Article
Engineering, Electrical & Electronic
Wenqi Cui, Kechen Song, Hu Feng, Xiujian Jia, Shaoning Liu, Yunhui Yan
Summary: Researchers propose a novel autocorrelation-aware aggregation network (A3Net) for salient object detection of strip steel surface defects. The use of attention mechanism and scale interaction module contributes to the superior performance of the proposed method on both public and newly built datasets.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Computer Science, Artificial Intelligence
Yongping Zhang, Sijie Shen, Sen Xu
Summary: In this study, a lightweight YOLOv5 strip steel surface defect detection algorithm is proposed, which achieves a good balance between detection speed and accuracy by introducing efficient lightweight convolutional layers and a non-parametric attention mechanism. Experimental results show that the proposed algorithm outperforms existing methods in terms of detection accuracy and speed.
FRONTIERS IN NEUROROBOTICS
(2023)
Article
Chemistry, Multidisciplinary
Xinwen Zhou, Mengen Wei, Qianglong Li, Yinghua Fu, Yangzhou Gan, Hao Liu, Jing Ruan, Jiuzhen Liang
Summary: This paper proposes a defect detection method based on RASPP module and DPN network, which can accurately locate and classify defects on the surface of steel strip. The experimental results demonstrate that the proposed method achieves better performance in terms of mAP, classification accuracy, and detection speed.
APPLIED SCIENCES-BASEL
(2023)
Article
Mathematics
Xinglong Feng, Xianwen Gao, Ling Luo
Summary: This paper proposes a deep learning model for strip steel defect classification, achieving a high classification accuracy by introducing FcaNet and CBAM technologies. Additionally, through ensemble learning optimization, the recognition rate of oxide scale defects and overall defect classification accuracy were significantly improved.
Article
Multidisciplinary Sciences
Xinglong Feng, Xianwen Gao, Ling Luo
Summary: This study proposed a new hot rolled steel strip defect dataset X-SDD for the actual detection of defects on the surface of hot rolled steel strip. Various algorithms were tested on X-SDD, with the results showing that the proposed algorithm achieved high accuracy and outperformed other comparable algorithms.
Article
Chemistry, Multidisciplinary
Huaping Guo, Shanggui Zhan, Li Zhang, Wenbo Zhu, Yange Sun, Jing Wang
Summary: Strip steel is a crucial material for industries like aerospace, shipbuilding, and pipelines, and any defects in the strip steel can lead to significant economic losses. Detecting these defects is challenging due to the complex variations in strip steel. This paper proposes a novel method based on a U-shaped residual network, utilizing attention mechanisms in the encoder to extract multi-scale defect features and a decoder to capture contextual data. Experimental results demonstrate that this method effectively segments surface defect objects with clear boundaries compared to other advanced techniques.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Artificial Intelligence
Mohamed Ben Gharsallah, Ezzedine Ben Braiek
Summary: Quality control in silicon steel manufacturing is crucial, and image processing techniques have proven useful for defect detection. A novel algorithm based on anisotropic diffusion and saliency map was proposed for defect detection in hot rolled silicon steel images, showing improved accuracy and robustness compared to traditional methods.
JOURNAL OF INTELLIGENT MANUFACTURING
(2021)
Article
Engineering, Multidisciplinary
Guoqi Liu, You Jiang, Baofang Chang, Dong Liu
Summary: The paper proposes a superpixel-based SLSFS method to address the issues in region-based active contour models. The method improves segmentation accuracy under noise and protects weak edge information by generating adaptive initial contour and using improved saliency detection.
Article
Engineering, Multidisciplinary
Bo Liu, Bin Yang, Yelong Zhao, Jianqiang Li
Summary: This paper proposes a deep learning method called low-pass U-Net to improve the segmentation effects of strip steel defects. The method combines low-pass filters and adaptive variance Gaussian low-pass layers to effectively perform defect detection and segmentation. The proposed method achieves considerable performance improvement in practical datasets.
MEASUREMENT SCIENCE AND TECHNOLOGY
(2023)
Article
Chemistry, Multidisciplinary
Shudi Yang, Jiaxiong Wu, Zhipeng Feng
Summary: The research introduces a dual-fusion active contour model with semantic information to improve the accuracy of contour extraction in underwater images. Experimental results show that the proposed model achieves the best results in MAE, ER, and DR indicators, providing reliable prior knowledge for target tracking and visual information mining.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Information Systems
Vivek Kumar Singh, Nitin Kumar
Summary: This paper proposes a Convex Hull Based Random Walks (CoBRa) approach for salient object detection. In this approach, an image is segmented into superpixels and a Convex Hull is constructed to partition the image into foreground and background regions. The centroid of the foreground region is calculated and used to compute initial saliency. Two thresholds are applied to produce binary segmented images, and foreground and background seeds are collected and refined. A random walk is then constructed to generate a pixel-wise saliency map. Experimental results on six datasets demonstrate the superiority of the proposed approach.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Xiaofei Zhou, Hao Fang, Zhi Liu, Bolun Zheng, Yaoqi Sun, Jiyong Zhang, Chenggang Yan
Summary: This article proposes an end-to-end dense attention-guided cascaded network (DACNet) for the detection of salient objects (i.e., defects) on strip steel surfaces. DACNet is a U-shape network consisting of an encoder and a decoder. The encoder deploys multi-resolution convolutional branches in a cascaded way to fuse deep features, while the decoder integrates the multi-scale deep features into the saliency map using a dense attention mechanism. Experimental results show that our model outperforms state-of-the-art models in all evaluation metrics.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Computer Science, Artificial Intelligence
Ruiqi Wu, Feng Zhou, Nan Li, Haibo Liu, Naihong Guo, Rugang Wang
Summary: This paper proposes a lightweight YOLOX surface defect detection network with a Multi-scale Feature Fusion Attention Module (MFFAM). Experimental results show that this method improves the detection frame rate while maintaining accuracy and significantly improves the detection accuracy of small objects.
FRONTIERS IN NEUROROBOTICS
(2022)
Article
Engineering, Multidisciplinary
Ming Tang, Yuanyuan Li, Wei Yao, Lingyu Hou, Qichun Sun, Jiahang Chen
Summary: The study proposes a defect detection method based on attention mechanism and multi-scale maxpooling, using Resnet50 to construct a two-stage detection model. The method was trained and tested on the NEU-DET dataset, achieving a 3.65% mAP performance improvement compared to the baseline network. Additionally, the classification accuracy of the method reaches as high as 94.73%.
MEASUREMENT SCIENCE AND TECHNOLOGY
(2021)
Article
Engineering, Electrical & Electronic
Pradipta Sasmal, M. K. Bhuyan, Shashwata Gupta, Yuji Iwahori
Summary: This article introduces an automatic polyp detection system for endoscopic video frames which utilizes saliency maps and tracking mechanisms for localization. The method achieves high tracking efficiency and segmentation scores, proving to be effective for polyp detection and localization, with promising results in the CVC clinic Database.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2021)
Article
Computer Science, Artificial Intelligence
Shuai Ma, Kechen Song, Hongwen Dong, Hongkun Tian, Yunhui Yan
Summary: This paper proposes a Modal Complementary Fusion Network (MCFNet) to address the issue of sub-optimal detection results caused by direct fusion of low-quality RGB-T images. It introduces a modal reweight module to evaluate the global quality of images and adaptively reweight RGB-T features, as well as a spatial complementary fusion module to selectively fuse multi-modal features. Experimental results demonstrate the outstanding performance of MCFNet compared to state-of-the-art methods.
APPLIED INTELLIGENCE
(2023)
Article
Chemistry, Inorganic & Nuclear
Na Xu, Qianqian Huang, Li Shi, Jia Wang, Xiangrong Li, Wei Guo, Dong Yan, Tianjun Ni, Zhijun Yang, Yunhui Yan
Summary: In this study, PDA@FeS NPs were successfully synthesized using a simple predoping polymerization-coprecipitation strategy, and the intelligent PDA matrix effectively prevented the oxidation and agglomeration of FeS nanoparticles. The PDA@FeS NPs exhibited excellent photothermal antibacterial effects against both E. coli and S. aureus, and the release of ferrous ions under weakly acidic conditions triggered the Fenton reaction to produce toxic hydroxyl radicals, leading to cell membrane damage and cellular content leakage.
DALTON TRANSACTIONS
(2023)
Article
Automation & Control Systems
Han Wang, Kechen Song, Liming Huang, Hongwei Wen, Yunhui Yan
Summary: RGB-T salient object detection has achieved rapid development and excellent results in recent years. However, the current RGB-T datasets lack low-illumination data, leading to poor performance in detecting salient objects in extremely low-illumination scenes. To address this issue, we propose a T-aware guided early fusion network that leverages thermal images to enhance the detection performance of low-illumination data.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Review
Computer Science, Artificial Intelligence
Hongkun Tian, Kechen Song, Song Li, Shuai Ma, Jing Xu, Yunhui Yan
Summary: This paper presents a comprehensive survey of data-driven robotic visual grasping detection (DRVGD) for unknown objects. It reviews both object-oriented and scene-oriented aspects, providing detailed information about associated grasping representations and datasets. The challenges of DRVGD and future directions are also pointed out.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Engineering, Electrical & Electronic
Kechen Song, Yanqi Bao, Han Wang, Liming Huang, Yunhui Yan
Summary: This article introduces a new defect detection technology that combines RGB and thermal infrared images. The proposed information flow fusion network (IFFNet) method can detect surface and internal defects more comprehensively. Experimental results show that our method performs better than existing methods on three RGB-Thermal infrared datasets.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Computer Science, Artificial Intelligence
Hongwei Wen, Kechen Song, Liming Huang, Han Wang, Yunhui Yan
Summary: Cross-modality salient object detection is improved for universality and anti-interference by proposing a network with feature extraction strategy, graph mapping reasoning module (GMRM), and mutual guidance fusion module (MGFM). Experimental results show good performance in universality and anti-interference.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Engineering, Multidisciplinary
Wenli Zhao, Kechen Song, Yanyan Wang, Shubo Liang, Yunhui Yan
Summary: This paper proposes a feature-aware network (FaNet) for few shot defect classification, which can effectively distinguish new classes with a small number of labeled samples. In FaNet, ResNet12 is used as the baseline, and the feature-attention convolution module (FAC) is applied to extract comprehensive feature information from the base classes. An online feature-enhance integration module (FEI) is adopted during the test phase to average the noise from defect images, further enhancing image features among different tasks. In addition, a large-scale strip steel surface defects few shot classification dataset (FSC-20) with 20 different types is constructed. Experimental results show that the proposed method achieves the best performance compared to state-of-the-art methods for 5-way 1-shot and 5-way 5-shot tasks. The dataset and code are available at: https://github.com/VDT-2048/FSC-20.
Review
Materials Science, Multidisciplinary
Xin Wen, Jvran Shan, Yu He, Kechen Song
Summary: Steel surface defect recognition is important in industrial defect detection and has gained increasing attention. This paper discusses the key hardware and options for steel surface defect detection systems, and provides a literature review of algorithms for steel surface defect recognition, including traditional machine learning algorithms based on texture and shape features, as well as supervised, unsupervised, and weakly supervised deep learning algorithms. Common datasets and algorithm performance evaluation metrics in this field are also summarized. Lastly, the challenges and corresponding solutions for current steel surface defect recognition algorithms are discussed, along with the future work focus.
Article
Automation & Control Systems
Hongkun Tian, Kechen Song, Song Li, Shuai Ma, Yunhui Yan
Summary: This paper proposes a framework for rotation adaptive grasping estimation based on a novel RGB-D fusion strategy, which can better utilize the advantages of RGB and depth modes and suppress disadvantages, achieving spatial and rotational adaptiveness to unknown objects and poses. The method has been validated on different datasets, showing excellent accuracy and outperforming other methods.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Automation & Control Systems
Kechen Song, Ying Zhao, Liming Huang, Yunhui Yan, Qinggang Meng
Summary: RGB-T image analysis has gained wide attention and made significant research progress in various applications. This paper provides a comprehensive review of the technology and applications in the fields of image fusion, salient object detection, semantic segmentation, pedestrian detection, object tracking, and person re-identification. It extensively reviews more than 400 papers across over 10 different application tasks, analyzing various methods and presenting the performance of state-of-the-art techniques. Additionally, it offers an in-depth analysis of challenges and potential technical improvements for future RGB-T image analysis.
ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Hongkun Tian, Kechen Song, Jing Xu, Shuai Ma, Yunhui Yan
Summary: This paper proposes an antipodal-points grasping representation model and presents a new network model (APDNet) for grasp detection in multi-object scenes. The proposed method achieves state-of-the-art performance with well-balanced accuracy and efficiency, as demonstrated on a public dataset and real robot platform.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Engineering, Electrical & Electronic
Ying Zhao, Kechen Song, Yiming Zhang, Yunhui Yan
Summary: Few-shot semantic segmentation (FSS) is the task of segmenting target regions of query images using a few labeled support samples. This study introduces thermal infrared images (T) to handle complex outdoor lighting environments and proposes a bidirectional modality difference elimination network (BMDENet) to enhance segmentation performance. The network achieves this by reducing heterogeneity between RGB and thermal images, fusing cross-modal information, and addressing issues in advanced models.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS II-EXPRESS BRIEFS
(2023)
Article
Materials Science, Multidisciplinary
Xin Wen, Wenli Zhao, Zhenhao Yu, Jianxun Zhao, Kechen Song
Summary: This study proposes a novel anomaly detection method based on multi-scale knowledge distillation (Ms-KD) and a block domain core information module (BDCI) to quickly screen abnormal images in the surface inspection of strip steel. By utilizing the multi-scale knowledge distillation technique and the optimal storage of block-level features, the proposed method enables the student network to learn normal image information and better capture abnormal data to solve the imbalance problem. Experimental results showed the effectiveness of this method in strip steel defect anomaly detection, achieving high performance in image-level AUROC and pixel-level PRO indicators compared to state-of-the-art methods.
Article
Computer Science, Artificial Intelligence
Shuai Ma, Kechen Song, Menghui Niu, Hongkun Tian, Yanyan Wang, Yunhui Yan
Summary: This paper proposes a feature-based domain disentanglement and randomization (FDDR) framework to improve the generalization of deep models in unseen datasets. The framework successfully addresses the appearance difference issue between training and test images by decomposing the defect image into domain-invariant structural features and domain-specific style features. It also utilizes randomly generated samples for training to further expand the training sample.
ADVANCED ENGINEERING INFORMATICS
(2024)
Article
Engineering, Multidisciplinary
Tonglei Cao, Kechen Song, Likun Xu, Hu Feng, Yunhui Yan, Jingbo Guo
Summary: This study constructs a high-resolution dataset for surface defects in ceramic tiles and addresses the scale and quantity differences in defect distribution. An improved approach is proposed by introducing a content-aware feature recombination method and a dynamic attention mechanism. Experimental results demonstrate the superior accuracy and efficiency of the proposed method.