A thermal image detection method for insulators incorporating within-class sparse prior knowledge and improved YOLOv8

doi:10.11996/JG.j.2095-302X.2025061247

Abstract

Abstract:

To achieve high-precision detection of low-resolution thermal image of insulators, an infrared multi-scale insulator detection method was proposed based on intra-class sparse prior and improved YOLOv8. To address the issue of missed detections due to local density in insulator images, a universal intra-class sparse prior was proposed, combining prior knowledge with training data. This enabled the model to perceive the unique geometric features and morphological information of the insulator, enhancing the accuracy of the target detection model without additional computational cost, and provided a standardized annotation method for insulator data samples. To tackle the difficulty of feature extraction from low-resolution infrared images, a robust feature downsampling module was employed to replace convolutional downsampling, preserving fine-grained detail information and enhancing the robust representation of key feature maps. For the problem of large-scale variations and occlusion in insulator targets, a wise-MPDIoU was utilized, improving the bounding-box loss function and the model’s ability to localize insulators of different sizes. Experimental data demonstrated that, compared to the baseline model, the proposed method achieved improvements of 3.3 and 3.5 percentage points in AP50 and AP50:95 metrics, respectively, providing a new solution for insulator thermal image detection.

Key words: insulator, thermal image, within-class sparse prior knowledge, object detection, deep learning

CLC Number:

ZHAO Zhenbing, Ouyang Wenbin, FENG Shuo, LI Haopeng, MA Jun. A thermal image detection method for insulators incorporating within-class sparse prior knowledge and improved YOLOv8[J]. Journal of Graphics, 2025, 46(6): 1247-1256.

Figures/Tables 13

References 44

[1]	和敬涵, 罗国敏, 程梦晓, 等. 新一代人工智能在电力系统故障分析及定位中的研究综述[J]. 中国电机工程学报, 2020, 40(17): 5506-5515.
	HE J H, LUO G M, CHENG M X, et al. A research review on application of artificial intelligence in power system fault analysis and location[J]. Proceedings of the CSEE, 2020, 40(17): 5506-5515 (in Chinese).
[2]	徐舒玮, 邱才明, 张东霞, 等. 基于深度学习的输电线路故障类型辨识[J]. 中国电机工程学报, 2019, 39(1): 65-74.
	XU S W, QIU C M, ZHANG D X, et al. A deep learning approach for fault type identification of transmission line[J]. Proceedings of the CSEE, 2019, 39(1): 65-74 (in Chinese).
[3]	赵振兵, 冯烁, 赵文清, 等. 融合知识迁移和改进YOLOv6的变电设备热像检测方法[J]. 智能系统学报, 2023, 18(6): 1213-1222.
	ZHAO Z B, FENG S, ZHAO W Q, et al. Thermd image detection method for substation equipment by incorporating knowledge migration and improved YOLOv6[J]. CAAI Transactions on Intelligent Systems, 2023, 18(6): 1213-1222 (in Chinese).
[4]	张血琴, 周志鹏, 郭裕钧, 等. 不同材质绝缘子污秽等级高光谱检测方法研究[J]. 电工技术学报, 2023, 38(7): 1946-1955.
	ZHANG X Q, ZHOU Z P, GUO Y J, et al. Detection method of contamination grades of insulators with different materials based on hyperspectral technique[J]. Transactions of China Electrotechnical Society, 2023, 38(7): 1946-1955 (in Chinese).
[5]	赵振兵, 孔英会, 戚银城, 等. 面向智能输变电的图像处理技术[M]. 北京: 中国电力出版社, 2014: 1-4.
	ZHAO Z B, KONG Y H, QI Y C, et al. Image processing technology for intelligent power transmission and transformation[M]. Beijing: China Electric Power Press, 2014: 1-4 (in Chinese).
[6]	刘传洋, 吴一全. 基于红外图像的电力设备识别及发热故障诊断方法研究进展[J]. 中国电机工程学报, 2025, 45(6): 2171-2195.
	LIU C Y, WU Y Q. Research progress of power equipment identification and thermal fault diagnosis based on infrared images[J]. Proceedings of the CSEE, 2025, 45(6): 2171-2195 (in Chinese).
[7]	ZHENG H B, CUI Y H, YANG W Q, et al. An infrared image detection method of substation equipment combining iresgroup structure and CenterNet[J]. IEEE Transactions on Power Delivery, 2022, 37(6): 4757-4765. DOI URL
[8]	ZHOU X Y, WANG D Q, KRÄHENBÜHL P. Objects as points[EB/OL]. [2024-08-20]. https://arxiv.org/abs/1904.07850.
[9]	HE H Y, HU Z, WANG B Z, et al. A contactless zero-value insulators detection method based on infrared images matching[J]. IEEE Access, 2020, 8: 133882-133889. DOI URL
[10]	ZHOU S J, LIU J F, FAN X H, et al. Thermal fault diagnosis of electrical equipment in substations using lightweight convolutional neural network[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72: 5005709.
[11]	RONNEBERGER O, FISCHER P, BROX T. U-net: convolutional networks for biomedical image segmentation[C]// The 18th International Conference on Medical Image Computing and Computer-Assisted Intervention. Cham: Springer, 2015: 234-241.
[12]	YIN J G, LU Y P, GONG Z X, et al. Edge detection of high-voltage porcelain insulators in infrared image using dual parity morphological gradients[J]. IEEE Access, 2019, 7: 32728-32734. DOI URL
[13]	ZHAO Z B, FENG S, MA D Y, et al. A weakly supervised instance segmentation approach for insulator thermal images incorporating sparse prior knowledge[J]. IEEE Transactions on Power Delivery, 2024, 39(5): 2693-2703. DOI URL
[14]	赵振兵, 郭广学, 王艺衡, 等. 融合边缘感知与统计纹理知识的输电线路金具锈蚀检测[J]. 智能系统学报, 2024, 19(5): 1228-1237.
	ZHAO Z B, GUO G X, WANG Y H, et al. Rust detection in transmission line fittings via fusion of edge perception and statistical texture knowledge[J]. CAAI Transactions on Intelligent Systems, 2024, 19(5): 1228-1237 (in Chinese).
[15]	李刚, 蔡泽浩, 孙华勋, 等. 基于改进YOLOv8与语义知识融合的金具缺陷检测方法研究[J]. 图学学报, 2024, 45(5): 979-986. DOI
	LI G, CAI Z H, SUN H X, et al. Research on defect detection of transmission line fittings based on improved YOLOv8 and semantic knowledge fusion[J]. Journal of Graphics, 2024, 45(5): 979-986 (in Chinese). DOI
[16]	HAO S, AN B Y, MA X, et al. PKAMNet: a transmission line insulator parallel- gap fault detection network based on prior knowledge transfer and attention mechanism[J]. IEEE Transactions on Power Delivery, 2023, 38(5): 3387-3397. DOI URL
[17]	ZHAO Z B, LIU N, WANG L. Localization of multiple insulators by orientation angle detection and binary shape prior knowledge[J]. IEEE Transactions on Dielectrics and Electrical Insulation, 2015, 22(6): 3421-3428. DOI URL
[18]	ZHENG H B, SUN Y H, LIU X H, et al. Infrared image detection of substation insulators using an improved fusion single shot multibox detector[J]. IEEE Transactions on Power Delivery, 2021, 36(6): 3351-3359. DOI URL
[19]	ZHENG H B, LIU Y, SUN Y H, et al. Arbitrary-oriented detection of insulators in thermal imagery via rotation region network[J]. IEEE Transactions on Industrial Informatics, 2022, 18(8): 5242-5252. DOI URL
[20]	LIU X L, RAO Z Y, ZHANG Y X, et al. UAVs images based real-time insulator defect detection with transformer deep learning[C]// 2023 IEEE International Conference on Robotics and Biomimetics. New York: IEEE Press, 2023: 1-6.
[21]	GUO Z Y, ZHU T, ZHANG X Y, et al. Infrared insulator object detection algorithm based on improved deformable DETR[C]// The 10th International Forum on Electrical Engineering and Automation. New York: IEEE Press, 2023: 147-151.
[22]	ZHU X Z, SU W J, LU L W, et al. Deformable DETR: deformable transformers for end-to-end object detection[EB/OL]. (2021-05-18) [2024-11-03]. https://arxiv.org/abs/2010.04159.pdf.
[23]	LU W, CHEN S B, TANG J, et al. A robust feature downsampling module for remote-sensing visual tasks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 4404312.
[24]	HE K M, GKIOXARI G, DOLLÁR P, et al. Mask R-CNN[C]// 2017 IEEE International Conference on Computer Vision. New York: IEEE Press, 2017: 2980-2988.
[25]	李文璞, 毛颖科, 廖逍, 等. 基于旋转目标检测的变电设备红外图像电压致热型缺陷智能诊断方法[J]. 高电压技术, 2021, 47(9): 3246-3253.
	LI W P, MAO Y K, LIAO X, et al. Intelligent diagnosis method of infrared image for substation equipment voltage type thermal defects based on rotating target detection[J]. High Voltage Engineering, 2021, 47(9): 3246-3253 (in Chinese).
[26]	王有元, 李后英, 梁玄鸿, 等. 基于红外图像的变电设备热缺陷自调整残差网络诊断模型[J]. 高电压技术, 2020, 46(9): 3000-3007.
	WANG Y Y, LI H Y, LIANG X H, et al. Self-adjusting residual network diagnosis model for substation equipment thermal defects based on infrared image[J]. High Voltage Engineering, 2020, 46(9): 3000-3007 (in Chinese).
[27]	SELVARAJU R R, DAS A, VEDANTAM R, et al. Grad-CAM: why did you say that?[EB/OL]. [2024-08-20]. https://arxiv.org/abs/1611.07450.
[28]	王海群, 魏培旭, 解浩龙, 等. 基于改进YOLOv8的红外船舶检测[J]. 电光与控制, 2025, 32(1): 61-67.
	WANG H Q, WEI P X, XIE H L, et al. Infrared ship detection based on improved YOLOv8[J]. Electronics Optics & Control, 2025, 32(1): 61-67 (in Chinese).
[29]	崔克彬, 焦静颐. 基于MCB-FAH-YOLOv8的钢材表面缺陷检测算法[J]. 图学学报, 2024, 45(1): 112-125. DOI
	CUI K B, JIAO J Y. Steel surface defect detection algorithm based on MCB-FAH-YOLOv8[J]. Journal of Graphics, 2024, 45(1): 112-125 (in Chinese). DOI
[30]	魏陈浩, 杨睿, 刘振丙, 等. 具有双层路由注意力的YOLOv8道路场景目标检测方法[J]. 图学学报, 2023, 44(6): 1104-1111. DOI
	WEI C H, YANG R, LIU Z B, et al. YOLOv8 with bi-level routing attention for road scene object detection[J]. Journal of Graphics, 2023, 44(6): 1104-1111 (in Chinese).
[31]	易磊, 黄哲玮, 易雅雯. 改进YOLOv8的输电线路异物检测方法[J]. 电子测量技术, 2024, 47(15): 125-134.
	YI L, HUANG Z W, YI Y W. Improved YOLOv8 foreign object detection method for transmission lines[J]. Electronic Measurement Technology, 2024, 47(15): 125-134 (in Chinese).
[32]	TONG Z J, CHEN Y H, XU Z W, et al. Wise-IoU: bounding box regression loss with dynamic focusing mechanism[EB/OL]. (2023-08-08) [2024-11-03]. https://arxiv.org/abs/2301.10051.
[33]	MA S L, XU Y. MPDIoU: a loss for efficient and accurate bounding box regression[EB/OL]. (2024-08-20) [2024-11-03]. https://arxiv.org/abs/2307.07662.
[34]	HENDRYCKS D, GIMPEL K. Gaussian error linear units (GELUs)[EB/OL]. [2024-08-20]. https://arxiv.org/abs/1606.08415.
[35]	WANG A, CHEN H, LIU L H, et al. YOLOv10:real-time end-to-end object detection[EB/OL]. [2024-08-20]. https://arxiv.org/abs/2405.14458.
[36]	WANG C Y, BOCHKOVSKIY A, LIAO H Y M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2023: 7464-7475.
[37]	LI C, ZHOU A J, YAO A B. Omni-dimensional dynamic convolution[EB/OL]. [2024-08-20]. https://arxiv.org/abs/2209.07947.
[38]	XU G P, LIAO W T, ZHANG X, et al. Haar wavelet downsampling: a simple but effective downsampling module for semantic segmentation[J]. Pattern Recognition, 2023, 143: 109819. DOI URL
[39]	NING J L, SPRATLING M. The importance of anti-aliasing in tiny object detection[EB/OL]. [2024-08-20]. https://arxiv.org/abs/2310.14221.
[40]	ZHANG X, SONG Y Z, SONG T T, et al. LDConv: linear deformable convolution for improving convolutional neural networks[J]. Image and Vision Computing, 2024, 149: 105190. DOI URL
[41]	ZHANG H, ZHANG S J. Shape-IoU: more accurate metric considering bounding box shape and scale[EB/OL]. [2024-04-01]. https://arxiv.org/abs/2312.17663.
[42]	ZHANG H, ZHANG S J. Focaler-IoU: more focused intersection over union loss[EB/OL]. [2024-08-20]. https://arxiv.org/abs/2401.10525.
[43]	YU Z P, HUANG H B, CHEN W J, et al. YOLO-FaceV2: a scale and occlusion aware face detector[J]. Pattern Recognition, 2024, 155: 110714. DOI URL
[44]	WANG C Y, YEH I H, LIAO H Y M. YOLOv9:learning what you want to learn using programmable gradient information[EB/OL]. [2024-08-20]. https://doi.org/10.48550/arXiv.2402.13616.

模型	标注方法	P	R	AP50/%	AP50:95/%	FPS/(帧·s^-1)	GFLPOs	Parameters/M
	①	86.9	72.4	80.9	57.6	87.72
RT-DETRr18	②	88.9	77.6	85.5	59.9	75.19	56.9	19.9
	③	92.3	78.4	87.9	62.2	90.91
	①	82.6	70.5	82.1	57.6	161.29
YOLOv10n	②	80.0	75.7	82.2	56.4	123.46	8.2	2.7
	③	84.5	79.9	86.7	60.4	151.52
	①	91.1	75.2	82.9	49.2	588.24
YOLOv7t	②	87.9	78.5	85.1	54.0	555.56	13.0	6.0
	③	89.8	82.2	87.1	54.8	555.56
	①	94.8	81.7	89.1	61.7	123.46
YOLOV8n	②	87.6	80.6	88.0	61.4	151.52	8.1	3.0
	③	93.0	86.5	91.4	63.2	140.85

模型	标注方法	P	R	AP50/%	AP50:95/%	FPS/(帧·s^-1)	GFLPOs	Parameters/M
	①	86.9	72.4	80.9	57.6	87.72
RT-DETRr18	②	88.9	77.6	85.5	59.9	75.19	56.9	19.9
	③	92.3	78.4	87.9	62.2	90.91
	①	82.6	70.5	82.1	57.6	161.29
YOLOv10n	②	80.0	75.7	82.2	56.4	123.46	8.2	2.7
	③	84.5	79.9	86.7	60.4	151.52
	①	91.1	75.2	82.9	49.2	588.24
YOLOv7t	②	87.9	78.5	85.1	54.0	555.56	13.0	6.0
	③	89.8	82.2	87.1	54.8	555.56
	①	94.8	81.7	89.1	61.7	123.46
YOLOV8n	②	87.6	80.6	88.0	61.4	151.52	8.1	3.0
	③	93.0	86.5	91.4	63.2	140.85

YOLOv8n	类内稀疏先验	RFD	Wise-MPDIoU	P	R	AP50/%	AP50:95/%	FPS/(帧·s^-1)	GFLPOs	Parameters/M
√				94.8	81.7	89.1	61.7	123.46	8.1	3.01
√	√			93.0	86.5	91.4	63.2	140.85	8.1	3.01
√	√	√		93.5	86.2	92.1	63.9	121.95	8.1	3.02
√	√		√	95.6	85.8	92.4	64.1	144.93	9.6	3.01
√	√	√	√	93.6	87.6	92.4	65.2	120.48	9.6	3.02

YOLOv8n	类内稀疏先验	RFD	Wise-MPDIoU	P	R	AP50/%	AP50:95/%	FPS/(帧·s^-1)	GFLPOs	Parameters/M
√				94.8	81.7	89.1	61.7	123.46	8.1	3.01
√	√			93.0	86.5	91.4	63.2	140.85	8.1	3.01
√	√	√		93.5	86.2	92.1	63.9	121.95	8.1	3.02
√	√		√	95.6	85.8	92.4	64.1	144.93	9.6	3.01
√	√	√	√	93.6	87.6	92.4	65.2	120.48	9.6	3.02

方法	AP50/%	AP50:95/%	GFLPOs
ODConv	87.3	60.7	5.8
HWD	91.5	62.6	7.7
WaveletPool	91.1	63.4	7.4
LDConv	91.0	62.3	8.0
RFD	92.1	63.9	8.1