Early Identification of Cotton Fields Based on Gf-6 Images in Arid and Semiarid Regions (China)

Round 1

Reviewer 1 Report

Early identification of cotton fields based on GF-6 images in 2 Arid and Semiarid Regions (China)

1.     The main problem is that it appears that the paper does not start with a hypothesis to test, but rather focused solely on applying a machine learning method to the data. A more appealing scientific story writing is required.

2.       Once authors defined the hypothesis, the conclusion must support the hypothesis

3.       The model does not seem to have been developed for prospective use, but rather a post-hoc analysis. The discussion needs to propose the value of the results

4.       The author must explain the class label because random Forest is supervised learning. Has the dataset previously been labeled? If the dataset did not already have a class label, writers must explain how to label data for the dataset.

5.       Although the authors explained the purpose of this study, it needs to be clarified for the reader. Therefore, it is necessary to clarify the purpose of the study.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear authors, there are alot of work that have been done.

Nevertheless the use of English need alot of work and your work cannot be adequate presented.

A few of my thoughts..

Introduction: "Cotton as agricultural commodities reserve resources"??

Material and methods: "The source of water vapor is relatively small"?

Results and analysis: "MIoU can be used to judge the accuracy of the model's prediction"? Also i believe that the results are poor..

Conclusion and discussion: Poor discussion with no future research..

Discussion. (Again discussion?): "In this study, the accuracy was improved by adding vegetation index and texture feature set". Not good for closing lines..

The quality of the English Language is poor..

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

1. The numbering is chaotic. The title of Chapter 4 is Summary and Discussion, but there is only one section. Chapter 4 is Summary without Discussion, while Chapter 5 is Discussion.

2. There are few descriptions of attention mechanism in the paper, and the DAM model is proposed as an innovation in the paper, which lacks detailed explanation.

3. The model architecture diagram is not clear enough, please use high-definition images.

4. The structure and organization of the paper still need to be readjusted, and there is a lack of analysis of the experimental results. Apart from the numerical comparison of identification accuracy, the comparison chats of identification of different models were not presented, making it impossible to determine the true detection effect of the model.

Extensive editing of English language required

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Thanks for working hard to improve the manuscript. There are also english editing that should be done and some new relevant references in the introduction section.

Quality of English Language should be improved

Reviewer 3 Report

1. From the latest revised draft, it can be seen that the DAM model proposed in the article is the already proposed CBAM model. If there is no difference between the two attention mechanisms, the proposal of DAM in this paper will not be innovative. Please provide analysis.

2. If there is a difference between DAM and CBAM, then comparative experiments and effect analysis with current popular attention mechanisms such as CBAM need to be added to the paper.

Moderate editing of English language required