Image Data Classification via Consensus Marking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine learning methods require manual marking of feature points in images, which can lead to inaccuracies and affect the reliability of image classification results.
Innovation Solution
An image data classification system that distributes images to multiple users for marking, uses image recognition algorithms to identify circled objects, and calculates a consensus mark rate to determine image quality, allowing for redistribution if the quality is below a certain threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If manual marking of feature points is used, then ease of operation is improved, but measurement precision deteriorates due to human error
Solution Approach 1:
The system implements a feedback mechanism where multiple users mark feature points independently, the system calculates consistency rates among different users' markings, and uses this feedback to verify or correct individual markings. This allows manual marking to remain easy while improving precision through collective verification.
Solution Approach 2:
The patent combines multiple manual markings from different users into a unified result. By merging the markings from multiple users and calculating consistency rates, the system achieves higher precision than a single user could achieve alone, while maintaining the ease of manual operation.
2Measurement precision
If multiple users mark images for consensus verification, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The system performs self-verification by automatically calculating consistency rates among different users' markings and determining whether to accept or reject each marking based on predefined thresholds. This self-service mechanism reduces the need for complex external validation systems.
Solution Approach 2:
The patent replaces complex manual verification processes with automated computational methods. Instead of requiring complex review procedures, the system uses algorithmic consistency calculation to verify markings, simplifying the overall system complexity while improving precision.
3Measurement precision
If images are redistributed for re-marking when quality is low, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The system performs preliminary calculations of consistency rates and quality assessments during the marking process itself, rather than requiring separate verification passes. This preliminary action identifies low-quality markings early, allowing for efficient redistribution only when necessary and minimizing overall time loss.
Data Source
AI summary
An image data classification method which includes distributing image data to n users is provided. N marks corresponding to the image data are collected by collecting the mark made by each of the n users on the image data. Once target marks are determined from the n marks and a rate of the target marks is calculated, a quality of the image data is determined according to the rate of the target marks.