Detection of Pine Wilt Disease Using Drone Remote Sensing Imagery and Improved YOLOv8 Algorithm: A Case Study in Weihai, China
, Changbo Yi 1, Zheng Zhang 1, Hang Fei 1, Hongwei Zheng 2,3, Haoran Jiang 6, Yanchun Jiang 7, Xianfeng Zhao 8, Xiaojing Zhao 9 and Pengsen Yang 10
Round 1
Reviewer 1 Report
This study is a significant contribution in the field of Smart Agriculture. The authors opted to a combination of Drone Remote Sensing Imagery along with a well tuned version of YOLOv8 Algorithm for the detection of Pine Wilt Disease. I have suggested some minor changes for the improvement of the manuscript.
Comments for author File: 
Comments.docx
Author Response
Please see the attachmentAuthor Response File: Author Response.docx
Reviewer 2 Report
This study is amied to detect pine wilt disease using drone remote sensing imagery and improved YOLOv8 Algorithm. The study design is acceptable. The study contains some valuable results that can be considered after suitable revisions.
Comments/Suggestions:
L26-29: Abstract contains only one sentence about the results. Please give more detailes about your results in the Abstract.
L32: Keywords: please avoid keywords that is already in the title.
L162, L172, L243, L311, L394: Figures 1,2,4,5 and Table 2 titles are too short and not well informative. In case of Figure 4 give also the meaning of abbreviations.
L312: Results and Discussion section is mainly interpretation of results. Please create a separate Discussion section and compare your results with the previous literatures.
L505: Figure 8. Letter size is too small in the figures.
L527: Please give you conclusion in bullet points.
Number of references should be increased.
Author Response
please see the attachment
Author Response File: Author Response.docx
Reviewer 3 Report
The manuscript entitled "Detection of Pine Wilt Disease Using Drone Remote Sensing Imagery and Improved YOLOv8 Algorithm: A Case Study inWeihai, China" developed an improved YOLOv8 model for detection and identification of pine wilt disease (PWD). Overall, the manuscript was well prepared, with details provided. The following comments are recommended to be addressed for revising the manuscript:
1.Figure 1 is not an algorithm flowchart. Author should rename it.
2. In line 206, 1446 datasets represent image or diseased pine tree instance?
3. Figure 4 shows the input images of 640*640 pixels, but the author cropped and divided images into 451*451-pixel segments (in line 200).
4. In line 244, the attention modules CBAM, ECA, and GAM should be spelled out at the first mention. In addition, could the author explain why these three attention modules were selected and compared?
5. In line 379, the author stated that comparing to YOLOv8s, YOLOv8s-Atten outperforms in all metrics, including mAP50, mAP50-95, F1-Score, and Mean. Could the author elaborate why YOLOv8s-atten outperformed, but YOLOv8n-atten didn't improve the detection performance compared to YOLOv8n and YOLOv8n-small? Especially, YOLOv8n-atten applied both improved mechanisms, i.e., small object detection layer and attention mechanism.
6. The introduction section should provide relevant literature about attention module insertion location.
7. Figure 6 is too small to recognize. The detection results are hard to see clearly.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Congratulations to authors, they have done well under the light of my suggestions. Best of Luck
Reviewer 2 Report
The authors have prepared the requested changes. The study has improved a lot.
Reviewer 3 Report
The author made corresponding corrections and explanations. The manuscript can be accepted.
