Visual Guide for Consistent Image Classification Training Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The inconsistent and unclear definitions of defects in visual inspection processes lead to inefficient and costly quality control, especially when transitioning from human inspectors to machine learning models, due to deficient training data.
Innovation Solution
A user-generated visual guide provides a consistent framework for image classification, allowing administrators and qualified human classifiers to define and seed a framework with 'good quality' images, which is then used to train machine learning models and improve the accuracy of both human and AI classifiers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If human inspectors define defects based on institutional knowledge, then visual inspection can be performed, but the definitions are inconsistent and unclear leading to poor reliability
Solution Approach 1:
The patent creates a visual guide that copies and standardizes defect definitions from human expert knowledge into a structured format with representative images and descriptions. This copying process transforms subjective institutional knowledge into objective, reusable training data that maintains reliability while improving consistency.
Solution Approach 2:
The patent performs preliminary action by pre-defining defect categories and creating a visual guide before machine learning model training. This preliminary structuring of defect information ensures consistent definitions are established upfront, preventing inconsistency during the actual inspection process.
2Extent of automation
If machine learning models are trained with deficient training data, then automation can be achieved, but the model performance is poor
Solution Approach 1:
The visual guide serves as an intermediary between human expert knowledge and machine learning models. It translates subjective defect definitions into objective training data with representative images, descriptions, and categories, enabling automated inspection while maintaining high accuracy through standardized training examples.
Solution Approach 2:
The patent changes the parameters of training data by structuring defect information into standardized categories, adding representative images, and creating consistent descriptions. This parameter transformation converts unstructured expert knowledge into structured training data that improves model accuracy while enabling automation.
3Reliability
If visual inspection is performed manually, then defect definitions can be applied, but the process is time consuming and expensive
Solution Approach 1:
The visual guide copies defect definition knowledge from human inspectors into a reusable digital format. This copying enables the same quality standards to be applied automatically by machine learning models, maintaining inspection quality while dramatically improving productivity by eliminating manual review time.
Data Source
AI summary
A process and a system for creating a visual guide for developing training data for a classification of image, where the training data includes images tagged with labels for the classification of the images. A processor may prompt a user to define a framework for the classification. For an initial set of images within the training data, qualified human classifiers are prompted to locate the images within the framework and to tag the images with labels. The processor determines whether the tagged images have consistent labels, and, if so, the processor adds images to the training data. The processor may add the images by providing a visual guide, the visual guide including tagged images arranged according to their locations within the framework their labels, and prompting human classifiers to tag the additional images with labels for the classification, according to the visual guide.