Multi-Class Pixel Certainty Active Learning Model for Classification of Land Cover Classes Using Hyperspectral Imagery

Round 1

Reviewer 1 Report

My comments are in the attached document. 

Comments for author File: Comments.pdf

Author Response

Dear Sir

We were pleased to have an opportunity to revise our paper now entitled, “Multi-class Pixel Certainty Active Learning Model for Classification of Land Cover Classes using Hyperspectral Imagery”. In revising the paper, we have carefully considered your comments and suggestions, as well as those of the reviewers. As instructed, we have attempted to succinctly explain changes made in reaction to all comments. We reply to each comment in a point-by-point fashion.

Author Response File: Author Response.docx

Reviewer 2 Report

The structure of introduction is not clear and satisfied. The motivation and novelty of this paper are not clear. Too much related work was introduced, but please focus on the main topic.

Please check the grammar errors throughout the paper.

The image qualities were low the caption of figures provided useless information with the example of figure 3.

Please highlight the best accuracies by the different method in each class.

Please list a table to demonstrate the parameters of proposed method.

Please provide the information of calculation platform.

 

Author Response

Dear Sir

We were pleased to have an opportunity to revise our paper now entitled, “Multi-class Pixel Certainty Active Learning Model for Classification of Land Cover Classes using Hyperspectral Imagery”. In revising the paper, we have carefully considered your comments and suggestions, as well as those of the reviewers. As instructed, we have attempted to succinctly explain changes made in reaction to all comments. We reply to each comment in a point-by-point fashion.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Review of “Multi-class Pixel Certainty Active Learning Model for Classification of Land Cover Classes using Hyperspectral Imagery” [V2] authored by Chandra Shekhar Yadav, Monoj Kumar Pradhan, SyamMachinathuParambil Gangadharan, Jitendra Kumar Chaudhary, Jagendra Singh, Arfat Ahmad Khan, MohdAnul Haq, Ahmed Alhussen, Chitapong Wechtaisong, Hazra Imran, Zamil S. Alzamil, and Himansu Sekhar Pattanayak

Overall, the manuscript is much improved and easier to read. The authors summarize and cite previous relevant research in depth. The authors demonstrate how their improvements to previous approaches has resulted in better model predictability using a variety of performance metrics. I do, however, still have some minor issues that need to be addressed before publication.

 

Firstly, there are still many instances where the language is not correct. This is a dense paper. The language, therefore, should make it as easy as possible for the reader to understand the presented research. I have attached an edited pdf of the manuscript; where I have highlighted in yellow indicates sections/sentences where the language needs to be revised. Sections highlighted in pink are addressed in the comments below:

 

In the abstract, I would change “The effectiveness” to “the ability of PCAL to accurately categorize land cover types is demonstrated…”

 

I am not sure I necessarily agree with the authors’ first statement. What exactly do they mean by “clear”? And fine spatial resolution is required only if identifying land cover types that vary at fine spatial scales.

 

Lns. 223-227: So much of the paper so far has built up to this statement: how does the PCAL method improve upon these earlier methods? I would use this moment to highlight the significant improvements this study presents. I am not sure how lines 224-227 support the previous statement in lines 223-224.

 

Ln. 238: “photos” should be “images”.

 

Lns. 245-246: I would say “that are required for processing of optical data”.

 

Ln. 246: I would say “accuracy” rather than “efficiency”.

 

What wavelengths are being shown in Fig. 2?

 

A colorbar might be helpful in Figs. 4 and 5? What are the units associated with results being shown?

 

Lns. 319-320 still do not make sense to me. The cells with 33% size?

 

Equation 4 is still unclear to me. “double”? Could the authors possibly use i and j to indicate row and column? It looks like the authors are referencing/indexing a temporary matrix.

 

More explanation in the caption for Fig. 6 would be helpful. It is hard to conceptualize what is being presented here, especially since the axes are not labeled.

 

The Algorithm presented after Fig. 6 is still being presented in [partial] code. There’s for loops, an elseif loop, with variables indexed. Is there a way to make this more readable? Perhaps a flow chart might be an easier way for the authors to present this?

 

Ln. 479: I think the authors need to be quantitative here. By how much did their method improve predictability? Was this significant?

 

Figure 12: Do the blue lines/dots indicate that there is a Zero false positive rate for SVM? It would be helpful if the authors made a comment about the SVM results compared to the PCAL results since both are presented in this figure.

Comments for author File: Comments.pdf

Author Response

Dear Sir

Good Evening

 

We have updated the article as per your suggestion. Hope we have presented our work as per your expectation.  We are sorry for delay from our side.

Author Response File: Author Response.docx

Reviewer 2 Report

The authors have given the comprehensive response and the reviewer agreed to accept it.

Author Response

Dear Sir

Good Evening

 

Please find updated article.

Author Response File: Author Response.docx