UAV-Based 3D-Calibration of Thermal Cameras for Bat Flight Monitoring in Large Outdoor Environments
Round 1
Reviewer 1 Report (Previous Reviewer 1)
Comments and Suggestions for AuthorsI noticed that the last time I reviewed this manuscript, the overall quality of the manuscript has improved a lot since the last time. However, there are still some problems for the author to modify
1. The text in Figure 2 is confused.
2. The author needs to add information on how the internal orientation elements are calibrated, which may have been calibrated before installation, but the description should be added.
3. 2.5Automated 2D image points detection, the manuscript lacks the description of the data processing Results of this part, which should be supplemented in 3. Results. In other words, the author needs to supplement the drone images taken (or partial magnification images).
4. The author takes the drone image as the control point, but the impact captured should be a planar factor. How should the author handle it？
5. The purpose of this study is to monitor the movement of bats. But there are no images of bats in the text. It is suggested that the author make a supplement in the appendix.
6. The number of references seems to be 28, which is relatively small.
7. Figure A5-7. Please include units for the axes
8. The coordinate system of the UAV is WGS84 coordinate, while the scene captured by the camera is 3D object coordinate system. In the process of processing, the problem of coordinate system transformation is involved, so it is recommended to describe this process in more detail
9. In Figures 9 and 10, please note the normalization of the title and unit writing of all relevant axes.
Author Response
Thank you for your review. We used your comments to improve our manuscript. We gave a detailed point-by-point response in the attached PDF.
Author Response File: 
Author Response.pdf
Reviewer 2 Report (Previous Reviewer 2)
Comments and Suggestions for AuthorsWell-written, previous responses to review comments have answered questions
Author Response
Thank you very much for your review and also for your helpful comments in the last round.
Reviewer 3 Report (New Reviewer)
Comments and Suggestions for AuthorsI found this work really sound and of high practical importance.
Author Response
Thank you for your positive feedback and for recognizing the practical importance of our work.
Reviewer 4 Report (New Reviewer)
Comments and Suggestions for AuthorsThe paper proposes a method for 3D calibration based on unmanned aerial vehicles (UAVs) equipped with two thermal imaging cameras, aiming to automatically detect flight paths in wind farms. But there are the following issues:
1. The contribution expression is not clear, and only the goal is presented at the end of the Introduction. It is recommended to display the contribution of the paper item by item;
2. The background introduction does not elaborate on the characteristics of wind power plants, nor does it explain the differences between the paper method and traditional non wind power plants. It is suggested to optimize the wording;
3. The introduction of the relevant work is too simple, only providing an overview of the methods used. The existing methods have not been clearly explained on how to solve the existing problems. It is suggested to add more relevant explanations;
4. The layout of the entire text needs to be further optimized, as the table layout is quite messy;
5. The references are relatively old and must be replaced with the latest ones;
6. The paper lacks comparison with other latest methods. It is suggested to add relevant experimental comparisons.
Author Response
Thank you for your review. We used your comments to improve our manuscript. We gave a detailed point-by-point response in the attached PDF.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report (Previous Reviewer 1)
Comments and Suggestions for AuthorsI have reviewed this manuscript for three times, the author has made many optimization and improvement, and added the necessary data and discussion, I think it is ready for publication.
Comments on the Quality of English Languageok
Author Response
Thank you for your time and constructive feedback throughout the review process. We are glad to hear that you are now satisfied with our revisions.
Reviewer 4 Report (New Reviewer)
Comments and Suggestions for AuthorsMost of the questions have been revised as required, but there have only been 5 references in the past three years. The proportion of references in the past three years must be increased to over 80%, otherwise the topic studied in this article has no research trend and is already outdated.
Author Response
Thank you for reviewing our manuscript again, we attached our response.
Author Response File: Author Response.pdf
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
Comments and Suggestions for AuthorsI don't think this is a well-rounded research topic.
1.The author's research content mainly involves using drones for 3D Calibration of Two Thermal Cameras. While this can be used for tracking bats and analyzing the impact of wind farms, the entire paper lacks methods for tracking bats and analyzing the impact of wind farms, as well as relevant experiments to support these methods.
2.If the research content is 3D Calibration of Two Thermal Cameras, it seems more suitable for journals in photogrammetry or computer vision, rather than remote sensing journals.
3.The paper includes numerous schematic diagrams, but no actual drone images or 3D Calibration of Two Thermal Cameras.
4.Using the GPS location of the drone as a control point for camera calibration seems like a feasible idea. However, the GPS location of the drone itself has errors, with a positioning accuracy of a few meters. Using this as a control point for 3D calibration seems impractical. The author needs more thorough experiments and data to support this. Although using the calibration results directly for bat tracking may partially offset the GPS errors to some extent.
However, the paper lacks any content related to tracking bats. Therefore, I believe this research is very incomplete and not suitable for publication.
5. The first three sentences of the abstract are confusing: the first sentence is about 3D camera calibration, the second sentence is about the impact of wind farms, and the third and fourth sentences talk about the problem caused by small targets. The second sentence seems out of place. The author needs to reorganize the language to introduce the research problem. For example, they could start by discussing the impact of wind farms on bats and then introduce the need to track bats and the camera calibration issues they face. The title is "Bats," and the abstract mentions thermal cameras. The author needs to be more precise in their wording throughout the paper, including in the title.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper presents a novel unmanned aerial vehicle (UAV)-based 3D camera calibration method for tracking nocturnal flying bats around wind turbines. This method is highly original for animal flight path tracking using thermal imaging cameras in a wide range of outdoor environments
1. the method mentioned in the article relies on the accuracy of the UAV's GPS signal, which may affect the quality of the 3D calibration and reconstruction if there are large errors in the GPS signal?
2. the article mentions that the UAV flight is manually operated by a human, which may introduce operational variability and affect the consistency and accuracy of the calibration?
3. Existing methods use reference structures of known size for calibration, but these methods may not be applicable in large-scale outdoor environments where the size of the reference object limits the distance between cameras.
4. The article mentions that relying on 2D reprojection errors for 3D calibration may not fully resolve viewpoint and scale ambiguities, which may lead to uncertainty in the final position of 3D points.
5. the flight behavior of bats can be very complex, and the methods in the article may require further validation to ensure robustness under different flight behaviors and environmental conditions
6. although the article provides an overview of the method, some implementation details may be missing, especially for the specific parameter selection and optimization process of the algorithm.
7. The article has already mentioned some limitations in the discussion section and suggested possible solutions. Further details on the impact of these limitations on the results of the study could be provided, and where possible additional data or analyses to improve the methodology could be provided.
8. Is the number of references too low?
9. could the Introduction section be divided into paragraphs.
Author Response
Comments 1: the method mentioned in the article relies on the accuracy of the UAV's GPS signal, which may affect the quality of the 3D calibration and reconstruction if there are large errors in the GPS signal?
Response 1: Yes, the errors are usually within 1.5 meters for our used drone (line 130) and we do cross-validation by different flights for 3D calibration and error evaluation because of that reason (line 85). The drone error is included then. But we mentioned it as a limitation also (line 283).
Comments 2: the article mentions that the UAV flight is manually operated by a human, which may introduce operational variability and affect the consistency and accuracy of the calibration?
Response 2: Yes, that is also a trade-off between cheap equipment and accuracy. Doing automated flights would be more precise. This is the reason for the variability in the error plot of Figure 8. Doing automated drone flights would reduce the errors though. We addressed that in the limitations on line 257
Comments 3: Existing methods use reference structures of known size for calibration, but these methods may not be applicable in large-scale outdoor environments where the size of the reference object limits the distance between cameras.
Response 3: Yes that is true, therefore we devolped the method (line 7).
Comments 4: The article mentions that relying on 2D reprojection errors for 3D calibration may not fully resolve viewpoint and scale ambiguities, which may lead to uncertainty in the final position of 3D points.
Response 4: Yes, we mentioned that on line 54. This is well-known and can be found in "Multiple View Geometry" from Hartley and Zisserman. We will add the source here, thank you.
Comments 5: the flight behavior of bats can be very complex, and the methods in the article may require further validation to ensure robustness under different flight behaviors and environmental conditions
Response 5: The frame rate of the cameras is up to 30 images per second. You can see a .GIF image of a bat passing the turbine on our GITHUB repository: https://github.com/christofhapp/batflight3D_publication (please scroll down)
Using 30 Hz as a frame rate is enough for most flight conditions of bats, but you are right, as the distance to the camera system gets higher, the flight path is not exact anymore on a small scale to the 3D resolution (Figure 3). We will add that in the discussion.
Comments 6: although the article provides an overview of the method, some implementation details may be missing, especially for the specific parameter selection and optimization process of the algorithm.
Response 6: Yes, a lot of implementation details are missing, because the full code is 6000 lines and there are lots of chosen parameters. But we provide a reduced code on our GITHUB repo to reproduce all the findings of the publication including the data. Discussing all parameters would have been too much for this publication as it is already 16 pages long. We only discussed the important ones. Many others are camera or site specific e.g.
Comments 7: The article has already mentioned some limitations in the discussion section and suggested possible solutions. Further details on the impact of these limitations on the results of the study could be provided, and where possible additional data or analyses to improve the methodology could be provided.
Response 7: More accurate RTK drones (line 308) and automated flight patterns would improve the results (line 304). They have the drawback of being much more expensive, but it would be interesing to see the improvements. We did not do it, because the errors were low enough for our purpose and much better than related work (and we do not own a RTK drone).
Comments 8: Is the number of references too low?
Response 8: It is a very special interdisciplinary research topic. We tried to mention all research topics which are close enough to ours, but yes, i think we could add some more sources here.
Comments 9: could the Introduction section be divided into paragraphs.
Response: Yes, thank you, that may help to give it a better visual structure.
