Review Reports
by
- Brett Lawrence 1,*,†,
- Emerson de Lemmus 2,† and
- Hyuk Cho 2
Reviewer 1: Qing Li Reviewer 2: Anonymous Reviewer 3: Buddhika Madurapperuma
Round 1
Reviewer 1 Report (Previous Reviewer 2)
Thanks again. However, the authors still refused to consider the previous comments, contents cannot meet the requirements of a SCI journal at all.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report (New Reviewer)
Manuscript Number: remotesensing-2121376
Full Title: Towards UAS-based real-time detection of small wildlife objects in heterogeneous landscapes: finding Red-cockaded Woodpecker cavities using computer vision.
The paper submitted to Remote Sensing MDPI by Brett Lawrence et al. describes an application of a real-time object detection algorithm to effectively detect red-cockaded woodpecker tree cavities, via UAV, in Sam Houston National Forest, near Houston, USA.
The topic is of considerable interest to the scientific community and the paper is analysed in original submission. The paper has been edited according to the RS MDPI instructions for authors and is set up correctly in all its paragraphs, except references.
In my opinion, the article is enjoyable to read and the study is adequately original; the methodological approach is rigorous. The introduction synthesizes the scientific background. The methods and tools are described with details to allow another researcher to reproduce the results. On the basis of what has been mentioned, there is only few minor comment, and I permit to suggest (line by line), in agreement with the notes for MDPI RS reviewers:
1. Page 1, (Pennekamp and Schtickzelle, 2013; Weinstein, 2018; Borowiec et al., 2022). Please check MDPI instructions for authors, in wich the references is [n]. Please modify in the body of the main text;
2. Page 2, after Guirado et al., 2019, please delete second parenthesis.
3. Page 4, figure 1. Please modifies the figure as it is in low resolution. It would also be useful for the reader to know the datum (which I assume WGS84).
4. Page 8, figure 4. Please modifies the figure as it is in low resolution.
5. Page 9, figure 5. Please modifies the figure as it is in low resolution.
6. Page 11, figure 6. Please modifies the figure as it is in low resolution.
7. Page 11-12, eq. 1 to 4. Please modify with word mathematical tool.
8. Page 12, please insert the technical specifications of the Garmin GPSMAP 64s.
9. Page 14, figure 7. Please modifies the figure as it is in low resolution.
10. Page 15, figure 8. Please modifies the figure as it is in low resolution.
In conclusion the work deserves publication after minor revisions, in my opinion.
Best regards
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report (New Reviewer)
MS: Towards UAS-based real-time detection of small wildlife objects in heterogeneous landscapes: finding Red-cockaded Woodpecker cavities using computer vision
This study detects the Red-cockaded Woodpecker cavities in the Sam Houston National Forest in Texas Using You Only Look Once (Yolo) object detection algorithms. The model is useful since it can detect cavities in real time, where forest managers and wildlife biologists can implement best management practices. I made comments through the pdf using sticky notes since there are no line numbers to comment in this report.
So, I listed major comments here.
The title consists of some waste words like “small wildlife objects” which is given as Red-cockaded Woodpecker cavities in the latter part of the title. Computer vision is a broad term which can specify using the Yolo algorithms.
Here is my version of the title feel free to make necessary changes as authors wish.
UAS-based real-time detection of Red-cockaded Woodpecker cavities in heterogeneous landscapes using Yolo object detection algorithms
1. The authors are not properly follow the journal guidelines which can be found through this link (https://www.mdpi.com/journal/remotesensing/instructions).
- In the main text, reference numbers should be placed in square brackets [ ]
- The reference list should follow the ACS style guide including the full title, journal abbreviation, bold year etc.
2. Some images/figures are blurred. So, add high quality images/figures with at least 300 dpi.
3. Did you complete 1,264 hectares survey using multi-rotor UAS, which seems a big area.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report (New Reviewer)
Dear Authors,
Thank you so much for addressing all the comments suggested and the current version looks good. So I recommend the article for a publication.
All the Best!
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
Good paper with relevant statistics and clear results
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
In this paper, the cavity detection performance of red-cockaded woodpeckers (RCW) was investigated by two commonly used YOLO models, named YOLO-V4 and YOLO-V5. However, this paper does not have any theoretical depth at all, for example, no formulas are involved. YOLO-V4 and YOLO-V5 with many executable codes and programs have been reported for plenty of applications in many areas including computer vision, image processing, and signal processing, etc. Therefore, the reviewer does not recommend this paper due to poor innovation and lacks models’ improvement.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
General:
Manuscript needs proofreading.
Materials and Methods:
All figures are low quality, please put better quality photos at least 300 dpi. For some of the flowcharts it's impossible to read the text.
In Figure 3, right hand image, where is the cavity? the cavity cannot be seen in the aerial image?
I am unsure how it is possible to see a small cavity in a tree from 76 meters above ground. what is the ground sampling distance? what is the average pixel size of cavities? How was the ground truth bounding box defined? was it the whole tree or just the cavity?
The authors are claiming that the FPR is 1, meaning that the probability of labelling a positive image as negative is 1. what does that mean regarding the reproducibility of their results? what are the implications? they did not expand on this in the discussion section.
The results are presented based on average precision. Average precision is not really a good indicator for showing accuracy. For target detection, IOU, TOP1, TOP3, TOP5, are usually used. The loss function was not delved into as well.
Did the authors perform any tuning of the model? How did they make sure the model they are using fits your objective? YOLOV5 has a large number of parameters, did the authors perform any pruning? Was there any complexity control done on the training process? How did you make sure the complexity of the model fits the complexity of the problem?
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
The reviewer reads the response carefully and meanwhile reads the manuscript again carefully, the authors provide an explanation reading the wildlife management using classic YOLO model without any new ideas. However, the authors refused to improve and innovate the depth of the classic model, and only natural images such as human faces are replaced with the wildlife pictures using classical model. This is not always innovative, just a kind of handling or transporting to some extent, the revised version still needs to be rejected! In addition, scientific research in remote sensing areas with zero formula or eq’s support, it is unnormal.