Advantages of Using Transfer Learning Technology with a Quantative Measurement

Round 1

Reviewer 1 Report

1. Transfer learning has already been applied in land-cover mapping tasks. However, the authors only evaluate fine-tuning techniques for building mapping, which is not enough. Actually, unsupervised domain adaptation (UDA) algorithms in remote sensing are suitable for transfer learning, and there already exist many methods [1, 2, 3]. These advanced methods need to be included.

[1] Wang J, Ma A, Zhong Y, et al. Cross-sensor domain adaptation for high spatial resolution urban land-cover mapping: From airborne to spaceborne imagery[J]. Remote Sensing of Environment, 2022, 277: 113058.

[2] Zhang L, Lan M, Zhang J, et al. Stagewise unsupervised domain adaptation with adversarial self-training for road segmentation of remote-sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 60: 1-13.

[3] Yan L, Fan B, Liu H, et al. Triplet adversarial domain adaptation for pixel-level classification of VHR remote sensing images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 58(5): 3558-3573.

2. The experiments only compare the models without pre-training and the models pre-trained on open-source remote sensing data. This is unfair because existing general methods are already equipped with many pre-trained weights on ImageNet, CityScapes, etc. It is unfair to discard these as baseline settings. More comparison experiments about UDA methods need to be compared.

3. As a method test for transfer learning, I am concerned about the comparison between the proposed method and the recent SAM model[4]. The latter can be applied to remote sensing mapping in a zero-shot way.

[4] Kirillov A, Mintun E, Ravi N, et al. Segment anything[J]. arXiv preprint arXiv:2304.02643, 2023.

[5] Osco L P, Wu Q, de Lemos E L, et al. The Segment Anything Model (SAM) for Remote Sensing Applications: From Zero to One Shot[J]. arXiv preprint arXiv:2306.16623, 2023.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Title of the paper: "Advantages of using transfer learning technology with a quantative measurement"

     As a researcher working in the same field, I am impressed by the technique introduced in the paper because it sheds new light on the earlier results of several authors and obviously can be successfully used in practice. From this point of view, the subject of the paper fits well with the scope of the journal (Remote Sensing).

 The paper is ended with numerical simulations that corroborate the theoretical results.

This manuscript contains new ideas and good results that help other researchers.

The decision is too a major revision for publication in "Remote Sensing".

 Therefore, I recommend publishing this work after taking these points into account.

1. Introduction needs to explain the main contributions of the work more clearly.

2. The novelty of this paper is not clear. The difference between the present work and previous Works should be highlighted.

3.  What does the research address the main question?

4. Do you consider the topic original or relevant in the field? Does it

address a specific gap in the field?

5. What does it add to the subject area compared with other published

material?

6. What specific improvements should the authors consider regarding the

methodology? What further controls should be considered?

7. Are the conclusions consistent with the evidence and arguments presented, and do they address the main question posed?

8. Please include any additional comments on the tables and figures.

9. The authors can add the following reference to enrich the introductory section:

*Two computational algorithms for the numerical solution for system of fractional differential equations, Arab Journal of Mathematical Sciences, 2015, 21(1), pp. 39–52.

*Optimal control, signal flow graph, and system electronic circuit realization for nonlinear Anopheles mosquito model, International Journal of Modern Physics C, 2020, 31(9), 2050130.

5- Future recommendations should be added to assist other researchers in extending the presented research analysis.

Sincerely

Yours

Minor editing of English language required

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Dear authors 

I am willing to see how the findings would change after you resolve the limiting factors ( if possible ). That could be your next paper. Interesting reading! 

Fine.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

The manuscript’s presentation is clear, and its conclusion is generally robust and well-supported by the experiments. However, I suggest enhancing it further by incorporating the following: 1) While you've discussed the utilization of transfer learning techniques based on U-Net, providing more detailed information about the specific transfer learning methods employed and their rationale could offer readers a deeper understanding of your approach. 2) Consider dedicating a section to discussing potential future directions for research in this area. This could spark further interest and collaboration among researchers working on similar topics.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors have addressed all my concerns.

The writing seems good.