Improved MViTv2-T model for insulator defect detection

Fuhong Meng; Guowu Yuan; Hao Zhou; Hao Wu; Yi Ma; Fuhong Meng; Guowu Yuan; Hao Zhou; Hao Wu; Yi Ma

doi:10.3934/electreng.2025001

AIMS Electronics and Electrical Engineering

2025, Volume 9, Issue 1: 1-25. doi: 10.3934/electreng.2025001

Previous Article Next Article

Research article Special Issues

Improved MViTv2-T model for insulator defect detection

1.
School of Information Science and Engineering, Yunnan University, Kunming 650504, Yunnan, China
2.
Electric Power Research Institute, Yunnan Power Grid Co., Ltd, Kunming 650214, Yunnan, China

Academic Editor: Qichun Zhang

Received: 05 September 2024 Revised: 04 November 2024 Accepted: 18 November 2024 Published: 04 December 2024

Insulators play a crucial role in transmission lines. Insulators exposed to natural environments are prone to various malfunctions. These faults will seriously affect the safety and stability of the power grid system operation, so intelligent detection of insulator defects has become increasingly important. This paper presents an insulator defect detection model based on the improved MViTv2-T (Multiscale Vision Transformers Version 2 Tiny). The new model utilizes the sore penalty mechanism (SPM) cluster non-maximum suppression (NMS) algorithm instead of the batched non-maximum suppression (NMS) algorithm from the original model. Additionally, it introduces the stage query recollection method, which integrates high-level and low-level module queries within each stage, along with various experimentation on integration functions between the two. The experimental results indicate that the improved MViTv2-T model attains an mAP (mean average precision)@0.5:0.95 of 76.1$ \% $, mAP@0.5 of 96.1$ \% $, and mAR@0.5 of 97.2$ \% $ in insulator defect detection. Compared to the original model, there is a 1.8$ \% $ increase in mAP@0.5:0.95 and a 17$ \% $ decrease in the detection error rate at an Intersection over Union (IoU) threshold of 0.5. Furthermore, when compared to standard two-stage detection models and YOLO series models, the improved MViTv2-T model also exhibits distinct performance advantages.
- insulator,
- deep learning,
- defect detection,
- ViT,
- MViTv2
Citation: Fuhong Meng, Guowu Yuan, Hao Zhou, Hao Wu, Yi Ma. Improved MViTv2-T model for insulator defect detection[J]. AIMS Electronics and Electrical Engineering, 2025, 9(1): 1-25. doi: 10.3934/electreng.2025001

Related Papers:

Abstract

Insulators play a crucial role in transmission lines. Insulators exposed to natural environments are prone to various malfunctions. These faults will seriously affect the safety and stability of the power grid system operation, so intelligent detection of insulator defects has become increasingly important. This paper presents an insulator defect detection model based on the improved MViTv2-T (Multiscale Vision Transformers Version 2 Tiny). The new model utilizes the sore penalty mechanism (SPM) cluster non-maximum suppression (NMS) algorithm instead of the batched non-maximum suppression (NMS) algorithm from the original model. Additionally, it introduces the stage query recollection method, which integrates high-level and low-level module queries within each stage, along with various experimentation on integration functions between the two. The experimental results indicate that the improved MViTv2-T model attains an mAP (mean average precision)@0.5:0.95 of 76.1$ \% $, mAP@0.5 of 96.1$ \% $, and mAR@0.5 of 97.2$ \% $ in insulator defect detection. Compared to the original model, there is a 1.8$ \% $ increase in mAP@0.5:0.95 and a 17$ \% $ decrease in the detection error rate at an Intersection over Union (IoU) threshold of 0.5. Furthermore, when compared to standard two-stage detection models and YOLO series models, the improved MViTv2-T model also exhibits distinct performance advantages.

References

[1]	Zhao ZB, Jiang ZG, Li YX, Qi YC, Zhai YJ, Zhao WQ, et al. (2021) Overview of visual defect detection of transmission line components. Journal of Image and Graphics 26: 2545–2560. https://doi.org/10.11834/jig.200689 doi: 10.11834/jig.200689
[2]	Chen C, Yuan GW, Zhou H, Ma Y (2023) Improved YOLOv5s model for key components detection of power transmission lines. Math Biosci Eng 20: 7738–7761. https://doi.org/10.3934/mbe.2023334 doi: 10.3934/mbe.2023334
[3]	Liu HY, Yuan GW (2022) Cigarette appearance defect detection method based on improved YOLOv5s. Comput Technol Dev 32: 161–167. https://doi.org/10.3969/j.issn.1673-629X.2022.08.026 doi: 10.3969/j.issn.1673-629X.2022.08.026
[4]	Zhang Y, Dou Y, Yang K, Song X, Wang J, Zhao L (2024) Insulator defect detection based on BaS-YOLOv5. Multimed Syst 30: 212. https://doi.org/10.1007/s00530-024-01413-w doi: 10.1007/s00530-024-01413-w
[5]	Chen H, He Z, Shi B, Zhong T (2019) Research on recognition method of electrical components based on YOLO V3. IEEE Access 7: 157818–157829. https://doi.org/10.1109/ACCESS.2019.2950053 doi: 10.1109/ACCESS.2019.2950053
[6]	Su J, Yuan Y, Przystupa K, Kochan O (2024) Insulator defect detection algorithm based on improved YOLOv8 for electric power. Signal Image Video Process 18: 6197–6209. https://doi.org/10.1007/s11760-024-03307-w doi: 10.1007/s11760-024-03307-w
[7]	Li D, Yang P, Zou Y (2024) Optimizing Insulator Defect Detection with Improved DETR Models. Mathematics 12: 1507. https://doi.org/10.3390/math12101507 doi: 10.3390/math12101507
[8]	Yuan H, Wang J (2023) Power Insulator Defect Detection Based on Multi-scale Dense Adaptive Sensing. Mathematics 2661: 012001. https://doi.org/10.1088/1742-6596/2661/1/012001 doi: 10.1088/1742-6596/2661/1/012001
[9]	Zhang T, Zhong S, Xu W, Yan L, Zou X (2024) Catenary Insulator Defects Detection: A Dataset and an Unsupervised Baseline. IEEE T Instrum Meas 73: 1–15. https://doi.org/10.1109/TIM.2024.3390695 doi: 10.1109/TIM.2024.3390695
[10]	Wang S, Liu Y, Qing Y, Wang C, Lan T, Yao R (2020) Detection of insulator defects with improved ResNeSt and region proposal network. IEEE Access 8: 184841–184850. https://doi.org/10.1109/ACCESS.2020.3029857 doi: 10.1109/ACCESS.2020.3029857
[11]	Mei H, Jiang H, Chen J, Yin F, Wang L, Farzaneh M (2021) Detection of internal defects of full-size composite insulators based on microwave technique. IEEE T Instrum Meas 70: 1–10. https://doi.org/10.1109/TIM.2021.3085111 doi: 10.1109/TIM.2021.3085111
[12]	Cao Z, Chen K, Chen J, Chen Z, Zhang M (2024) CACS-YOLO: A Lightweight Model for Insulator Defect Detection based on Improved YOLOv8m. IEEE T Instrum Meas 73: 1–10. https://doi.org/10.1109/TIM.2024.3453332 doi: 10.1109/TIM.2024.3453332
[13]	Han G, Yuan Q, Zhao F, Wang R, Zhao L, Li S, et al. (2023) An improved algorithm for insulator and defect detection based on yolov4. Electronics 12: 933. https://doi.org/10.3390/ELECTRONICS12040933 doi: 10.3390/ELECTRONICS12040933
[14]	Zhao ZB, Li Y, Qi YC, Kong YH, Nie LQ (2020) Insulator defect detection method based on dynamic focus loss function and sample balance method. Electric Power Automation Equipment 40: 205–211. https://doi.org/10.16081/j.epae.202010008 doi: 10.16081/j.epae.202010008
[15]	Guo L, Liao Y, Yao H, Chen J, Wang M (2018) An electrical insulator defects detection method combined human receptive field model. J Control Sci Eng 2018: 2371825. https://doi.org/10.1155/2018/2371825 doi: 10.1155/2018/2371825
[16]	Li T, Hao T (2022) Damage detection of insulators in catenary based on deep learning and Zernike moment algorithms. Appl Sci 12: 5004. https://doi.org/10.3390/app12105004 doi: 10.3390/app12105004
[17]	Qi Y, Li Y, Du A (2023) Research on an insulator defect detection method based on improved yolov5. Appl Sci 13: 5741. https://doi.org/10.3390/app13095741 doi: 10.3390/app13095741
[18]	Zhang H, Huang G, Yang L (2023) Insulator defect detection algorithm based on multi-scale feature fusion optimization. International Conference on Algorithms, High Performance Computing, and Artificial Intelligence (AHPCAI 2023) 12941: 226–232. https://doi.org/10.1117/12.3011642 doi: 10.1117/12.3011642
[19]	Fan H, Xiong B, Mangalam K, Li Y, Yan Z, Malik J, et al. (2021) Multiscale vision transformers. Proceedings of the IEEE/CVF international conference on computer vision 2021: 6824–6835. https://doi.org/10.48550/arXiv.2104.11227 doi: 10.48550/arXiv.2104.11227
[20]	Li Y, Wu CY, Fan H, Mangalam K, Xiong B, Malik J, et al. (2022) Mvitv2: Improved multiscale vision transformers for classification and detection. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition 2022: 4804–4814. https://doi.org/10.48550/arXiv.2112.01526 doi: 10.48550/arXiv.2112.01526
[21]	Zheng Z, Wang P, Ren D, Liu W, Ye R, Hu Q, et al. (2021) Enhancing geometric factors in model learning and inference for object detection and instance segmentation. IEEE Trans Cybern 52: 8574–8586. https://doi.org/10.1109/TCYB.2021.3095305 doi: 10.1109/TCYB.2021.3095305
[22]	Lou M, Zhou HY, Yang S, Yu Y (2023) TransXNet: learning both global and local dynamics with a dual dynamic token mixer for visual recognition. arXiv preprint arXiv: 2310.19380. https://doi.org/10.48550/arXiv.2310.19380
[23]	Bodla N, Singh B, Chellappa R, Davis LS (2017) Soft-NMS–improving object detection with one line of code. Proceedings of the IEEE international conference on computer vision 2017: 5561–5569. https://doi.org/10.48550/arXiv.1704.04503 doi: 10.48550/arXiv.1704.04503
[24]	Zheng Z, Wang P, Liu W, Li J, Ye R, Ren D (2020) Distance-IoU loss: Faster and better learning for bounding box regression. Proceedings of the AAAI conference on artificial intelligence 34: 12993–13000. https://doi.org/10.1609/aaai.v34i07.6999 doi: 10.1609/aaai.v34i07.6999
[25]	Bolya D, Zhou C, Xiao F, Lee YJ (2019) Yolact: Real-time instance segmentation. Proceedings of the IEEE/CVF international conference on computer vision 2019: 9157–9166. https://doi.org/10.48550/arXiv.1904.02689 doi: 10.48550/arXiv.1904.02689

Reader Comments

Your name:*

Email:*
© 2025 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)