LPASS-Net: Lightweight Progressive Attention Semantic Segmentation Network for Automatic Segmentation of Remote Sensing Images

Round 1

Reviewer 1 Report

Title: LPASS-Net: Lightweight Progressive Attention Semantic Segmentation Network for Automatic Segmentation of Remote Sensing Images

 

The semantic segmentation network (LPASS-Net) is proposed in the article. The presentation is very accurate and the information presented sequentially.

1. In the abstract section, the aim, the methodology of the study, and the quantitative values of the results should be presented. Reduce the Abstract section. It is too big.

2. Keywords should not be the same as title words. Please correct it.

3. The Introduction chapter explains the methods consistently. The purpose of the research was not clearly presented. Separately, add the purpose of the research.

4. In the entire text, abbreviations should be explained where they are first used, and only abbreviations should be used afterward. Please, check the entire text.

5. Provide references for mathematical expressions not derived from this work. All the mathematical equations need references, check it.

6. I recommend that the Experiments and Discussion chapter be changed to separate: 4. Results; Conclusion and Discussion. Formulas and methodology should be in 3. Methodologies. Rearrange the text of 3, 4 chapters.

7. The list of references could be more extensive. Now, it is only 33 pieces.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors propose the LPASS-Net for category segmentation of large-size and high-resolution remote sensing images. The experiments are abundant, and the results are convincing. The following minor comments should be addressed before this work can be published.

(1) The abstract is too redundant. It should be simplified.

(2) The ground truth is suggested to show in the visual results, for example, Fig. 15 and Fig. 16.

(3) For EPCP, please add detailed descriptions. How to perform edge filling?

(4) The experimental results demonstrate its advantage over other networks. Maybe other methods can also obtain better performance using EPCP post-processing, please clarify this via experiments.

 

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

I have carefully read the paper titled “LPASS-Net: Lightweight Progressive Attention Semantic Segmentation Network for Automatic Segmentation of Remote Sensing Images” which proposes a light segmentation algorithm for remote sensing imagery segmentation. Several issues should be addressed by the authors.

-       -  There are typos and grammatical errors that should be solved. For instance, it is written, “The proposed lightweight non-local convolutional attention network (LNCA-Net) is a spatial dimensional attention mechanism that autocorrelation on global feature maps by integrating global information.” Autocorrelation is a noun!!

-        - The quality of Figure 1 is low. It is not readable.

-      - Why is the size of the final output of your model in Figure 2 (256,256,3)? Why did you flatten the results before the final output? It is a bit strange to me. Does it improve the results?

-        - The quality of Figure 13 is too low, and fonts are too small.

-       -  In the figures, ground-truth images are required.

Overall, the paper is well-written and well-described.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The manuscript is appropriat after correction.