Steel Surface Defect Diagnostics Using Deep Convolutional Neural Network and Class Activation Map

Seungchul Lee,Soo Young Lee,Bayu Adhi Tama,Seok Jun Moon

doi:10.3390/app9245449

Abstract

Steel defect diagnostics is considerably important for a steel-manufacturing industry as it is strongly related to the product quality and production efficiency. Product quality control suffers from a real-time diagnostic capability since it is less-automatic and is not reliable in detecting steel surface defects. In this study, we propose a relatively new approach for diagnosing steel defects using a deep structured neural network, e.g., convolutional neural network (CNN) with class activation maps. Rather than using a simple deep learning algorithm for the classification task, we extend the CNN diagnostic model for being used to analyze the localized defect regions within the images to support a real-time visual decision-making process. Based on the experimental results, the proposed approach achieves a near-perfect detection performance at 99.44% and 0.99 concerning the accuracy and F-1 score metric, respectively. The results are better than other shallow machine learning algorithms, i.e., support vector machine and logistic regression under the same validation technique.

Highlights

Quality inspection and control in the steel-manufacturing industry have been a critical issue for assuring the product quality and increasing its productivity
We briefly explain the outline of the research, followed by details of deep learning-based proposed architecture using a convolutional neural network and class activation map, as well as parametric measures used in the experiment
We proposed a steel surface defect detection based on a convolutional neural network with a class activation map

Summary

Introduction

Quality inspection and control in the steel-manufacturing industry have been a critical issue for assuring the product quality and increasing its productivity. For an immediate response and control about the flaws, detecting steel defects should be preceded to analyze the failure causes. To this end, a sophisticated diagnostic model is required to detect the failures properly and to enhance the capability of quality control. The development of a visual inspection system for identifying steel surface defects should be conducted to secure the reliability of the process and the product. Gray level co-occurrence matrix (GLCM), proposed by Haralick et al [18], is a two-dimensional matrix that contains statistical information about the texture of the single-channel image. Owing to its advantages of extracting the features from the images, it has been utilized to the image analysis tasks in a variety of fields, e.g., medical, material and manufacturing [19,20,21]

Methods

Results

Conclusion