Manual surface inspection methods performed by quality inspectors do not satisfy the continuously increasing quality standards of industrial manufacturing processes. Machine vision provides a solution by using an automated visual inspection (AVI) system to perform quality inspection and remove defective products. Numerous studies and works have been conducted on surface inspection algorithms. With the advent of deep learning, a number of new algorithms have been developed for better inspection. In this paper, the state-of-the-art in surface defect inspection using deep learning is presented. In particular, we focus on the inspection of industrial products in semiconductor, steel, and fabric manufacturing processes. This work makes three contributions. First, we present the prior literature reviews on vision-based surface defect inspection and analyze the recent AVI-related hardware and software. Second, we review traditional surface defect inspection algorithms including statistical methods, spectral methods, model-based methods, and learning-based methods. Third, we investigate recent advances in deep learning-based inspection algorithms and present their applications in the steel, fabric, and semiconductor industries. Furthermore, we provide information on publicly available datasets containing surface image samples to facilitate the research on deep learning-based surface inspection.

质量检查员执行的手动表面检查方法无法满足工业生产过程不断提高的质量标准。机器视觉通过使用自动视觉检查（AVI）系统执行质量检查并清除有缺陷的产品，提供了一种解决方案。关于表面检查算法已经进行了许多研究和工作。随着深度学习的到来，已经开发了许多新算法来更好地进行检查。本文介绍了使用深度学习进行表面缺陷检测的最新技术。特别是，我们专注于检查半导体，钢铁和织物制造过程中的工业产品。这项工作做出了三点贡献。第一的，我们介绍了基于视觉的表面缺陷检查的现有文献综述，并分析了最近与AVI相关的硬件和软件。其次，我们回顾了传统的表面缺陷检查算法，包括统计方法，光谱方法，基于模型的方法和基于学习的方法。第三，我们研究基于深度学习的检查算法的最新进展，并介绍其在钢铁，织物和半导体行业中的应用。此外，我们提供了包含表面图像样本的公开可用数据集的信息，以促进基于深度学习的表面检测的研究。第三，我们研究基于深度学习的检查算法的最新进展，并介绍其在钢铁，织物和半导体行业中的应用。此外，我们提供了包含表面图像样本的公开可用数据集的信息，以促进基于深度学习的表面检测的研究。第三，我们研究基于深度学习的检查算法的最新进展，并介绍其在钢铁，织物和半导体行业中的应用。此外，我们提供了包含表面图像样本的公开可用数据集的信息，以促进基于深度学习的表面检测的研究。