CNN-Based Product Quality Incident Counting from Appearance Defects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the number of product quality incidents are inaccurate and slow due to reliance on statistical methods and customer feedback, which involve subjective analysis and delayed data processing.
Innovation Solution
A method using a convolutional neural network (CNN) to analyze appearance quality data from products under inspection, building a quality evaluation model to detect product appearance defects, and calculating an incident occurrence coefficient to determine the number of product quality incidents.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If statistical methods and customer feedback are used to determine the number of product quality incidents, then the process is simple to implement, but the accuracy and speed of determination are low
Solution Approach 1:
The patent replaces manual statistical methods and subjective customer feedback analysis with an automated convolutional neural network system. The CNN model processes product appearance images to automatically detect and count quality incidents, substituting human-based mechanical processes with intelligent automated systems that provide both high accuracy and efficiency
Solution Approach 2:
The patent introduces a convolutional neural network model as an intermediary between raw product images and quality incident determination. This intermediary automatically extracts features, identifies defects, and quantifies quality incidents, eliminating the need for manual statistical analysis while maintaining system implementability
2Productivity
If manual statistical analysis of quality data is performed, then the system complexity is low, but the processing speed and timeliness are slow
Solution Approach 1:
The patent replaces slow manual statistical analysis with automated deep learning-based image processing. The CNN model processes multiple product images simultaneously, automatically counting quality incidents in real-time, thereby dramatically improving processing speed while managing system complexity through modular architecture
Solution Approach 2:
The patent performs preliminary feature extraction and defect identification through the trained CNN model before final quality incident determination. This preliminary processing prepares data in advance, enabling rapid real-time quality assessment without complex manual analysis during the inspection phase
3Measurement precision
If customer feedback data is collected and analyzed, then comprehensive quality understanding is achieved, but the time delay and subjectivity increase
Solution Approach 1:
The patent performs quality incident detection immediately during the product inspection phase using the CNN model, rather than waiting for customer feedback. This preliminary action provides objective, real-time quality data at the source, eliminating both time delays and the subjectivity inherent in customer perception-based methods
Solution Approach 2:
The patent enables the quality inspection system to autonomously detect and count quality incidents through automated image processing and CNN-based defect recognition. This self-service capability eliminates dependency on external customer feedback, providing immediate objective measurements without human subjectivity or time delays
Data Source
AI summary
Disclosed is a method for determining a number of product quality incidents based on a convolutional neural network, the method includes the following steps: building an appearance quality dataset, building a quality evaluation model, and analyzing a number of product quality incidents; a preset model and a preset optimized model are trained and validated according to the built appearance quality dataset to obtain initial quality evaluation models, which are filtered to obtain a quality evaluation model according to model performance coefficient and comprehensive model coefficient, product appearance defect data are obtained according to appearance defect data detected and obtained by the quality evaluation model, appearance defect ratio and significant appearance defect ratio of the products under quality inspection are obtained according to the product appearance defect data, and a number of product quality incidents of the products under quality inspection is finally obtained according to an incident occurrence coefficient.