Motion Identification Using Multi-Channel Feature Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing motion identification methods struggle to accurately distinguish between various user motions, especially when there are many kinds of motions or when motion differences are insufficient, leading to confusion and difficulty in identifying key features.
Innovation Solution
The system employs a processor with a preprocessing module, data cutting module, channel matching module, data integration module, and model training module to capture and process movement data using convolution and pooling algorithms, generating an identification model that effectively differentiates sustained and non-sustained motions by simplifying feature groups and reducing calculation complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If machine learning and statistical analysis are used for motion identification, then model training and state identification can be performed, but motion differences are insufficient and motions are easily confused
Solution Approach 1:
The patent segments motion data into multiple channels based on different feature groups (e.g., acceleration, gyroscope, magnetometer data separated into different channels). Each channel processes specific feature segments independently, allowing the system to capture subtle differences between motions that would be lost in aggregated analysis. This segmentation enables precise identification of key features by analyzing each motion dimension separately.
Solution Approach 2:
The patent introduces a channel dimension to organize feature groups, transforming the data structure from flat feature vectors to multi-channel feature maps. By adding this dimensional organization, the system can process and compare motions across multiple dimensions simultaneously, improving the ability to distinguish between similar motions through convolution operations across channels.
2Measurement precision
If detailed feature analysis is performed to distinguish motion differences, then identification precision improves, but calculation amount increases
Solution Approach 1:
The patent divides feature data into segments that are processed independently through convolution and pooling operations. By segmenting the data processing pipeline, the system can apply efficient algorithms to smaller data chunks rather than processing all features at once, reducing overall computational load while maintaining precision through the structured organization of feature groups into channels.
Solution Approach 2:
The patent applies pooling operations that summarize feature groups by selecting representative values (e.g., max pooling or average pooling) rather than processing every individual feature point. This partial processing approach captures essential motion characteristics while discarding redundant details, achieving good identification precision with reduced calculation requirements.
Data Source
AI summary
An identification method includes: sensing movement data; capturing multiple feature data from the movement data; cutting the first feature data into a plurality of first feature segments, dividing the first feature segments into a plurality of first feature groups, and calculating multiple first similarity parameters of the first feature groups respectively corresponding to a plurality of channels; making the first feature groups correspond to the channels according to the first similarity parameters; simplifying the first feature groups corresponding to the channels respectively by a convolution algorithm to obtain a plurality of first convolution results corresponding to the first feature groups; simplifying the first convolution results corresponding to the first feature groups respectively by a pooling algorithm to obtain multiple first pooling results corresponding to the first feature groups; and combining the first pooling results corresponding to the first feature groups to generate a first feature map.