TRDet: Two-Stage Rotated Detection of Rural Buildings in Remote Sensing Images
Round 1
Reviewer 1 Report
1.Fast and accurate acquisition of the outline of rural buildings on remote sensing images is an efficient method to monitor illegal rural buildings, and the topic is very meaningful. However, the research goal of the paper is not clear, and the problems in relevant research of acquisition of the outline of rural buildings are not clear enough.
2.If the paper aims to solve the problem that the rotated object detection methods are easy to lose location information and sensitive to noise in the abstract, but the experiment doesn’t design any indicators for assessment of the accuracy of the location information.
3.The results and methods should be separated into different sections. In the results section, there are many contents about methods in all parts of 3.1, some parts of 3.2 and 3.3, which need to be moved to the methods section.
Author Response
Dear reviewer :
We thank you very much for reading our article carefully and making comments. These opinions are very important to us. Please see the attachment.
Sincerely.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Dear Authors,
This paper presents a two-stage rotated object detection model for rural buildings 381 based on deep feature fusion and pixel attention modules. The paper is well written and the topic is of great importance, but it presents some issue that should be solved before its publication.
GENERAL COMMENTS
- The proposed method shoul de better explained because it is hard to replicate the study based on what is reported in the methods;
- Discussions section should be improved with some comparison between the obtained results and those obtained in similar studies;
SPECIFIC COMMENTS
-lines 26-27: Remote sensing (RS) is not used only in hydrological context. Please add more usage of RS (e.g., agricultural application (doi:10.1080/22797254.2021.1951623); fire severity (doi: 10.3390/rs11141735); habitat monitoring (doi:10.1016/S0169-2046(03)00028-8); biomass estimation (doi:10.1007/978-3-319-92099-3_21); ecc...);
- line 88: Please use the full name for TRDet;
- lines 104-106: Please give more information about the UAV dataset used. It is a free available dataset? How is it structured? What are the technical specifications of the UAV and the sensor used? What are the parameters of the flight (mean flight altitude, percentage of overlap between images, ecc...)? And so on...
- line 140: The TRDet network achitecture proposed in Figure 3 is not easily readble (What C1 to C4 are?, ecc...). Please try to re-draw the workflow in order to improve the understanding of the proposed method for any potential readers;
- lines 142-147: Please move these lines to Introduction section among the reasons that prompted the authors to propose their method;
- line 225-233: These lines are not reporting results. Please move to Method section.
Author Response
Dear reviewer :
We thank you very much for reading our article carefully and making comments. These opinions are very important to us. Please see the attachment.
Sincerely.
Author Response File: Author Response.docx
Reviewer 3 Report
This manuscript reports the use of a DFF network for building detection in remote areas based on remote sensing images.
The manuscript is generally well written and explains the problem, the methods and the results.
There is not a significant novelty, since it is just an application of a set of techniques that we know being able to provide good results.
Yet, very good results are attained and, therefore, I believe that this manuscript can be accepted.
Author Response
Dear reviewer :
We thank you very much for reading our article carefully and making comments. These opinions are very important to us. Please see the attachment.
Sincerely.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
Dear Authors,
thanks for the effort in taking into account all comments.
The point n.2 has been only partially addressed. Please add some comparison with similar studies.
All other points have been adequately addressed.
Author Response
Dear reviewer:
We thank you very much for your suggestions. We are so sorry that some problems were not completely corrected due to our misunderstanding last time. At present, they have been corrected. Please see the attachment. thank you.
Author Response File: Author Response.docx
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.
