Insulator Infrared Image Segmentation Algorithm Based on Dynamic Mask and Box Annotation

Li, Tong; Zhou, Jinhui; Guoliang, Song; Wen, Yonghua; Ye, Yuanhao; Chen, Sijun

doi:10.1109/icpes53652.2021.9683904

Cited by 5 publications

(3 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Yan et al [2] studied a substation equipment instance segmentation model based on Mask R-CNN, which has a better segmentation effect on the substation equipment image dataset. Li et al [17] proposed an IR insulator image segmentation algorithm based on dynamic mask and box annotation to achieve the overall segmentation of insulator strings by marking insulators with rectangular boxes in the infrared image. Zhou et al [3] established the intrinsic connection between the parameters and the statistics of the image by optimizing the pulse-coupled neural network model, combining the information of the gradient of the image and the principle of great likelihood estimation, thus improving the precision of the fault segmentation of electrical equipment.…”

Section: Literature Reviewmentioning

confidence: 99%

An Improved Solov2 Based on Attention Mechanism and Weighted Loss Function for Electrical Equipment Instance Segmentation

Wu,

Liu,

Jiang

et al. 2024

CMC

View full text Add to dashboard Cite

The current existing problem of deep learning framework for the detection and segmentation of electrical equipment is dominantly related to low precision. Because of the reliable, safe and easy-to-operate technology provided by deep learning-based video surveillance for unmanned inspection of electrical equipment, this paper uses the bottleneck attention module (BAM) attention mechanism to improve the Solov2 model and proposes a new electrical equipment segmentation mode. Firstly, the BAM attention mechanism is integrated into the feature extraction network to adaptively learn the correlation between feature channels, thereby improving the expression ability of the feature map; secondly, the weighted sum of CrossEntropy Loss and Dice loss is designed as the mask loss to improve the segmentation accuracy and robustness of the model; finally, the non-maximal suppression (NMS) algorithm to better handle the overlap problem in instance segmentation. Experimental results show that the proposed method achieves an average segmentation accuracy of mAP of 80.4% on three types of electrical equipment datasets, including transformers, insulators and voltage transformers, which improve the detection accuracy by more than 5.7% compared with the original Solov2 model. The segmentation model proposed can provide a focusing technical means for the intelligent management of power systems.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

An Improved Solov2 Based on Attention Mechanism and Weighted Loss Function for Electrical Equipment Instance Segmentation

Wu,

Liu,

Jiang

et al. 2024

CMC

View full text Add to dashboard Cite

show abstract

“…In this context, Light Detection and Ranging (LiDAR) airborne scanning technology is the ideal alternative for creating accurate 3D models of the power lines and is preferred to Mobile-and Terrestrial-Laser Scanning methods, which cannot be applied in dense forest and steep mountainous landscapes [2], [5]. Additionally, during the last few years, Unmanned Aerial Vehicles (UAVs) have been made widely available, proposing a low cost and flexible solution for power infrastructure inspection applications covering the entire spectrum of power components [6], [7], power line [8], [9], utility poles detection [10], [11], as well as the development of autonomous systems to rapidly acquire high-quality data in a repeatable manner using multiple sensors, e.g. RGB, thermal, multispectral [12], [13].…”

Section: Introductionmentioning

confidence: 99%

A LiDAR-Based Method to Identify Vegetation Encroachment in Power Networks with UAVs

Savva

Papageorgiou

Kyrkou

et al. 2023

2023 International Conference on Unmanned Aircraft Systems (ICUAS)

View full text Add to dashboard Cite

Vegetation encroachment in power transmission and distribution networks constitutes a major hazard for the environment and the networks' integrity, but also for the society at large, with multifaceted consequences. On many occasions the vegetation near the power lines, in conjunction with the aged infrastructure, caused and spread fires leading to large-scale disasters. To this end, 3D representations are proactively created using LiDAR sensors to identify locations of vegetation encroachment. Of particular interest is the use of UAVs, which propose a cost-effective alternative to employing airplanes. In this study, UAVs were employed to acquire LiDAR data from the power distribution network and a subtractive data-driven methodology is proposed, whereby irrelevant points are discarded, aiming to identify power lines without employing 3D modelling methods. In this context, geometric features are calculated and a rigorous analysis is conducted over the feature set, different classifiers and parameters to investigate the robustness of the proposed approach. Extensive evaluation suggests that the Random Forest classifier is able to identify power lines with high performance (F1-Score=97.74% and Accuracy=99.09%), using both geometric and color-based features, being also robust to the presence of moderate noise and down-sampling levels.

show abstract

“…Li vd. [11] izolatör disklerini bölütlemek için doğru olmayan konumlandırma, düşük tanıma verimliliği ve karmaşık arka planda yalıtkan görüntünün zor bölütlenmesi problemlerini çözmeyi amaçlayan dinamik maske kullanan bir kızılötesi izolatör görüntü bölütleme algoritması önermişlerdir. Önerdikleri algoritmada kızılötesi görüntüdeki izolatörleri dikdörtgen bir çerçeve ile işaretleyerek izolatör dizisinin genel bölümlenmesini gerçekleştirmişlerdir.…”

Section: Introductionunclassified

Detection of Insulator Faults in Catenary Systems with Deep Learning

Karaduman

Akın

BİNAY

et al. 2022

Demiryolu Mühendisliği

View full text Add to dashboard Cite

Insulators are the most important components of catenary systems in electrified railway lines. Fractures or burns in insulators cause interruptions in transportation. These interruptions also prevent safe operation, especially on high-speed rail lines. Detecting faults in insulators at an early stage will enable to intervene in catenary systems at the most appropriate time and prevent insulator-related accidents. In this article, a deep learning-based method is proposed to classify insulators in catenary systems as faulty or intact. A data set containing 1100 isolator images was used in the study. The images in this dataset are trained and tested with the ResNet34 deep learning architecture. With the proposed architecture, faults in isolators are classified with 95,7% accuracy, 99% precision and 96,6% recall values. These values show that the performed study is a reliable method for fault detection in isolators in catenary systems.

show abstract

Insulator Infrared Image Segmentation Algorithm Based on Dynamic Mask and Box Annotation

Cited by 5 publications

References 14 publications

An Improved Solov2 Based on Attention Mechanism and Weighted Loss Function for Electrical Equipment Instance Segmentation

An Improved Solov2 Based on Attention Mechanism and Weighted Loss Function for Electrical Equipment Instance Segmentation

A LiDAR-Based Method to Identify Vegetation Encroachment in Power Networks with UAVs

Detection of Insulator Faults in Catenary Systems with Deep Learning

Contact Info

Product

Resources

About