Review Reports

Round 1

Reviewer 1 Report

The authors have developed a promising architecture for the detection of small target in transmission lines. I have following suggestions to improve the paper comprehension:

1.      Figure comprehension is very difficult. For example, Figure 1, the blue circles are not visible. The figure shows the comparison between the existing methods and the paper presented. Are those images generated from existing methods and the new algorithm respectively? If images are presented from existing methods, what specific model is being used from the literature. Similarly, Figure 8 and 9, the changes are not evidently visible.

2.      The description for dataset is missing. Are the models of comparison trained and tested on the same data.

3.         Comparison with state of art models, would help better compare the model performance. You may consider models such as, Cao, S.; Deng, J.; Luo, J.; Li, Z.; Hu, J.; Peng, Z. Local Convergence Index-Based Infrared Small Target Detection against Complex Scenes. Remote Sens. 2023, 15, 1464. https://doi.org/10.3390/rs15051464, and YEAR=2023 URL=https://www.frontiersin.org/articles/10.3389/fnbot.2022.1074862    DOI=10.3389/fnbot.2022.1074862

4.      Consider having a GitHub to share models developed for use to the community.

   

Author Response

Thank you very much for your comments and suggestions, here are some of our responses:

Point 1: Graphics are very difficult to understand. For example, in Figure 1, the blue circle is not visible. The figure shows a comparison between the existing methods and the papers presented. Are these images generated by existing methods and new algorithms? If the image is presented from an existing method, what is the specific model used in the literature? Again, in Figures 8 and 9, the change is not obvious.

Response 1: Modifications have been made to the image not being clear, and the network architecture used by both methods has been indicated.

Point 2: Missing data set description. Compare whether models are trained and tested on the same data.

Response 2: A detailed description of the data set has been added, and the comparison model is trained and tested on the same data set.

Point 3: Comparison with the most advanced models will help to better compare model performance. You can consider examples such as Cao, S. Deng Jun; Roger; Li Zhiqiang; Hu Jun; Infrared small target detection in complex scene based on local convergence index. Rs. 2023, 15, https://doi.org/10.3390/rs15051464 and 1464. YEAR = 2023 = HTTPS URL: / / www.frontiersin.org/articles/10.3389/fnbot.2022.1074862 DOI = 10.3389 / fnbot. 2022.1074862

Response 3: We have done a comparative experiment, which is indicated in the paper. Since the author who proposed the infrared small target detection method only gave the processing time of each image in the NUAA-SIRST dataset, but not the processing time of each image in the NUDT-SIRST dataset, this paper only compared the  frames per second (FPS) of each image in the NUAA-SIRST dataset.

Point 4: Consider having a GitHub to share models developed for community use.

Response 4: A github repository has been created. The repository address is：https://github.com/cheng9809/yolov5.git

Reviewer 2 Report

The authors needs to implement the comments which is attached as pdf

Comments for author File: Comments.pdf

Extensive english corrections required

Author Response

Thank you very much for your comments and suggestions, here is our response to them:

Point 1: In abstract line 11 numerous is bold unnecessarily. Is there any

specific reason for making it bold.

Response 1: Have been modified.

Point 2: what is accuracy P in abstract line 24.

Response 2: The accuracy P is Precision, which is one of the evaluation indicators.

Point 3: Typo error “Ito” line 94.

Response 3: Have been modified.

Point 4: Line 95 “wen enhance the network’s capability to focus” check the

Grammar

Response 4: Have been revised.

Point 5: Figure 1 has to be improved. Instead of giving figures, the proposed

methodology can be explained as a block diagram for better clarity

and understanding.

Response 5: Figure 1 has been modified.

Point 6: In-text citations should be based on mdpi format. The citations are

not visible and they should not be in superscript.

Response 6: Have been revised

Point 7: Line 153-155 “In summary, it can be found that the above research

can’t simultaneously take into account the detection accuracy and

detection speed when carrying out small target detection of

transmission lines, which is needed to achieve small target detection

of transmission lines.” Rewrite the sentence as it is very long.

Response 7: Have been revised.

Point 8: Overall English writing should be improved. It is very poor.

Response 8: The whole paper has been adjusted.

Point 9: All the figures need to be improved as it is not satisfactory.

Response 9: Have been modified.

Point 10: Performance metrics can be given in tabular form

Response 10: Have been modified.

Point 11: The graphs used does not have the x and y axis.

Response 11: It has been indicated in the paper.

Point 12: All the figures should have the same size some are zoomed in or

zoomed out.

Response 12: The size of the graph has been unified.

Point 13: The discussion needs to be improved.

Response 13: The discussion has been improved.

Point 14: Give a tabular form comparing your work with previous works for

better readability.

Response 14: The algorithm in this paper has been compared with previous algorithms in tabular form.

Point 15: References are not in mdpi format please check.

Response 15: Have been modified.

Author Response File: Author Response.docx

Reviewer 3 Report

Transmission Lines Small Target Detection Algorithm Research Based on YOLOv5

“In recent years, UAV technology” line 1, define UAV as its first instance.

I find related work is very limited, can be improved to offer a substantial study to the readers of the field.

“ECA attention mechanism to solve this problem”, is it proposed by authors? As no citation support.

Some of the augmented images can be shown here to show the new dataset images.

Work is good, as improved results are significant from yolo and others.

Need to explain results section with more clarity, Fugure 14 and why such changes?

Can be explained

 

Training loss comparison can be accompanied by exhaustive explanation.

Minor edits

Author Response

Thank you very much for your comments and suggestions, here is our response：

Point 1: “In recent years, UAV technology” line 1, define UAV as its first instance.

Response 1: Have been modified.

Point 2: I find related work is very limited, can be improved to offer a substantial study to the readers of the field.

Response 2: This section has been added in related work.

Point 3: “ECA attention mechanism to solve this problem”, is it proposed by authors? As no citation support.

Response 3: This module was not proposed by us, and its source is indicated in the paper.

Point 4: Some of the augmented images can be shown here to show the new dataset images.

Response 4: It is not clear what picture needs to be shown here, we guess is it the comparison effect picture before and after adding ECA module?

Point 5: Need to explain results section with more clarity, Fugure 14 and why such changes?

Response 5: The comparison results in Figure 14 and the reasons for such changes have been explained, which have been indicated in the paper.

Point 6: Training loss comparison can be accompanied by exhaustive explanation.

Response 6: A detailed description has been added for this.

Reviewer 4 Report

Authors presented a new approach “Transmission Lines Small Target Detection Algorithm search Based on YOLOv5”. But paper needs some minor modifications.

 

1.      Include latest references from 2023 focusing on Small Detection YOLOv5.

2.      Contribution is not properly highlighted.

3.      There is no mention of limitations in the study.

4.      Please correct typo mistakes

5.      Please cross check the scale of all figures in the manuscript to make them clear  and more presentable.

 Please correct typo mistakes

Author Response

Thank you very much for your comments and suggestions, here is our response to them:

Point 1: Include latest references from 2023 focusing on Small Detection YOLOv5.

Response 1: This paper has added new references from 2023, which focus on the research of small target detection of transmission lines based on YOLOv5.

Point 2: Contribution is not properly highlighted.

Response 2: The author contribution has been re-adjusted.

Point 3: here is no mention of limitations in the study.

Response 3: The detection method proposed in this study is specifically aimed at the detection difficulties of small targets (including bolts, insulators, fixtures, etc.) of transmission lines, and has certain pertinence.

Point 4: Please correct typo mistakes

Response 4: Modifications have been made.

Point 5: Please cross check the scale of all figures in the manuscript to make them clear  and more presentable.

Response 5: Modifications have been made.