Article Classification Method Using Iterative Evaluation Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing visual inspection methods face challenges in accurately classifying articles due to the complexity of changing evaluation methods when dealing with images of small defects or low-contrast defects, and require user intervention for re-classification, complicating the learning process.
Innovation Solution
A method that determines an evaluation method for classifying articles by using sample images, adjusts the classification group for singularly valued samples, and iteratively refines the evaluation method based on new classifications, allowing for improved accuracy and reduced user intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the evaluation method is determined using sample images with small defects or low-contrast defects, then the classification performance may deteriorate, but changing the evaluation method requires re-classifying all samples which complicates the learning process
Solution Approach 1:
The patent segments the sample images into two groups: those with singular evaluation values (outliers) and those with normal evaluation values. By handling these groups differently - changing the group assignment only for samples with singular values while keeping others unchanged - the system avoids the complexity of re-classifying all samples when modifying the evaluation method.
Solution Approach 2:
Instead of re-classifying all samples when changing the evaluation method (excessive action), the patent applies partial action by only re-classifying the subset of samples that have singular evaluation values. This partial approach reduces the learning process complexity while still improving classification performance for the affected samples.
2Measurement precision
If all samples are re-classified when changing evaluation method, then classification accuracy improves, but the learning process becomes more complex and time-consuming
Solution Approach 1:
The patent applies partial action by identifying and re-classifying only the subset of samples with singular evaluation values rather than re-classifying all samples. This significantly reduces the time required for the learning process while still improving classification accuracy for the affected samples, as demonstrated by the improved detection rate of defective products.
Solution Approach 2:
The patent implements a feedback mechanism where samples with singular evaluation values are identified, their group assignments are changed, and the evaluation method is updated based on this feedback. This iterative feedback process improves classification accuracy efficiently without requiring complete re-classification of all samples each time.
Data Source
AI summary
The present invention provides a classifying method of classifying an article into one of a plurality of groups based on an image of the article, comprising determining an evaluation method for obtaining an evaluation value of an image by using at least some of sample images, obtaining evaluation values for the sample images by the determined evaluation method, changing a group to which a sample of at least one sample image having a singular evaluation value among the sample images belongs, changing the evaluation method using the sample images after changing the group to which the sample of the at least one sample image belongs, obtaining an evaluation value for the image of the article using the changed evaluation method, and classifying the article into one of the plurality of groups based on the evaluation value for the image of the article.