skeleton point coordinates in a two-dimensional coordinate system of the captured image; a front coordinate estimation unit machine learned such that, by receiving a distance between preset skeleton points and an angle of a skeleton line obtained by connecting the skeleton points with respect to a reference line, a front coordinate that is an arrangement of the skeleton points when the occupant is viewed from front is estimated and output; and a physique determination unit determining a physique of the occupant based on the front coordinate.