Object Detection Using Joint Angle and Part Length Ratio
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection techniques, such as those used for human posture estimation from image data, face accuracy issues when the subject's body type differs from the average, leading to decreased detection accuracy and posture estimation reliability, especially when applied to objects with joints like humans or robots.
Innovation Solution
An object detection apparatus and method that extracts joint angles and part length ratios from image data to accurately detect information about objects with articulated parts, using a joint angle extraction section and part length ratio estimation section to improve detection accuracy regardless of body type.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a likelihood function statistically derived from images of multiple persons is used for posture estimation, then the detection method can be applied to various objects, but the accuracy decreases when the subject's body type differs from the average body type
Solution Approach 1:
The patent applies local quality by extracting and utilizing specific geometric features (joint angles and part length ratios) that are locally characteristic of each object's body structure. Instead of using a general likelihood function for all objects, the method adapts to local structural properties of individual objects, allowing accurate detection for objects with diverse body types while maintaining versatility across different object categories
Solution Approach 2:
The patent changes the detection parameters from general likelihood scores to specific geometric parameters (joint angles and part length ratios) that can be objectively measured and compared. By transforming the detection approach to use these invariant geometric parameters, the system achieves both versatility across object types and precision for individual body types
2Device complexity
If average body type-based detection methods are used, then the detection process is simplified, but the accuracy of posture estimation decreases for subjects with non-average body types
Solution Approach 1:
The patent enables self-service detection by allowing each object to define its own body constraint information based on its specific joint angles and part length ratios. The detection system automatically adapts to each object's unique characteristics without requiring manual configuration or average-based assumptions, achieving both simplicity and accuracy through automated self-adaptation
Data Source
AI summary
Disclosed is an object detection method capable of detecting with high precision information relating to a jointed object from image data. An object detection device (160) detects information relating to an object from image data of images captured of an object having multiple parts connected by joints. The disclosed object detection device (160) is provided with a joint angle extraction unit (161) which extracts the angle of a joint connecting two parts from candidates of the positions of two neighboring parts obtained from the image data, and a part length ratio estimation unit (165) which uses the joint angle to perform the detection described above.