A Novel CA-RegNet Model for Macau Wetlands Auto Segmentation Based on GF-2 Remote Sensing Images

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Use the following parameters like mAcc, mIoU, aAcc, Precision, Recall and F-1 for evaluation indicators. This study  will  then explore the potential and effectiveness of multiple methods, including data image processing, machine learning and deep learning.

Whether the methodologies used by you were able to efficiently and accurately extracting small-scale features for final prediction.

You can use common class imbalance in wetlands and can check wehther the loss function using the combination of CE_Loss and FocalLoss can improve the classification accuracy of difficult classes and enhance the performance and generalization ability of the model or not. 

Discussion part need to be improved including different parameters.

Author Response

We would like to express our great appreciation to you for comments on our manuscript. The details of the responses are in the document below. Thank you very much and have a nice day. Looking forward to hearing from you.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

A Novel CA-RegNet model for Macau Wetlands Auto Segmentation Based on GF-2 Remote Sensing Images

I think that the study on wetlands is very valuable and informative. However, comments regarding the paper are included below.

·         There are some concerns in terms of grammar and semantic integrity throughout the paper, especially in the abstract section of the study. I recommend that the study be proofreading by experts whose native language is English.

·         Previous studies on wetlands need to be included more. Thus, readers and referees can understand the place of the study in the literature.

·         The contribution of the study to the literature, its innovation and the gap it will fill should be stated more clearly.

·         After introducing the ResNet, Res2Net, RegNet, EfficientNet, ConvNeXt, Inception-RegNet, SE-RegNet and CA-RegNet models, it should be clearly stated for what purpose it was decided to use which deep learning algorithm. Apart from this, these sections should be shortened by referring to studies in the literature within the scope of introducing the models. Otherwise, the study may be classified as a method review paper.

·         It should be explained what is taken into consideration when choosing epoch numbers. It should be stated how underfitting and overfitting situations are avoided.

·         What the measurement parameters mean should be clearly explained in order to provide guidance for readers who are not experts in this field.

·         Normalized matrices should be presented so that the relationship between confusion matrices can be understood.

Author Response

We would like to express our great appreciation to you for comments on our manuscript. The details of the responses are in the document below. Thank you very much and have a nice day. Looking forward to hearing from you.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This is a valuable study in the light of major shifts in global equilibrium engineered by the disruption of the earth's heat balance by widespread contamination of both the atmosphere and oceans.

The Macau methods are a well chosen example and the modelling approach clearly documented and illustrated.  This work is supported by that of others and adequately referenced.

The modelling provides a definitive approach to the role of wetlands and the analysis utilizes multiple methods that support the findings.

The manuscript is well structured and well presented and is recommended for publication as presented.

Author Response

We appreciate you taking the time to review our manuscript and thank you for your recognition. Thank you very much and have a nice day.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

You did a god job ! Congratulations!!!