A Review on Industrial Surface Defect Detection Based on Deep Learning Technology

Abstract

In recent years, with the rapid development of deep learning, computer vision technology based on convolutional neural network (CNN) is widely used in industrial fields. At present, surface defect detection by machine vision is one of the most mature applications of CNN in industry. This paper provides a comprehensive overview of deep learning in the field. First of all, we briefly introduce the major tasks of CNN in computer vision researches, including image classification, object detection, edge detection and image segmentation, which are frequently used techniques in surface defect inspection. After that, we describe in detail the applications of computer vision based on CNN models in a variety of industrial scenarios for surface defect detection tasks, which mainly cover the steel surface defect inspection, magnetic tile surface defect inspection, rail surface defect inspection, screen surface detect inspection, solar cell surface defect inspection, and some others. As an emerging representative of artificial intelligence technology, we believe that deep learning will gradually become one of the mainstream technologies for industrial vision in the future. Accordingly, this paper aims to present a reference and guidance for researchers in industry to apply the advanced technology of deep learning.