A Survey on Different Plant Diseases Detection Using Machine Learning Techniques
, Michal Jasinski 3,*, Zbigniew Leonowicz 3, Elzbieta Jasinska 4, Tomas Novak 5 and Arnab Kumar Maji 2,*
Round 1
Reviewer 1 Report
1. Could you provide the definition of 'accuracy' in table 4?
2. It would be better if you can add some run time analysis on each algorithm, how scalable is the algorithm?
3. Can you also be specific about the accuracy metrics in table 5? For example, there are methods which are addressing 1 class identification problem, there are methods which are measuring multi-class identification, so the measures are different, it is hard to tell which metric is used in the table. Sometimes there are two results in one cell, can you make them into two cells or label each of them?
4. In table 5, there are some methods that do not have accuracy, but still listed in the table, which is confusing, maybe add a note to the method which does not have an accuracy? do they have other measures?
5. In table 6, what is the threshold of 'Large number of images in dataset'?
6. In figure 3 and 4, could you add a time frame of the study? between a start year to an end year, what's the histogram?
Author Response
At the beginning I would like to thank Reviewers for their time and valuable contributions. All changes, that resulted from review were introduced and highlighted in the pdf.
Reviewer#1, Concern # 1: Could you provide the definition of 'accuracy' in table 4?
Author response: Thank you sir for your valuable remark
Author action: We have updated the manuscript and include the definition of accuracy of accuracy as note in below Table 4.
Reviewer#1, Concern # 2: It would be better if you can add some run time analysis on each algorithm, how scalable is the algorithm?
Author response: Dear Reviewer. Thank You for this remark. In this work, we have mainly focused on what dataset author used, how many classes are there in the dataset, what machine learning approaches they have used and the performances of their approach in terms of accuracy/precision/f1-score. In most of the paper run time analysis is not available and moreover the run time of models depends on the machine configuration. In our next work, we will analysis models with respect to run time. Thank You once again.
Author action:
Reviewer#1, Concern # 3: Can you also be specific about the accuracy metrics in table 5? For example, there are methods which are addressing 1 class identification problem, there are methods which are measuring multi-class identification, so the measures are different, it is hard to tell which metric is used in the table. Sometimes there are two results in one cell, can you make them into two cells or label each of them?
Author response: Thank you sir for your valuable remark. For both single class and multiclass approach, the term accuracy defines the ratio of number of correct recognitions to the total number of images in the test set.
Author action: In some work, author used precision/f1-score to measure the performance accuracy of their model. We have updated the manuscript and added the performance metrices along with which metric they have used. In the tables where two results were there in one cell, we have labeled them according to their datasets/classifiers/deep learning models. For example, in Table 5 the accuracy was given 95.65 and 94.3. We have updated this as 95.65(AlexNet) and 94.3 (SqueezeNet).
Reviewer#1, Concern # 4: In table 5, there are some methods that do not have accuracy, but still listed in the table, which is confusing, maybe add a note to the method which does not have an accuracy? do they have other measures?
Author response: Dear Reviewer. Thank You for this remark.
Author action: We have updated the manuscript and added the accuracies.
Reviewer#1, Concern # 5: In table 6, what is the threshold of 'Large number of images in dataset'?
Author response: Thank you sir for your valuable remark
Author action: We agree with reviewer’s comments and updated Table 6 in the manuscript and added the threshold criteria for large number of image dataset in the note section below the table.
Reviewer#1, Concern # 6: In figure 3 and 4, could you add a time frame of the study? between a start year to an end year, what's the histogram?
Author response: Dear Reviewer. Thank You for this remark.
Author action: We have updated the manuscript and include one figure as Figure 5 where we have put the histogram of work done with respect to number of years.
Reviewer 2 Report
Dear authors,
I enjoyed reading your manuscript "A study on different Plant Disease Detection using Machine Learning Techniques". It is a nice contribution worthy of publication, but I have some comments that I believe can improve your manuscript.
Title: This is a review paper and should be reflected in the title, otherwise the reader will be misled into thinking it is a research paper.
Figures: The quality of the figures in the paper is not good. It is suggested to redraw the pictures in the paper to make them more informative and attractive to readers.
Tables: The tables in the paper need to be revised to the journal's standard format. There is a question, what does the "NA" in many tables (e.g., Table 2, Table 4) stand for, which needs further explanation from the author.
Others: The author needs to further summarize when introducing the previous research done by the researcher. It is not acceptable that so many paragraph is about “who did what work...” as the beginning.
Author Response
At the beginning I would like to thank Reviewers for their time and valuable contributions. All changes, that resulted from review were introduced and highlighted in the pdf.
Reviewer#2, Concern # 1: This is a review paper and should be reflected in the title, otherwise the reader will be misled into thinking it is a research paper.
Author response: Dear Reviewer. Thank You for this remark.
Author action: We have updated the manuscript and renamed the manuscript from “A Study on different Plant Disease Detection using Machine Learning Techniques” to A Survey on different Plant Disease Detection using Machine Learning Techniques.
Reviewer#2, Concern # 2: The quality of the figures in the paper is not good. It is suggested to redraw the pictures in the paper to make them more informative and attractive to readers.
Author response: Dear Reviewer. Thank You for this remark.
Author action: We have updated the manuscript and redraw the pictures in the paper.
Reviewer#2, Concern # 3: The tables in the paper need to be revised to the journal's standard format. There is a question, what does the "NA" in many tables (e.g., Table 2, Table 4) stand for, which needs further explanation from the author.
Author response: Dear Reviewer. Thank You for this remark.
Author action: We have updated the manuscript and added the explanation of NA in the below of each table as note. NA defines the information not available in that particular paper.
Reviewer#2, Concern # 4: The author needs to further summarize when introducing the previous research done by the researcher. It is not acceptable that so many paragraph is about “who did what work...” as the beginning.
Author response: Dear Reviewer. Thank You for this remark. In this paper, we have summarized together those papers which used same techniques. In most of work, we have discussed those papers along with their advantages and disadvantages.
Author action:
