Assessment of Trees’ Structural Defects via Hybrid Deep Learning Methods Used in Unmanned Aerial Vehicle (UAV) Observations

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

I enjoyed reading your paper.  Both the introduction to the topic and the description of what you did were comprehensive and your methods and results clear.  I think for this reason your paper is worthy of processsing.  My only complaint is that the introduction could be stated in much fewer words.  I made an attempt to demonstrate what I mean by editing the first paragraph.  I believe that substantial improvements could be made in all of the remaining paragraphs in the introduction in particular. 

Comments for author File: Comments.pdf

Comments on the Quality of English Language

Please see comments and suggestions for Authors.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

In this study, the deep learning algorithms of the lightweight You Only Look Once (YOLO) and encoder-decoder network DeepLabv3+ are combined in the unmanned aerial vehicle (UAV) observations to evaluate the trees’ structural defects.

The introduction correctly presents new and relevant bibliographic references for specialized literature. The sample test chosen for the case study is representative of formulating conclusions.

 

However, even if these two chapters are correctly prepared, what the work lacks, in my opinion, is the final part. Specifically, the discussion and conclusion chapters must have a different structure. The discussion chapter must focus on the presentation of the results, with the critical highlighting of the obvious limitations of this study, and the conclusion must highlight the original elements of this paper, in relation to what has been previously published on this topic.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Dear Authors,

After a careful review of the manuscript titled “Assessment of tree’s structural defects via hybrid deep learning methods used in unmanned aerial vehicle (UAV) observations”, I have made the following suggestions and comments for improving the manuscript.

Strengths:

1. The manuscript provides a strong explanation in the introduction part about the applications of deep learning methods in the detection of tree defects.
2. The limitations of various deep learning methods are clearly explained.
3. Ablation analysis statements are well articulated and clearly explained.

Comments:

1. The literature on UAV applications in tree defect identification is not sufficient in the manuscript.
2. The authors did not mention the UAV parameters (i.e., flying altitude, resolution, focal length, camera specifications, etc.) used for conducting this study. Including these details is essential for reproducibility and understanding the study's context.
3. If the UAV was flown at a lower altitude, the specific field circumstances under which this study was conducted need to be detailed. This information is crucial for evaluating the practicality and adaptability of the methodology.
4. The manuscript includes images of trees captured from different angles, providing a 360° view of each tree. However, the authors did not mention the flight planning of UAVs. Details about flight planning are necessary to understand the consistency and thoroughness of the data collection process. (Not in terms of distance from UAV to object)
5. The overall study does not convincingly highlight the application of UAVs. It appears that the same observations could be made using normal mobile videos. The authors need to clarify how they used UAVs in this study compared to other publications that utilized conventional video recording methods.
6. The novelty of the study is lacking in this manuscript. Several studies have already been conducted on tree defect identification using images or videos and deep learning methods. The authors should emphasize what new insights or methodologies their study contributes to the field.
7. The manuscript should include a discussion on the future applications of this work. Highlighting potential advancements and applications will demonstrate the broader impact and relevance of the study.
8. The limitations of this work, especially in terms of field conditions, need to be explicitly stated.

Suggestions:

1. The following statements need to be properly cited to provide the necessary academic support:

Line No.: 210 – 212

Line No.: 214 – 215

Line No.: 536 – 538

Line No.: 576 – 577

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript has been sufficiently improved to warrant publication in Forests.

Author Response

Comment: The manuscript has been sufficiently improved to warrant publication in Forests.

Response:  Thank you for your comment on the revisions.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

After reviewing the comments addressed by the authors of the manuscript titled “Assessment of tree’s structural defects via hybrid deep learning methods used in unmanned aerial vehicle (UAV) observations”, I have the following suggestions:

1.      The authors mentioned that they flew the UAV at different altitudes to capture the tree defects, but they did not provide information about the field conditions, such as plant density, which is crucial for UAV operations (the UAV might hit the tree if not stabilized properly).

2.      Please include a table detailing the type of drone used and its specifications for better clarity. It is important to note that medium and large UAVs might not be suitable for areas with high plant density.

3. The authors did not clarify the identification of tree defects that are not covered by UAV capturing by the complete rotation of 360⁰ angles.

 

4.      When capturing images of different trees at multiple altitudes and from various distances (3-5m), conventional mobile devices (e.g., cell phones, cameras mounted on a mobile car) could be more economical and accessible compared to UAVs.

Author Response

Please see the attachment.

Author Response File: Author Response.docx