YOLOv5n-ShuffleNetv2: A Lightweight Transmission Line Insulator Defect Detection Algorithm

Wanbo Luo
Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, PMLR 278:181-188, 2025.

Abstract

Insulators represent a pivotal component of transmission lines, with their functionality directly impacting the reliable operation of the power grid. Current insulator defect detection algorithms face significant challenges, including low accuracy and long latency, which hinder their practical application in the timely and reliable maintenance of power systems. Therefore, this paper proposed a lightweight detector to minimize the model’s parameters and calculations. First, the YOLOv5n algorithm was chosen as the foundation for the detection system’s lightweight design. Second, the ShuffleNetv2 backbone replaced the YOLOv5n backbone to further lightweight the model. Experimental results showed that the proposed YOLOv5n-ShuffleNetv2 model achieved 84.5% mAP50 at 87.7 FPS using the Nvidia Jetson Orin Nano 4G. Although the accuracy decreased by 4.1%, the model achieved a 44.7% reduction in the number of parameters and a 22% increase in detection speed, demonstrating a significant improvement in efficiency and practicality.

Cite this Paper

BibTeX
@InProceedings{pmlr-v278-luo25a, title = {YOLOv5n-ShuffleNetv2: A Lightweight Transmission Line Insulator Defect Detection Algorithm}, author = {Luo, Wanbo}, booktitle = {Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing}, pages = {181--188}, year = {2025}, editor = {Zeng, Nianyin and Pachori, Ram Bilas and Wang, Dongshu}, volume = {278}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v278/main/assets/luo25a/luo25a.pdf}, url = {https://proceedings.mlr.press/v278/luo25a.html}, abstract = {Insulators represent a pivotal component of transmission lines, with their functionality directly impacting the reliable operation of the power grid. Current insulator defect detection algorithms face significant challenges, including low accuracy and long latency, which hinder their practical application in the timely and reliable maintenance of power systems. Therefore, this paper proposed a lightweight detector to minimize the model’s parameters and calculations. First, the YOLOv5n algorithm was chosen as the foundation for the detection system’s lightweight design. Second, the ShuffleNetv2 backbone replaced the YOLOv5n backbone to further lightweight the model. Experimental results showed that the proposed YOLOv5n-ShuffleNetv2 model achieved 84.5% mAP50 at 87.7 FPS using the Nvidia Jetson Orin Nano 4G. Although the accuracy decreased by 4.1%, the model achieved a 44.7% reduction in the number of parameters and a 22% increase in detection speed, demonstrating a significant improvement in efficiency and practicality.} }
Endnote
%0 Conference Paper %T YOLOv5n-ShuffleNetv2: A Lightweight Transmission Line Insulator Defect Detection Algorithm %A Wanbo Luo %B Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing %C Proceedings of Machine Learning Research %D 2025 %E Nianyin Zeng %E Ram Bilas Pachori %E Dongshu Wang %F pmlr-v278-luo25a %I PMLR %P 181--188 %U https://proceedings.mlr.press/v278/luo25a.html %V 278 %X Insulators represent a pivotal component of transmission lines, with their functionality directly impacting the reliable operation of the power grid. Current insulator defect detection algorithms face significant challenges, including low accuracy and long latency, which hinder their practical application in the timely and reliable maintenance of power systems. Therefore, this paper proposed a lightweight detector to minimize the model’s parameters and calculations. First, the YOLOv5n algorithm was chosen as the foundation for the detection system’s lightweight design. Second, the ShuffleNetv2 backbone replaced the YOLOv5n backbone to further lightweight the model. Experimental results showed that the proposed YOLOv5n-ShuffleNetv2 model achieved 84.5% mAP50 at 87.7 FPS using the Nvidia Jetson Orin Nano 4G. Although the accuracy decreased by 4.1%, the model achieved a 44.7% reduction in the number of parameters and a 22% increase in detection speed, demonstrating a significant improvement in efficiency and practicality.
APA
Luo, W.. (2025). YOLOv5n-ShuffleNetv2: A Lightweight Transmission Line Insulator Defect Detection Algorithm. Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, in Proceedings of Machine Learning Research 278:181-188 Available from https://proceedings.mlr.press/v278/luo25a.html.

Related Material