Comparative Analysis of Classification Models for Steel Rods Using Convolution Neural Networks and Fuzzy Logic Systems
Main Article Content
Abstract
Maintaining high-quality steel rods is crucial in the competitive steel-strip production industry. Human visual inspection has traditionally been the primary method for detecting flaws, but it has limitations in accuracy, processing time, cost, and reliability. Automated visual inspection technologies have been developed to address these issues. However, in-depth research on vision-based approaches for identifying and categorizing surface flaws in steel products has also proven ineffective. This paper presents a comparative analysis of various classification models for different types of steel rods using a combination of pre-trained Convolutional Neural Networks (CNNs) and fuzzy logic systems. We employ VGG16 to extract features from grayscale images of steel rods and then use multiple classifiers such as Support Vector Machines (SVM), Decision Trees, Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), and K-Nearest Neighbors (KNN). Additionally, we explore fuzzy c-means clustering, fuzzy SVM, and fuzzy regression models to enhance classification accuracy. Experimental results indicate that fuzzy logic systems can significantly improve classification performance compared to traditional methods. Our study includes learning-driven, statistical, spectral, and texture segmentation methods, as well as diverse classification techniques. The paper also outlines future research directions.
Article Details

This work is licensed under a Creative Commons Attribution-NoDerivatives 4.0 International License.