AMM-FuseNet: Attention-Based Multi-Modal Image Fusion Network for Land Cover Mapping

Round 1

Reviewer 1 Report

The manuscript entitled: “AMM-FuseNet: Attention based Multi Modal Image Fusion

Network for Land Cover Mapping” overall provides an interesting account of the methods used. However, more could be done. The following are some of the comments which need critical attention for improving the paper:

 

- The authors should clearly explain what the innovative contributions of this manuscript to science is.

 

- Please include a methodological framework.

 

- What limitations does your study have? what have you done to minimize these limitations?

 

- The conclusions section should be expanded. More specifically, the authors should emphasize the contribution and implication of the study to science.

 

- Minor grammar and punctuation errors can be found throughout the text and need to be corrected.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Human activities have caused various environmental problems. Having reliable land cover information could provide support to solve the environmental problems. With the development of remote sensing technology, numerous data obtained by different kinds of sensors are adopted to observe the earth and produce the land cover map. In the past few decades, to improve the performance of land cover classification, scholars studied a variety of algorithms for multi-modal data fusion. However, there are still challenges for it, as the authors mentioned. In this paper, the authors proposed a novel dynamic deep network architecture, AMM-FuseNet, for the purposes of the land cover mapping application. And the AMM-FuseNet showed a better performance than other state-of-the-art methods. This method has merit and seems to fit well with the scope of the journal and the interest of Remote Sensing readers. However, several details need to be modified in the revised manuscript.

My minor concerns as follows:

1. At line 79, there is a typing mistake, please correct it.

2. Why choose Hunan, DFC2020, and Potsdam as the test dataset, are they representative or are they commonly used test datasets in the industry. It should be mentioned in the paper.

For the rest of the manuscript, it is well-written and solid.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

- The study focused on using a multi-modal remote sensing Image fusion network AMM-FuseNet for Land Cover mapping.

- The manuscript is a very good contribution in the field of remote sensing, particularly applying deep learning in land cover mapping applications

79- re-check this sentence "deep learning learning in computer vision"

310- Do you mean SAR (Sentinel-1)?

313- Did you check for the speckle filtering in the pre-processing of SAR data to reduce the noise in the SAR image before classification?

310- Re-check the spatial resolution of Sentinel-1 data is it 20m or you did downscale to reach 10m?

324- In which year this classification of the different land cover classes was carried out?

- What is the temporal resolution of the data used in the current study? I which years the data were acquired?

329- The same as point 324, in which year this classification of the different land cover classes in the study area was carried out? 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Thanks to the authors for the changes made to the manuscript. It has improved a lot. In my opinion the manuscript can be accepted for publication.

Reviewer 3 Report

Accept in present form