|
|
An improved defect recognition framework for casting based on DETR algorithm |
Long Zhang1,2,3, Sai-fei Yan2, Jun Hong3, Qian Xie2,4, Fei Zhou3, Song-lin Ran |
1 Anhui Province Key Laboratory of Metallurgical Engineering & Resources Recycling, Anhui University of Technology, Ma’anshan 243002, Anhui, China 2 School of Metallurgical Engineering, Anhui University of Technology, Ma’anshan 243002, Anhui, China 3 Technical Department, Anhui Highly Precision Casting Co., Ltd., Ma’anshan 238100, Anhui, China 4 Anhui Engineering Laboratory for Intelligent Applications and Security of Industrial Internet, Anhui University of Technology, Ma’anshan 243032, Anhui, China |
|
|
Abstract The current casting surface defect detection algorithms suffer from poor small target defect recognition and imbalance between detection performance and detection time. An improved algorithmic framework for casting defect detection was proposed based on the DEtection TRansformer (DETR) algorithm. The algorithm takes ResNet with an efficient channel attention (ECA)-Net module as the backbone network. In addition, based on the original algorithm architecture, dynamic anchor boxes, improved multi-scale deformable attention module, and SIoU loss function are introduced to improve the sensitivity of transformer structure to input location information and scale size, and the small target defect detection performance is effectively improved. The recognition performance of the algorithm in a self-built casting defect dataset was studied. The improved DETR algorithm has 97.561% accuracy in recognizing two defects, namely sandinclusion and notch, with the detection rate being improved by 65.854% and 17.073% compared with the original DETR and you only look once (Yolo)-V5, respectively. This algorithm verifies the applicability of the transformer architecture target detection algorithm for casting defect detection tasks and provides new ideas for detecting other similar application scenarios.
|
|
|
|
|
Cite this article: |
Long Zhang,Sai-fei Yan,Jun Hong, et al. An improved defect recognition framework for casting based on DETR algorithm[J]. Journal of Iron and Steel Research International, 2023, 30(05): 949-959.
|
|
|
|
[1] |
Guang-da Bao, Ting Wu, Duo-gang Wang, Xiao-bin Zhou, Hai-chuan Wang. Multi-model coupling-based dynamic control system of ladle slag in argon blowing refining process[J]. JOURNAL OF IRON AND STEEL RESEARCH,INTERNATIONAL, 2023, 30(05): 926-936. |
[2] |
Gong-hao Lian, Qi-hao Sun, Xiao-ming Liu, Wei-miao Kong, Ming Lv, Jian-jun Qi, Yong Liu, Ben-ming Yuan, Qiang Wang. Automatic recognition and intelligent analysis of central shrinkage defects of continuous casting billets based on deep learning[J]. JOURNAL OF IRON AND STEEL RESEARCH,INTERNATIONAL, 2023, 30(05): 937-948. |
[3] |
Xuan-dong Wang, Nan Li, Hang Su, Hui-min Meng. Prior austenite grain boundary recognition in martensite microstructure based on deep learning[J]. JOURNAL OF IRON AND STEEL RESEARCH,INTERNATIONAL, 2023, 30(05): 1050-1056. |
[4] |
Bing Han, Wei-hao Wan, Dan-dan Sun, Cai-chang Dong, Lei Zhao, Hai-zhou Wang. A deep learning-based method for segmentation and quantitative characterization of microstructures in weathering steel from sequential scanning electron microscope images[J]. JOURNAL OF IRON AND STEEL RESEARCH,INTERNATIONAL, 2022, 29(5): 836-845. |
[5] |
Wei‑gang Li, Lu Xie, Yun‑tao Zhao, Zi‑xiang Li, Wen‑bo Wang. Prediction model for mechanical properties of hot-rolled strips by deep learning[J]. JOURNAL OF IRON AND STEEL RESEARCH,INTERNATIONAL, 2020, 27(9): 1045-1053. |
|
|
|
|