Strip steel surface defect detection based on lightweight YOLOv5

Article Engineering, Electrical & Electronic

DCAM-Net: A Rapid Detection Network for Strip Steel Surface Defects Based on Deformable Convolution and Attention Mechanism

Haixin Chen et al.

Summary: This article proposes a rapid detection network for strip steel surface defects based on deformable convolution and attention mechanism (DCAM-Net). It introduces contrast limited adaptive histogram equalization (CLAHE) as a data augmentation method and a novel enhanced deformation feature extraction block (EDE-block) to improve defect detection. The coordination attention (CA) module is also introduced to further improve the network's ability to locate defects.

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT (2023)

添加到收藏夹

Article Engineering, Multidisciplinary

MSC-DNet: An efficient detector with multi-scale context for defect detection on strip steel surface

Rongqiang Liu et al.

Summary: In this paper, a defect detection module called MSC-DNet is proposed to localize and classify surface defects. The MSC-DNet utilizes a parallel structure of dilated convolution to capture the multi-scale context information of defects. A feature enhancement and selection module is also introduced to reduce confusing information. Experimental results show that the proposed MSC-DNet achieves high accuracy on benchmark datasets, meeting the quasi-real-time requirement.

MEASUREMENT (2023)

添加到收藏夹

Article Computer Science, Artificial Intelligence

A multi-scale pooling convolutional neural network for accurate steel surface defects classification

Guizhong Fu et al.

Summary: In this study, an innovative multi-scale pooling convolutional neural network was developed for high-accuracy steel surface defect classification, which can effectively capture defect location and obtain robust results.

FRONTIERS IN NEUROROBOTICS (2023)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Research on steel rail surface defects detection based on improved YOLOv4 network

Zengzhen Mi et al.

Summary: A deep learning algorithm is proposed to detect rail defects by performing sequential processes such as rail region extraction, improved Retinex image enhancement, background modeling difference, and threshold segmentation. The classification of defects is improved by introducing Res2Net and CBAM attention mechanism. Compared to other mainstream target detection algorithms, the improved YOLOv4 model shows excellent performance for rail defects detection and can be well-applied to rail defect detection projects.

FRONTIERS IN NEUROROBOTICS (2023)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Surface defect detection and classification of steel using an efficient Swin Transformer

Wei Zhu et al.

Summary: In this study, a new network architecture called LSwin Transformer is proposed for steel-surface defect detection. Various techniques, such as convolutional embedding module, attention patch merging module, and depth multilayer perceptron module, are introduced to enhance the adaptability of the Swin Transformer model. Experimental results demonstrate that the proposed model outperforms other methods with a mean average precision of 81.2%.

ADVANCED ENGINEERING INFORMATICS (2023)

添加到收藏夹

Article Computer Science, Artificial Intelligence

Focal and efficient IOU loss for accurate bounding box regression

Yi-Fan Zhang et al.

Summary: This paper focuses on the crucial step of bounding box regression in object detection and proposes two improvements, named EIOU loss and Focal-EIOU loss, to address the issues with previous loss functions. Experimental results demonstrate notable superiority in convergence speed and localization accuracy compared to other methods.

NEUROCOMPUTING (2022)

添加到收藏夹

Article Engineering, Electrical & Electronic

Efficient Fused-Attention Model for Steel Surface Defect Detection

Ching-Chi Yeung et al.

Summary: Steel surface defect detection is a crucial task in manufacturing. This article proposes a fused-attention network (FANet) to address the challenges of scale variations, shape variations, and detection efficiency in defect detection. The proposed method achieves state-of-the-art performance on two steel surface defect detection datasets by applying an attention mechanism, an adaptively balanced feature fusion method, and a fused-attention module to improve accuracy and speed.

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT (2022)

添加到收藏夹

Article Engineering, Multidisciplinary

Development of a YOLO-V3-based model for detecting defects on steel strip surface

Xupeng Kou et al.

Summary: A defect detection model based on YOLO-V3 was developed in this study, utilizing anchor-free feature selection mechanism and dense convolution blocks to improve feature reuse and characterization ability of the network, leading to higher performance compared to other models.

MEASUREMENT (2021)

添加到收藏夹

Article Mathematical & Computational Biology

A New Steel Defect Detection Algorithm Based on Deep Learning

Weidong Zhao et al.

Summary: In response to the low accuracy of current target detection algorithms in steel surface defect detection, the authors of this article proposed an improved target detection algorithm based on machine vision and implemented a series of improvement measures, including reconstructing the Faster R-CNN network structure, training the network with multiscale fusion, and using deformable convolution networks for complex target features.

COMPUTATIONAL INTELLIGENCE AND NEUROSCIENCE (2021)

添加到收藏夹

Article Engineering, Electrical & Electronic