Distributed Image Analysis Using Certainty-Weighted Small Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face issues with communication volume, privacy protection, and data management costs due to the transmission of full image data to servers, and lack the ability to process high-value detection results efficiently.
Innovation Solution
A system that converts original data into smaller, certainty-weighted data packets for transmission and analysis, using machine-learned models to ensure high reliability and value without transmitting the full data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If original image data is transmitted to the server, then analysis accuracy is improved, but communication volume and data management cost increase
Solution Approach 1:
The terminal extracts only the necessary feature amounts and detection results from the original image data, transmitting only these extracted elements to the server. This extraction process removes unnecessary data components while preserving the essential information needed for analysis, thereby reducing communication volume while maintaining analysis accuracy.
Solution Approach 2:
Instead of transmitting the complete original image data, the system creates a simplified copy consisting of feature amounts and detection results. This copy contains the critical information needed for server-side analysis while occupying significantly less communication bandwidth and storage space.
2Loss of information
If original image data is transmitted to the server, then complete information is available for analysis, but privacy protection is compromised
Solution Approach 1:
The terminal extracts only the essential feature amounts and detection results needed for analysis, leaving behind the original image data that contains sensitive personal information. This extraction ensures that complete analytical information is transmitted to the server while the privacy-sensitive original data remains locally on the terminal device.
3Adaptability or versatility
If original image data is stored on the server, then data availability is improved, but data management cost increases
Solution Approach 1:
The system extracts and transmits only the necessary feature amounts and detection results to the server for storage and analysis. This approach ensures that the server has access to all data needed for various analytical purposes while avoiding the burden of storing large volumes of original image data, thereby reducing data management costs.
4Quantity of substance
If small data is transmitted instead of original data, then communication volume is reduced, but information analysis capability is limited
Solution Approach 1:
The terminal transforms the original image data into a different parameter representation consisting of feature amounts and detection results. This parameter transformation maintains the essential analytical information while significantly reducing data size, enabling the server to perform comprehensive analysis on the transformed data without receiving the complete original image data.
Data Source
AI summary
An analysis system includes: a terminal that comprises a terminal controller; and a server that comprises a server controller and a communication interface, the terminal controller acquiring original data, converting the original data into a plurality of pieces of small data each having a size smaller than a size of each piece of the original data and into certainty factors of respective pieces of the small data, and transmitting the small data to the server, and the server controller receiving the small data from the terminal via the communication interface, executing analysis using the small data, and outputting an analysis result of the analysis.