Correction: Siddiqui, Z.A.; Park, U. A Drone Based Transmission Line Components Inspection System with Deep Learning Technique. Energies 2020, 13, 3348

The authors wish to make the following corrections to their paper [1]:

Changes in Figures

On page 14, Figure 8 should be changed from:

to the following correct version:

On page 15, Figure 9 should be changed from:

to the following correct version:

Change in Main Body Paragraphs

We found 58 inadvertent typos in our paper published in paper [1], as follows:

There are a few changes in the Abstract of the paper:

The sentence in the Abstract should be changed from:

damper-weights, power lines, and then analyze these transmission line components for

to the following correct version:

damper-weights, and power lines, and then analyze these transmission line components for

The sentence in the Abstract should be changed from:

by remote-controlled drone. The detected components are then analyzed using novel defect

to the following correct version:

by remotely controlled drone. The detected components are then analyzed using novel defect

The sentence in the Abstract should be changed from:

help of the proposed system automatic defect analyzing system, manual inspection time can be

to the following correct version:

help of the proposed automatic defect analyzing system, manual inspection time can be

Changes in the Introduction section (Section 1):

The sentence should be changed from:

such as power lines [6–9], electric poles [10–12], and insulators [13–16], and analyze their defects [1,17]

to the following correct version:

such as power lines [6–9], electric poles [10–12], and insulators [13–16], and analyzing their defects [1,17]

The sentence should be changed from:

The related work on electrical components detection can be classified into three categories,

to the following correct version:

The related studies on electrical component detection can be classified into three categories,

The sentence should be changed from:

variations in illumination and view-angle, which is inevitable in outdoor environments.

to the following correct version:

variations in illumination and view point, which are inevitable in outdoor environments.

The sentence should be changed from:

images and further segmentation of insulator using color features. The basic problem with using Haar features in insulator detection is that Haar features cannot handle large rotation and view-point angle variations.

to the following correct version:

images and then segment the insulator using color features. The basic problem with using Haar features in insulator detection is that Haar features cannot handle large rotation and view-point changes.

The sentence should be changed from:

to train binary SVM classifiers to detect insulator presence or absence. To avoid the sliding window-based approach, [26] proposed to binarize the input image and apply the SVM classifier to the areas

to the following correct version:

to train binary SVM classifiers to detect insulators. To avoid the sliding-window-based approach, [26] proposed to binarize the input image and apply SVM classifier to the areas

The sentence should be changed from:

These sliding window-based components detection methods lack generalizability and suffer from low detection speeds, because, in order to address the size and rotation variations, these method scales and rotates the input image to multiple sizes and multiple orientations, respectively.

to the following correct version:

These sliding window-based component detection methods lack generalizability and suffer from low detection speeds, because these methods scale and rotate the input image to multiple sizes and multiple orientations in order to address the size and rotation variations.

The sentence should be changed from:

(Single Shot Detector) [31] have not only shown staggering detection accuracies on unconstrained

to the following correct version:

(Single Shot Detector) [31] have not only shown high detection accuracies on unconstrained

The sentence should be changed from:

1.3.1. What are Convolutional Neural Networks (CNN)?

to the following correct version:

1.3.1. Convolutional Neural Network (CNN)

The sentence should be changed from:

function becomes very small or saturates.

to the following correct version:

function becomes very small or saturated.

The sentence should be changed from:

vision algorithms make them less viable.

to the following correct version:

algorithms make them less viable.

The sentence should be changed from:

equipped embedded platform to detect and classify different types of power line components form a

to the following correct version:

equipped embedded platform to detect and classify different types of power line components from a

The sentence should be changed from:

components defects analyzing methods were never covered in previous research works

to the following correct version:

component defect analyzing methods have never been covered in previous studies.

The sentence should be changed from:

5. A complete transmission line components inspection system is presented, whose robustness and real-time performance is evaluated and validated.

to the following correct version:

5. A complete transmission line component inspection system is presented and its robustness and real-time performance are evaluated.

The sentence should be changed from:

experimental setups used to train those detectors is discussed. The proposed power line detection

to the following correct version:

experimental setups used to train those detectors are discussed. The proposed power line detection

Changes in the Proposed System (Section 2):

The sentence should be changed from:

to acquaint the reader with the visual properties (i.e., shape, color, etc.) of the transmission line

to the following correct version:

to familiarize the reader with the visual properties (i.e., shape, color, etc.) of the transmission line

The sentence should be changed from:

our proposed system. (a) Transmission tower; (b) Lightning-Arrester (LA); (c) suspension type white

to the following correct version:

the proposed system. (a) Transmission tower; (b) lightning arrester (LA); (c) suspension-type—white

The sentence should be changed from:

area as efficiently. A drone has the benefit of flying up to the top of transmission towers, which are

to the following correct version:

area. A drone has the benefit of flying to the top of transmission towers, which are

The sentence should be changed from:

are passed to the defect analyzer that analyzes each type of defect independently. Once a particular defect is detected at the defect analysis stage, the defect analyzer returns the percentage of that defect

to the following correct version:

are passed to the defect analyzer that inspects each type of defect independently. Once a particular defect is detected at the defect analysis stage, the defect analyzer returns the probability of having that defect

The sentence should be changed from:

framework, the electrical power line are detected using a novel algorithm, which is presented in Section 2.2. In contrast to [8], here, the drone is flown 10~20 m away from the power line, which also prevents any possible damage to power line in case of sudden drone failure. Hence in total, the proposed system detects ten transmission line components and also analyzes for the corresponding defects. In the later sections, a detailed overview of the various steps shown in the system diagram of Figure 2 is given.

to the following correct version:

framework, the electrical power line is detected using a novel algorithm, which is presented in Section 2.2. In contrast to [8], the drone is flown 10~20 m away from the power line, which also prevents any possible damage to power line in case of sudden failure. In summary, the proposed system detects ten transmission line components and analyzes their defects. In the later sections, detailed explanations of those steps shown in Figure 2 are given.

Changes in Section 2.1:

The sentence should be changed from:

Transform (SIFT) descriptors produced big performance gains a decade ago, but due to diversity of appearance, illumination conditions, and background, it is difficult to manually design a robust feature descriptor to perfectly describe all kinds of object. Recently, deep convolutional features

to the following correct version:

Transform (SIFT) descriptors produced large performance gains a decade ago. However, due to diversity of appearance, illumination conditions, and background, it is difficult to manually design a robust feature descriptor to perfectly describe all kinds of objects. Currently, deep convolutional features

The sentence should be changed from:

detect nine different types of transmission line components, transmission towers, LA, porcelain

to the following correct version:

detect nine different types of transmission line components, i.e., transmission towers, LA, porcelain

The sentence should be changed from:

(PorSTI-R, PorSTI-W, and PolSTI are not standard abbreviations). The original Darknet-53 network

to the following correct version:

The original Darknet-53 network of YOLO V3 scales the object to three different levels during the training and testing time, and the

The sentence should be changed from:

Every components appear with a random orientation, view-point, lighting conditions, shapes, colors and scales, yet the CNN-based detector is able to detect them robustly.

to the following correct version:

Each component appears with a random orientation, view point, lighting conditions, shapes, colors, and scales, yet it is detected by the CNN-based detector robustly.

The sentence should be changed from:

The proposed CNN-based framework divides the input video frame into fixed-sized S × S regions, and from each region, it predicts 11 bounding box predictions using 11 anchor boxes. In order to generate good anchor box priors, the ground truth bounding box sizes are clustered into 11 groups using K-mean clustering [42]. The decision to clustered the size priors into 11 groups is based upon the aspect ratios in which these transmission line components can appear (Figure 1b–e) appears in similar aspect ratios while Figure 1a appears in totally different aspect ratio).

As both the PorSTI-R and PorSTI-W suffer from the same defects, they need not be differentiated at the detection step. Hence, the CNN-based transmission line components detector is trained with eight classes (combined the two classes PorSTI-R and PorSTI-W (Figure 1c,d) into one class). The detector is learned with a faster learning rate for the first half of the network training process so that the network can quickly learn to distinguish between different types of components. In the next stage, the learning rate is slowed down by a factor of 10 every 1000 iterations for the rest of the training process in order for the network to slowly learn the details of the shape, color, and context of the different types of components. Figure 3 shows some of the example detection results of the trained network.

to the following correct version:

The proposed CNN-based framework divides the input video frame into fixed-sized S×S regions, and from each region, it predicts 11 bounding boxes using 11 anchor boxes. To generate good anchor box priors, the ground-truth bounding box sizes are clustered into 11 groups using K-means clustering [42]. The decision to cluster the size priors into 11 groups is based upon the aspect ratios in which these transmission line components can appear. Figure 1b–e shows similar aspect ratios while Figure 1a shows an entirely different aspect ratio.

As both the PorSTI-R and PorSTI-W suffer from the same defects, they need not be differentiated at the detection step. Hence, the CNN-based transmission line component detector is trained with eight classes combining PorSTI-R and PorSTI-W into one class. The detector is taught with a faster learning rate for the first half of the network training process so that the network can quickly learn to distinguish between different types of components. In the next stage, the learning rate is slowed by a factor of 10 after every 1000 iterations for the remainder of the training process, which allows the network to slowly learn the details of the shape, color, and context of each component. Figure 3 shows example detection results of the trained network.

Changes in the Section 2.2:

The sentence should be changed from:

Detection of power line (rope wire) from aerial images is also a challenging task because of the cluttered background. The presence of fields, trees, bushes, and mountains causes enormous noise in the equivalent edge map (shown in Figure 3 and Figure 4a). Moreover, due to the natural sag in the power line, the edge equivalent of the power line is not a straight line (Figure 4b). Consequently, the power line detection methods that use Hough lines [34] end up with large false positive detections. However, it can be observed from the inspection videos that these power line appear as the connected component that appear horizontal to the camera (due to drone’s flying position). Hence a custom shaped kernel to compute the edge map (Figure 4c) is proposed in this paper. The algorithm to detect power line is given as Algorithm 1.

to the following correct version:

Detection of a power line (rope wire) from aerial images is also a challenging task because of the cluttered background. The fields, trees, bushes, and mountains causes enormous noise in the edge maps (shown in Figures 3 and 4a). Moreover, because of the natural sag in the power line, the edge equivalent of the power line is not a straight line (Figure 4b). Consequently, the power line detection methods that use Hough lines [34] end with large false positive detections. However, it can be observed from the inspection videos that these power lines appear as the connected components that are horizontal to the camera (due to drone’s flying position). Hence, a custom-shaped kernel to compute the edge map (Figure 4c) is proposed in this paper. The algorithm to detect power line is given in Algorithm 1.

The sentence should be changed from:

(here detected power lines are marked with red color, and labelled with an ID number).

to the following correct version:

Detected power lines are marked in red and labelled with ID numbers.

Changes in the Section 2.3:

The sentence should be changed from:

2.3. Defect Detection

The Previous sections presented the methods to detect transmission line components from the

to the following correct version:

2.3. Defect Analyzer

Previous sections presented the methods to detect transmission line components from the

The sentence should be changed from:

In every type of insulator (Figure 1b–d), the repeating circular-shaped part is called the “shed” or

to the following correct version:

In every type of insulator shown in Figure 1b–d, the repeating circular-shaped part is called the “shed” or

The sentence should be changed from:

to design a robust algorithm. As the sheds of the insulator appear as an equally spaced repeating pattern, hence the proposed algorithm iteratively searches for the missing shed using sample templates of the ceramic shed, drawn from the input insulator image. The missing shed/disk detection algorithm is elaborated using Figure 6.

to the following correct version:

As the sheds of the insulator appear as equally spaced repeating patterns, the proposed algorithm iteratively searches for the missing shed using sample templates of the ceramic shed, obtained from the input insulator image. The missing shed/disk detection algorithm is elaborated in Figure 6.

The sentence should be changed from:

then a LoG (Laplacian of Gaussian) filter is applied in order to detect the areas of rapid intensity

to the following correct version:

then a LoG (Laplacian of Gaussian) filter is applied to detect the areas of rapid intensity

The sentence should be changed from:

Figure 7. Steps to detect defect in balisor. (a) Input image after detection using CNN, (b) after eroding, (c) color clustering, (d) foreground mask after clusters selection and region filling, (e) masked foreground, (f) balisor image in R- and S-color space, (g) after filtering (f), (h) after removing noise from (g), (i) combination of R- and S-space outputs in (h), (j) mask for boundary regions, (k) removing boundary regions, and (l) marking the defected regions.

to the following correct version:

Figure 7. Steps to detect defects in balisor. (a) Input image after detection using CNN, (b) after eroding, (c) color clustering, (d) foreground mask after cluster selection and region filling, (e) masked foreground, (f) balisor image in red and saturation color space, (g) after filtering (f), (h) after removing noise from (g), (i) combination of red and saturation space outputs in (h), (j) mask for boundary regions, (k) removing boundary regions, (l) marking the defect regions.

The sentence should be changed from:

As their name suggests, sag adjusters are used to adjust the slackness in the transmission power

to the following correct version:

As their name suggests, sag adjusters are used to adjust the sagginess in the transmission power

The sentence should be changed from:

clusters the RGB image into two clusters, where the bigger cluster constitutes the background. A Gaussian mixture models (GMM) is learned using the means and covariance from the RGB and HSV space of the corroded regions of train images (random rusty images downloaded from the internet). The algorithm learns two GMM models using each color space and during test time it fuses the regions with a high probability of corrosion using a weighted sum approach. Figure 8 illustrates the different steps of the corrosion detection algorithm.

to the following correct version:

clusters the RGB image into two clusters, where the larger cluster constitutes the background. The Gaussian Mixture Models (GMM) are trained using the means and covariance from the RGB and HSV space of the corroded regions of train images (random rusty images downloaded from the internet). The algorithm trains two GMM models using each color space and it fuses the regions with a high probability of corrosion using a weighted sum approach. Figure 8 illustrates the corrosion detection algorithm.

Changes in the Experimental results and discussion section (Section 3):

The sentence should be changed from:

with those of state-of-the-art methods. Among the published studies, only [27,28,33,47–50]

to the following correct version:

with those of the state-of-the-art methods. Among the published studies, only [27,28,33,47–50]

The sentence should be changed from:

used in this paper, i.e., Pascal score.

The well-known Pascal score [51] is used to evaluate the performance of the proposed transmission line components detector. The Pascal score is calculated by taking the Intersection-over-

to the following correct version:

used in this paper, i.e., Pascal score [51].

The well-known Pascal score [51] is used to evaluate the performance of the proposed transmission line components detector. The Pascal score is calculated by taking the Intersection-over-

The sentence should be changed from:

and orientations, but the CNN-based detector shows reasonable detection performance for the

to the following correct version:

and orientations. However, the CNN-based detector shows reasonable detection performance for the

The sentence should be changed from:

is different in terms of number of test samples, resolution of images, evaluation metric, etc. and hence

to the following correct version:

is different in terms of number of test samples, resolution of images, evaluation metric, and hence

The sentence should be changed from:

the detection time and accuracy on the acquired dataset. The implementation of LSD at [53] and EDLines implementation at [54] is used. For evaluation purpose, three videos of the power line inspection

to the following correct version:

the detection time and accuracy on the acquired dataset. The implementation of LSD in [53] and EDLines implementation in [54] is used for the comparison. For the purpose of evaluation, three videos of the power line inspection

The sentence should be changed from:

that the proposed power line detector outperforms state-of-the-art, in terms of accuracy and detection

to the following correct version:

that the proposed power line detector outperforms the state-of-the-art methods, in terms of accuracy and detection

The sentence should be changed from:

Figure 14. First row: Line Segment Detector (LSD), second row: edge drawing algorithm (EDLines), and third row: proposed scheme.

to the following correct version:

Figure 14. Example power line detection results. First row: LSD, second row: EDLines, and third row: proposed scheme.

The sentence should be changed from:

One of the reasons why EDLines and LSD show poor power line detection result on this dataset

to the following correct version:

One of the reasons why EDLines and LSD show poor power line detection results on this dataset

The sentence should be changed from:

LSD, and the proposed method is ran on an image size of 1920 × 1080. The proposed algorithm is also

to the following correct version:

LSD, and the proposed method is performed on image with a size of 1920 × 1080. The proposed algorithm is also

The sentence should be changed from:

the slowest of all. EDLines (second row: Figure 14) performs better than LSD, due to better noisy edge removal capability than LSD. However, EDLines also yield many false positives on transmission

to the following correct version:

the slowest of all. EDLines (second row: Figure 14) performs better than LSD, owing to better noisy-edge removal capability than that of LSD. However, EDLines also yields many false positives on transmission

The sentence should be changed from:

Figure 13. PR-curve for power line detection shows dominant performance of proposed power line detection scheme.

to the following correct version:

Figure 13. PR-curve for power line detection shows superior performance in the proposed power line detection scheme.

The sentence should be changed from:

3.4. Defect Detection

to the following correct version:

3.4. Defect Analysis

The sentence should be changed from:

components is not feasible is the difficulty to collect a large number of faulty component images.

to the following correct version:

components is not feasible is the difficulty in collecting a large number of faulty component images.

The sentence should be changed from:

polymer insulators, and broken shed defect. The results of the proposed broken shed detection

to the following correct version:

polymer insulators, and broken shed defects. The results of the proposed broken shed detection

The sentence should be changed from:

As given in Table 4, the proposed method outperforms all the previous methods [49,50,55–57] in

to the following correct version:

As shown in Table 4, the proposed method outperforms all the previous methods [49,50,55–57] in

The sentence should be changed from:

approaches [50,56], and very near close precision and recall with CNN-based methods [49,50]. As

to the following correct version:

approaches [50,56], and comparable precision and recall with CNN-based methods [49,50]. As

The sentence should be changed from:

are designed to fly in extreme weather conditions (heavy rain, snow, winds, etc.) and can work properly, the authors do not recommend the inspection videos are taken in such weather to avoid any accidents.

to the following correct version:

are designed to fly in extreme weather conditions (e.g., heavy rain, snow, winds), the authors do not recommend the inspection videos are taken in such weather due to the chance for accidents.

Changes in the Conclusions and future directions section (Section 4):

The sentence should be changed from:

In this paper, an automatic transmission line components detection and defect detection system

to the following correct version:

In this paper, an automatic transmission line component detection and defect analysis system

The sentence should be changed from:

detection algorithm. The proposed power line algorithm shows dominant performance on the given

to the following correct version:

detection algorithm. The proposed power line detection algorithm shows superior performance on the given

The sentence should be changed from:

the power line components are detected, a defect detection system checks for potential defects. The performance of the proposed defect detectors suggests that handcrafted approaches can be used to detect some of the types of defects in situations where the availability of a large number of defected samples is not viable.

to the following correct version:

the power line components are detected, a defect analysis system checks for potential defects. The performance of the proposed defect analyzers suggests that handcrafted, feature-based approaches can be used to detect some of the types of defects in situations where the availability of a large number of defect samples is not viable.

The sentence should be changed from:

electric workers (physically climb transmission towers on regular basis) or use of expensive patrolling helicopters to inspect the conditions of transmission line components. The proposed defect analyzer

to the following correct version:

electric workers physically climbing transmission towers on regular basis or use of expensive patrolling helicopters to inspect the conditions of transmission line components. The proposed defect analysis

The sentence should be changed from:

deeper levels and formalizing for the detection of additional types of component defects. The future research will also consider combining the detection module and the defect analysis module under a

to the following correct version:

deeper levels and formalizing for the detection of additional types of defects. Future research will consider combining the detection module and the defect analysis module under a

We have found many inadvertent errors in our paper published in Journal [1].

These changes have no material impact on the conclusions of our paper. We apologize to our readers.

The funding remains same, but there were typos which we want to update from.

The Funding part should be changed from:

Funding: This research was supported by a grant (19PQWO-B153369-01) from Smart road lighting platform development and empirical study on test-bed Program funded by Ministry of the Interior and Safety of Korean government.

to the following correct version:

Funding: This research was supported by a grant (20PQWO-B153358-02) from the Smart Road Lighting Platform Development and Empirical Study on Test-bed Program, funded by Korean Ministry of the Interior and Safety.