Copper Nodule Defect Detection in Industrial Processes Using Deep Learning DOI Creative Commons
Zhicong Zhang, Xin Huang, Dandan Wei

и другие.

Information, Год журнала: 2024, Номер 15(12), С. 802 - 802

Опубликована: Дек. 11, 2024

Copper electrolysis is a crucial process in copper smelting. The surface of cathodic plates often affected by various electrolytic factors, resulting the formation nodule defects that significantly impact quality and disrupt downstream production process, making prompt detection these essential. At present, cathode plate nodules performed manual identification. In order to address issues with convex identification on industrial terms low accuracy, high effort, efficiency manufacturing lightweight YOLOv5 model combined BiFormer attention mechanism proposed this paper. employs MobileNetV3, feature extraction network, as its backbone, reducing parameter count computational complexity. Additionally, an introduced capture multi-scale information, thereby enhancing accuracy recognition. Meanwhile, F-EIOU loss function employed strengthen model’s robustness generalization ability, effectively addressing noise imbalance data. Experimental results demonstrate improved achieves precision 92.71%, recall 91.24%, mean average (mAP) 92.69%. Moreover, single-frame time 4.61 ms achieved model, which has size 2.91 MB. These metrics meet requirements practical provide valuable insights for process.

Язык: Английский

Adaptive shape imitation and selective semantic guidance for industrial surface defect detection DOI
Xiao Liang,

Y Y Li,

Xuewei Wang

и другие.

Expert Systems with Applications, Год журнала: 2025, Номер unknown, С. 127334 - 127334

Опубликована: Март 1, 2025

Язык: Английский

Процитировано

0
An efficient and scale-aware zero-shot industrial anomaly detection technique based on optimized CLIP DOI
Yahui Cheng,

Guojun Wen,

Aoshuang Luo

и другие.

Measurement, Год журнала: 2025, Номер unknown, С. 117443 - 117443

Опубликована: Апрель 1, 2025

Язык: Английский

Процитировано

0
DCUE-YOLO: A Lightweight Model in Industrial Defect Detection DOI Creative Commons

Jiajin Zhong,

H. Wang, J. H. Zou

и другие.

Research Square (Research Square), Год журнала: 2025, Номер unknown

Опубликована: Янв. 13, 2025

Abstract Accurate and rapid identification of defects in industrial products is essential for ensuring quality safety. However, the challenges presented by large-scale production environments, along with difficulty distinguishing between target complex backgrounds, complicate defect detection. Consequently, most detection models struggle to achieve an optimal balance accuracy efficiency. To improve efficiency, this paper proposes a lightweight network architecture, DCUE-YOLO, based on YOLOv10. The primary objective both efficiency product In addition, feature extraction module double convolutional path design hidden channels proposed under premise reducing computational complexity; capturing information different scales, model can enhance ability distinguish small from backgrounds. order further model's attention defects, also multifilter mechanism design. Meanwhile, effectively solve problem partial loss process downsampling, uses transposed convolution Extensive experiments were carried out using PCB, NEU-DET mixed-type WM38 public data sets, producing mean average precision (mAP) scores 94.3%, 90.5%, 98.7%, respectively. Compared YOLOv10s model, our mAP has improved 2.7%, 1.8%, 1.2%, respectively, while parameter count decreased 0.3M. Our demonstrates advantages recognition inference speed, thus validating its effectiveness

Язык: Английский

Процитировано

0
CTL-YOLO: A Surface Defect Detection Algorithm for Lightweight Hot-Rolled Strip Steel Under Complex Backgrounds DOI Creative Commons
Wenzheng Sun, Meng Na,

Longfa Chen

и другие.

Machines, Год журнала: 2025, Номер 13(4), С. 301 - 301

Опубликована: Апрель 7, 2025

Currently, in the domain of surface defect detection on hot-rolled strip steel, detecting small-target defects under complex background conditions and effectively balancing computational efficiency with accuracy presents a significant challenge. This study proposes CTL-YOLO based YOLO11, aimed at efficiently accurately blemishes steel industrial applications. Firstly, CGRCCFPN feature integration network is proposed to achieve multi-scale global fusion while preserving detailed information. Secondly, TVADH Detection Head identify textured backgrounds. Finally, LAMP algorithm used further compress network. The demonstrates excellent performance public dataset NEU-DET, achieving mAP50 77.6%, representing 3.2 percentage point enhancement compared baseline algorithm. GFLOPs reduced 2.0, 68.3% decrease baseline, Params are 0.40, showing an 84.5% reduction. Additionally, it exhibits strong generalization capabilities GC10-DET. can improve maintaining lightweight design.

Язык: Английский

Процитировано

0
Copper Nodule Defect Detection in Industrial Processes Using Deep Learning DOI Creative Commons
Zhicong Zhang, Xin Huang, Dandan Wei

и другие.

Information, Год журнала: 2024, Номер 15(12), С. 802 - 802

Опубликована: Дек. 11, 2024

Copper electrolysis is a crucial process in copper smelting. The surface of cathodic plates often affected by various electrolytic factors, resulting the formation nodule defects that significantly impact quality and disrupt downstream production process, making prompt detection these essential. At present, cathode plate nodules performed manual identification. In order to address issues with convex identification on industrial terms low accuracy, high effort, efficiency manufacturing lightweight YOLOv5 model combined BiFormer attention mechanism proposed this paper. employs MobileNetV3, feature extraction network, as its backbone, reducing parameter count computational complexity. Additionally, an introduced capture multi-scale information, thereby enhancing accuracy recognition. Meanwhile, F-EIOU loss function employed strengthen model’s robustness generalization ability, effectively addressing noise imbalance data. Experimental results demonstrate improved achieves precision 92.71%, recall 91.24%, mean average (mAP) 92.69%. Moreover, single-frame time 4.61 ms achieved model, which has size 2.91 MB. These metrics meet requirements practical provide valuable insights for process.

Язык: Английский

Процитировано

0