![](bg6.png)
Fig.6. illustrates a side-by-side comparison of the
detection results for samples in each defect category.
The results of crazing defect detection clearly
demonstrate that the algorithm presented in this paper
excels in accurately locating the detection bounding
box when compared to YOLOv5s. In the case of
inclusion defect detection, it is evident that our
improved algorithm surpasses the baseline model,
offering more robust detection capabilities and
outstanding performance for small targets.
4 CONCLUSION
To tackle the challenges associated with detecting
surface defects on strip steel within the context of
automated production, this paper introduces an
enhancement strategy based on the YOLOv5s
algorithm. Firstly, we propose an enhanced C3
module aimed at augmenting the model's feature
extraction capabilities. Secondly, we incorporate a
multi-scale pyramidal detection head to bolster the
model's proficiency in detecting small targets. Lastly,
we adopt the SIoU loss function to expedite the
model's convergence speed during training, thereby
improving its overall performance in defect detection.
The experimental results show that compared with the
benchmark model, the method proposed in this paper
effectively improves the detection accuracy and the
detection effect for small targets is improved
significantly. Furthermore, we plan to utilize
techniques like pruning and distillation to reduce the
model's size, facilitating its deployment on embedded
edge devices, thereby expanding its practical
applicability.
ACKNOWLEDGMENTS
This work was financially supported by Fujian
Provincial Natural Science Foundation Project
(2022J011247).
REFERENCES
Lei Yang, Shuai Xu, Junfeng Fan, En Li, Yanhong Liu. A
pixel-level deep segmentation network for automatic
defect detection [J].Expert Systems with Applications,
2023, 215: 119388.
https://doi.org/10.1016/j.eswa.2022.119388
Fityanul Akhyar, Ying Liu, Chao-Yung Hsu, Timothy K.
Shih, Chih-Yang Lin. FDD: a deep learning–based steel
defect detectors [J].The International Journal of
Advanced Manufacturing Technology, 2023: 1-15.
https://doi.org/10.1007/s00170-023-11087-9
Jianbo Yu, Xun Cheng, Qingfeng Li. Surface defect
detection of steel strips based on anchor-free network
with channel attention and bidirectional feature
fusion[J]. IEEE Transactions on Instrumentation and
Measurement, 2021, 71: 1-10.
https://doi.org/10.1109/TIM.2021.3136183
Bo Tang, Zi-Kai Song, Wei Sun, Xing-Dong Wang. An
end-to-end steel surface defect detection approach via
Swin transformer [J].IET Image Processing, 2023,
17(5): 1334-1345. https://doi.org/10.1049/ipr2.12715
Muwei Jian, Haodong Jin, Xiangyu Liu, Linsong Zhang.
Multiscale Cascaded Attention Network for Saliency
Detection Based on ResNet [J].Sensors, 2022, 22(24):
9950. https://doi.org/10.3390/s22249950
Chao Zhao, Xin Shu, Xi Yan, Xin Zuo, Feng Zhu. RDD-
YOLO: A modified YOLO for detection of steel surface
defects [J]. Measurement, 2023: 112776.
https://doi.org/10.1016/j.measurement.2023.112776
Marcelo Gennari, Roger Fawcett, Victor Adrian Prisacariu.
DSConv: Efficient Convolution Operator [J]. arXiv
preprint arXiv:1901.01928,
2019.https://doi.org/10.48550/arXiv.1901.01928
Jie Hu, Li Shen, Samuel Albanie, Gang Sun, Enhua Wu.
Squeeze-and-excitation networks[C]//Proceedings of
the IEEE conference on computer vision and pattern
recognition. 2018: 7132-7141.
https://doi.org/10.48550/arXiv.1709.01507
Chengqi Lyu, Wenwei Zhang, Haian Huang, et al.
RTMDet: An Empirical Study of Designing Real-Time
Object Detectors [J]. arXiv preprint arXiv:2212.07784,
2022.https://doi.org/10.48550/arXiv.2212.07784
Ionut Cosmin Duta, Li Liu, Fan Zhu, Ling Shao. Pyramidl
convolution: Rethinking convolutional neural networks
for visual recognition [J]. arXiv preprint
arXiv:2006.11538, 2020.
https://doi.org/10.48550/arXiv.2006.11538
Sasha Targ, Diogo Almeida, Kevin Lyman. Resnet in
resnet: Generalizing residual architectures [J]. arXiv
preprint arXiv:1603.08029, 2016.
https://doi.org/10.48550/arXiv.1603.08029
Zhaohui Zheng, Ping Wang, Wei Liu, Jinze Li, Rongguang
Ye, Dongwei Ren. Distance-IoU loss: Faster and better
learning for bounding box regression[C]//Proceedings
of the AAAI conference on artificial intelligence. 2020,
34(07): 12993-13000.
https://doi.org/10.48550/arXiv.1911.08287
Zhora Gevorgyan. SIoU loss: More powerful learning for
bounding box regression [J]. arXiv preprint
arXiv:2205.12740, 2022.
https://doi.org/10.48550/arXiv.2205.12740
Kai Chen, Jiaqi Wang, Jiangmiao Pang, et al.
MMDetection: Open MMLab Detection Toolbox and
Benchmark [J]. arXiv, 2019.
https://doi.org/10.48550/arXiv.1906.07155
ANIT 2023 - The International Seminar on Artificial Intelligence, Networking and Information Technology
380