3D Gesture Recognition via Local Coordinate System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3D hand gesture recognition technologies face challenges such as high error and low accuracy due to unconstrained global and local gesture changes, frequent occlusions, and self-similarity, especially when predicting absolute three-dimensional coordinates of hand key points.
Innovation Solution
A method and apparatus that obtain a feature map from a depth image, determine a target key point, establish a three-dimensional coordinate system, and predict other key points' coordinates within this system, allowing for accurate gesture recognition by indirectly calculating absolute coordinates of key points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If three-dimensional absolute coordinates of hand key points are predicted directly, then gesture recognition can be achieved, but recognition error is high and accuracy is low
Solution Approach 1:
The patent segments the hand gesture recognition task into two parts: first predicting relative coordinates of key points in a local coordinate system, then transforming to absolute coordinates. This segmentation avoids the difficulty of predicting absolute coordinates directly while maintaining recognition accuracy.
Solution Approach 2:
The patent introduces a local coordinate system as an intermediary between the camera coordinate system and the hand key point coordinates. By predicting coordinates in this intermediate local system and then transforming to absolute coordinates, the method reduces prediction error compared to directly predicting absolute coordinates.
2Adaptability or versatility
If absolute three-dimensional coordinates are predicted without coordinate system constraints, then gesture recognition can be performed, but the lack of coordinate constraints causes great variance and high error
Solution Approach 1:
The patent changes the parameter space from absolute coordinates to relative coordinates within a local coordinate system. This parameter transformation reduces the variance and improves prediction accuracy while maintaining the ability to represent gestures in absolute space through coordinate transformation.
Data Source
AI summary
Provided are a gesture recognition method and apparatus, and a storage medium. The method includes: obtaining a first feature map corresponding to a target region in a depth image, and determining a target key point from the first feature map; establishing a three-dimensional coordinate system in the first feature map by using the target key point, and predicting other three-dimensional coordinates of other key points in the three-dimensional coordinate system, the other key points being key points other than the target key point in the first feature map; and determining, based on the target key point and the other three-dimensional coordinates, a plurality of absolute coordinates of a plurality of key points, to determine a gesture of a target recognition object in the target region based on the plurality of absolute coordinates, the plurality of key points including the target key point and the other key points.