Plant Disease Diagnosis Using Deep Learning Based on Aerial Hyperspectral Images: A Review

Round 1

Reviewer 1 Report

-- Most of the figures are derived from published papers. I still don't think it is appropriate to use the figures created by other authors, even though you added references for the figures. I strongly recommend to re-design some figures and delete some figures

-- It is incomplete for the coordinate line of vertical axis for Figure 4(a).

-- Some well-known parts can be removed or shortened (e.g., 2.2. Deep Learning)

-- 3. Related Works. What is the role of the section? It is just a list to analyze the primary works of references.

-- The titles of the manuscript are inappropriate such as 2. Materials and Methods, 4. Discussion. It is a review article rather than a research article, so the organization structure should be specific and logical.

-- What roles do the UAVs place in Section 5. Conclusion? Most of the contents are focused on deep learning-based HRS.

Author Response

Thank you for your comments and suggestions regarding for our manuscript. We highly appreciate your revisions, since it directs our paper to the better quality. Therefore, we try our best to response to your comments and improve our manuscript. We send the revised version of our manuscript, along with the response letter that address your comments one by one.

Author Response File: Author Response.docx

Reviewer 2 Report

1- The title is not appropriate. Because it seems to be a research type article by saying "A Deep Learning Based ....".

2- There should be one or more benchmark tables for the methods based on Plant Disease Diagnosis.

3- The feature extraction part is not sufficient. What kind of different features are extracted from the data and how are they extracted? This part should be enriched more and more.

4-The regression between the vegetation indexes can be given in order to clarify which ones can be or should be used for disease detection. Some references can be the followings:

A. Jopia, F. Zambrano, W. Pérez-Martínez, P. Vidal-Páez, J. Molina, and F. de la Hoz Mardones, ‘‘Time-series of vegetation indices (VNIR/SWIR) derived from sentinel-2 (A/B) to assess turgor pressure in kiwifruit,’’ ISPRS Int. J. Geo-Inf., vol. 9, no. 11, p. 641, Oct. 2020, doi: 10.3390/ijgi9110641.

Y. Cimtay, B. Özbay, G. Yilmaz and E. Bozdemir, "A New Vegetation Index in Short-Wave Infrared Region of Electromagnetic Spectrum," in IEEE Access, vol. 9, pp. 148535-148545, 2021, doi: 10.1109/ACCESS.2021.3124453.

5-The introduction part should also make discussion about the usage of hyperspectral cameras in land cover classification, the advantages of using hyperspectral camera instead of using multispectral and/or RGB images should be mentioned.

6- What kind of diseases can be diagnosed from hyperspectral images ? I mean the methods based on detecting different kinds of diseases should also be discussed.

7- Is the disease diagnosis problem considered pixel-based ? How one can implement a precision detection (when the target is in sub-pixel) ?Related literature should be included.

 

 

 

 

 

 

Author Response

Thank you for your comments and suggestions regarding for our manuscript. We highly appreciate your revisions, since it directs our paper to the better quality. Therefore, we try our best to response to your comments and improve our manuscript. We send the revised version of our manuscript, along with the response letter that address your comments one by one.

Author Response File: Author Response.docx

Reviewer 3 Report

The article presents an in-depth review over the possibilities of plant pathogen detection via UAV-based remote sensing and data analysis with a focus on deep learning. In addition to reviewing the available literature the manuscript also provides insight on related themes, such as data preprocessing and model evaluation.

The article is well written and structured, giving a cohesive overview of the relevant information of topics from sensor types, UAVs and data analysis methodology. The reviews of the respective theme complexes are well crafted and relate interesting information.

 

Specific comments:

-        Line 42: “by the same microorganisms”, as there are no microorganisms mentioned earlier in the text it might be prudent to rephrase without the use of “same”.

-     Line 124-126: Due to the potential of snapshot-based hyperspectral sensors for UAV-based applications and their usage in one of the investigated studies later in the article (Zhang et al. 2019) it would be advantageous to include them in the sensor type overview.

-          Figure 3: The resolutions of the leaf images presented in the figure does not seem to match the labelled pixel sizes?

-          Line 318: Replace “the tarp” with “a tarp”.

-      Line 611: Change “lower resolution” into “lower spatial resolution” to clarify.

-       Line 611: The reviewer suggests replacing “high-resolution bands” with “high spectral resolution”.

 

Author Response

Thank you for your comments and suggestions regarding for our manuscript. We highly appreciate your revisions, since it directs our paper to the better quality. Therefore, we try our best to response to your comments and improve our manuscript. We send the revised version of our manuscript, along with the response letter that address your comments one by one.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors have carefully considered all the issues raised in the previous round of review.

Reviewer 2 Report

Paper is improved with respect to my previous comments.