Hand Position Estimation from Skeleton Joints in Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI algorithms for human posture estimation cannot directly detect hand position information in images, leading to incomplete recognition of human posture.
Innovation Solution
A method and apparatus that utilize a human skeleton detection model to determine key joint points, calculate height differences between head and first joint points, and use elbow and wrist joint points to determine hand size and position information of feature points surrounding the hand area, allowing rapid hand position recognition without complex calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a human skeleton detection model is used to detect key joint points, then the position information of joint points can be obtained, but the hand position information cannot be directly detected
Solution Approach 1:
The patent uses the human skeleton detection model as an intermediary tool to indirectly obtain hand position information. By detecting key joint points (head, first part, elbow, wrist) and calculating geometric relationships, the hand position is inferred without requiring direct hand detection capability, thus resolving the contradiction between measurement precision and detection difficulty
Solution Approach 2:
The patent performs preliminary detection of key joint points using the human skeleton detection model before calculating hand position information. This preliminary action provides the necessary input data (joint point positions) that enables subsequent hand position determination through geometric calculations, thereby achieving hand position detection without direct hand sensing capability
2Measurement precision
If hand position information is determined using complex calculations, then accurate hand position can be obtained, but algorithm delay and computing capacity requirements increase
Solution Approach 1:
The patent changes the approach from complex pixel-level hand detection to parameter-based geometric calculation. By transforming the problem into calculating hand position based on joint point coordinates and predetermined geometric relationships, the algorithm achieves accurate hand position determination with significantly reduced computational complexity and time consumption
Solution Approach 2:
The patent creates a simplified geometric model of the hand based on detected joint points. Instead of performing complex direct hand detection, the system copies the essential geometric relationships from the skeleton joint points to determine hand position, thereby achieving accurate results with much simpler calculations and reduced algorithm delay
Data Source
AI summary
The present application discloses a method for determining hand position information and a related device. The method includes: determining, according to a human skeleton detection model, position information of key part joint points of a human object in an image in a preset coordinate system; determining, according to position information of the head key part joint point and position information of the first part joint point, a height difference between the head key part joint point and the first part joint point; determining, according to the height difference, a hand size of the human object; determining position information of feature points according to the hand size, position information of the elbow joint point and position information of the wrist joint point, wherein the feature points are feature points of a preset shape surrounding a hand area; and determining the position information of the feature points as hand position information.