A steel surface defect inspection approach towards smart industrial monitoring

Hao, Ruiyang; Lu, Bingyu; Cheng, Ying; Li, Xiu; Huang, Biqing

doi:10.1007/s10845-020-01670-2

Cited by 100 publications

(45 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In the manufacturing industry, computer vision is widely used with applications in, magnetic tile surface defects (Huang et al 2018), additive manufacturing anomaly detection (Scime and Beuth 2018;Davtalab et al 2020), steel surface defect inspection (Hao et al 2020;Sun et al 2016), defect detection in lithium ion battery electrodes (Badmos et al 2020), roughness prediction (Grzenda and Bustillo 2019) etc. GANs models, especially after improvement of its training algorithm (Gulrajani et al 2017), have been proposed for anomaly detection in structured and arbitrary textured surfaces (Lai et al 2018) and unsupervised inspection of surfaces (Zhai et al 2018).…”

Section: Literature Surveymentioning

confidence: 99%

Synthetic data augmentation for surface defect detection and classification using deep learning

et al. 2020

View full text Add to dashboard Cite

Deep learning techniques, especially Convolutional Neural Networks (CNN), dominate the benchmarks for most computer vision tasks. These state-of-the-art results are typically obtained through supervised learning, for which large annotated datasets are required. However, acquiring such datasets for manufacturing applications remains a challenging proposition due to the time and costs involved in their collection. To overcome this disadvantage, a novel framework is proposed for data augmentation by creating synthetic images using Generative Adversarial Networks (GANs). The generator synthesizes new surface defect images from random noise which is trained over time to get realistic fakes. These synthetic images can be used further for training of classification algorithms. Three GAN architectures are trained, and the entire data augmentation pipeline is implemented for the Northeastern University (China) Classification (NEU-CLS) dataset for hot-rolled steel strips from NEU Surface Defect Database. The classification accuracy of a simple CNN architecture is measured on synthetic augmented data and further it is compared with similar state-of-the-arts. It is observed that the proposed GANs-based augmentation scheme significantly improves the performance of CNN for classification of surface defects. The classically augmented CNN yields sensitivity and specificity of 90.28% and 98.06% respectively. In contrast, the synthetically augmented CNN yields better results, with sensitivity and specificity of 95.33% and 99.16% respectively. Also, the use of GANs is demonstrated to disentangle the representation space and to add additional domain knowledge through synthetic augmentation that can be difficult to replicate through classic augmentation. The proposed framework demonstrates high generalization capability. It may be applied to other supervised surface inspection tasks, and thus facilitate the development of advanced vision-based inspection instruments for manufacturing applications.

show abstract

Section: Literature Surveymentioning

confidence: 99%

Synthetic data augmentation for surface defect detection and classification using deep learning

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Hence, the intra-class diversity and inter-class similarity might exacerbate the misleading of the training model. To illustrate the complex variance of defects, Hao et al [2] demonstrated the ratio distribution and size distribution of steel surface defects based on the NEU-DET dataset. Intuitively, up to about three-quarters of the surface defects were measured to have a ratio of the long side to the short side between one to three.…”

Section: Datasets Analysismentioning

confidence: 99%

“…In reality, owing to external factors such as equipment fatigue, human negligence, and external force, the steel surface may contain various types of defects. Consequently, these surface defects potentially affect the capability of steel products such as wear resistance, fatigue strength, and residual life [1,2], leading to huge economic losses for manufacturers and posing a high risk to worker safety. As such, defect recognition is an essential task for assuring product quality in manufacturing.…”

Section: Introductionmentioning

confidence: 99%

Design of Multi-Receptive Field Fusion-Based Network for Surface Defect Inspection on Hot-Rolled Steel Strip Using Lightweight Dataset

et al. 2021

View full text Add to dashboard Cite

With the advancement of industrial intelligence, defect recognition has become an indispensable part of facilitating surface quality in the steel manufacturing process. To assure product quality, most previous studies were typically trained with many defect samples. Nonetheless, a large quantity of defect samples is difficult to obtain, owing to the rare occurrence of defects. In general, deep learning-based methods underperformed as they have inherent limitations due to inadequate information, thereby restraining the application of models. In this study, a two-level Gaussian pyramid is applied to decompose raw data into different resolution levels simultaneously filtering the noises to acquire compact and representative features. Subsequently, a multi-receptive field fusion-based network (MRFFN) is developed to learn the hierarchical features and synthesize the respective prediction scores to form the final recognition result. As a result, the proposed method is capable of exhibiting an outstanding performance of 99.75% when trained using a lightweight dataset. In addition, the experiments conducted using the disturbance defect dataset showed the robustness of the proposed MRFFN against common noises and motion blur.

show abstract

“…In addition, a study [6] performed defect detection using sliding window methods to distinguish poor surface conditions (such as scratches and poor junctions). Furthermore, in the case of finding the locations of the defects, there are studies [5,20,21] on the detection of defects on the surface of steel based on yolo series [22,23], variational auto-encoder (VAE) [24], or R-CNN series [25], such as Fast R-CNN [26] and Mask R-CNN [27]. Similarly, [28] applied 3DCNN to analyze three dimensional point cloud data.…”

Section: Product Inspection With Deep Learningmentioning

confidence: 99%

“…In addition, there are papers that help researchers apply deep learning by creating open datasets. Teh NEU surface defect database is a steel plate defect inspection dataset opened by [32] and the aforementioned papers [19,20] also used the dataset. Moreover, KolektorSDD(Kolektor Surface-Defect Dataset) was created by [33] and PCB scans dataset, which are laser scans of PCBs, were generated by [28].…”

Section: Product Inspection With Deep Learningmentioning

confidence: 99%

Product Inspection Methodology via Deep Learning: An Overview

Kim

Cho

2021

Sensors

View full text Add to dashboard Cite

In this study, we present a framework for product quality inspection based on deep learning techniques. First, we categorize several deep learning models that can be applied to product inspection systems. In addition, we explain the steps for building a deep-learning-based inspection system in detail. Second, we address connection schemes that efficiently link deep learning models to product inspection systems. Finally, we propose an effective method that can maintain and enhance a product inspection system according to improvement goals of the existing product inspection systems. The proposed system is observed to possess good system maintenance and stability owing to the proposed methods. All the proposed methods are integrated into a unified framework and we provide detailed explanations of each proposed method. In order to verify the effectiveness of the proposed system, we compare and analyze the performance of the methods in various test scenarios. We expect that our study will provide useful guidelines to readers who desire to implement deep-learning-based systems for product inspection.

show abstract

A steel surface defect inspection approach towards smart industrial monitoring

Cited by 100 publications

References 37 publications

Synthetic data augmentation for surface defect detection and classification using deep learning

Synthetic data augmentation for surface defect detection and classification using deep learning

Design of Multi-Receptive Field Fusion-Based Network for Surface Defect Inspection on Hot-Rolled Steel Strip Using Lightweight Dataset

Product Inspection Methodology via Deep Learning: An Overview

Contact Info

Product

Resources

About