Tomato Leaf Disease Classification by Combining EfficientNetv2 and a Swin Transformer
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe manuscript has a good structure and development. The writing is simple and easy to understand for the topic. The introduction is well presented and gives a general overview of the work that has been done on the topic. The presentation of schemes in the methodology facilitates the understanding of the processes. I think the EfficientNetV2 model is the most suitable.
I consider that sections 3.2 and 3.3 be moved to methodology, since they are more descriptive than results. In the discussion, the studies that have been carried out and their precision are mentioned, but it is not mentioned whether all of these studies are in tomatoes or in some other type of plant.
In general, the manuscript is relevant, it shows how technology can be used in the detection of diseases in plants.
Author Response
We are very grateful to your comments and thoughtful suggestions.Based on these comments and suggestions, we have made careful modification on the original manuscript.Here are our responses to your comments.
Comments 1:I consider that sections 3.2 and 3.3 be moved to methodology, since they are more descriptive than results. In the discussion, the studies that have been carried out and their precision are mentioned, but it is not mentioned whether all of these studies are in tomatoes or in some other type of plant.
Response 1:Thank you for pointing this out. In the methodology, the proposed model is mainly described, so the processing of the data and the metrics are placed in the experiment. And other articles of the same type are also distributed in this way. In the discussion section, we are very sorry that we did not clearly state the specific object of the study and the subjects of the studies mentioned in the discussion were all tomato studies,which I have already marked in red in my submissions.
Thank you very much for your valuable comments.
Reviewer 2 Report
Comments and Suggestions for AuthorsDear Authors,
The presented work concerns the automatic recognition of diseases appearing on tomato leaves. In the analyzed text, the authors rely on data provided from two datasets. However, the description of these data is brief.
Here are a few comments on the text:
- In Figure 4, the authors distinguish different pathogenic factors on tomato leaves, but the text does not clarify whether the recognition result differentiates based on the type of pathogen.
- In Chapter 4, the authors very briefly point to different results in other similar studies. In this case, it would be appropriate to refer more extensively to the work and results obtained by other researchers.
Author Response
We are very grateful to your comments and thoughtful suggestions.Based on these comments and suggestions, we have made careful modification on the original manuscript.Here are our responses to your comments.
Comments1: The presented work concerns the automatic recognition of diseases appearing on tomato leaves. In the analyzed text, the authors rely on data provided from two datasets. However, the description of these data is brief.
Response 1:Thank you for pointing this out. We are very sorry that we did not provide a clear description of these two datasets. We have revised the text to address your concerns and marked it in red, and hope that it is now clearer. Please see 3.1 in the page 5 of the revised manuscript.
Comments2:In Figure 4, the authors distinguish different pathogenic factors on tomato leaves, but the text does not clarify whether the recognition result differentiates based on the type of pathogen.
Response 2: Thank you very much for your valuable feedback. We are very sorry that we did not specify whether the identification results were differentiated based on the type of pathogen. Based on your suggestion, we have added explanations and differentiation methods for Figure 4 in section 3.1 and highlighted it in red. Thank you for your feedback. I hope this revision will meet your satisfaction.
Comments 3:In Chapter 4, the authors very briefly point to different results in other similar studies. In this case, it would be appropriate to refer more extensively to the work and results obtained by other researchers.
Response 3:We sincerely appreciate your valuable feedback. In Chapter 4, we did not refer enough to the research results of other individuals, so we carefully read and searched for literature, and referenced the work of more people and added more references. We have revised the Chapter 4 to address your concerns and hope that it is now clearer. Please see the Chapter 4 in page 9 and page 10 of the revised manuscript.
Thank you for providing us with useful feedback. We hope this revision will meet your satisfaction.
Reviewer 3 Report
Comments and Suggestions for AuthorsIn this manuscript an Eff-Swin model that integrates the enhanced features of the EfficientNetV2 and Swin-Transformer networks, aiming to harness the local feature extraction capability of CNNs and the global modeling ability of transformers was proposed to extract local features while attending to global information have emerged as a novel research direction. Here some issues which I encourage the authors to consider:
1 – Regarding the results section it was demonstrated that the proposed approach outperforms other classical model but when comparison with other works in the literature it is difficult to see the real performance, all works uses the same dataset? All the works were tested using the same parts of the dataset? Discussion about this could be a nice complement of the current manuscript.
2 - Regarding the processing speed and computational resources usage, there is a lack of discussion about this, in my opinion the interment in accuracy is a little bit low comparing with previous works and if this increment of accuracy affects other issues such as the mention before, the accuracy increment could be irrelevant.
Comments on the Quality of English Language1 – There are some minor grammatical and style errors. I suggest a detailed revision of the English language.
Author Response
We are very grateful to your comments and thoughtful suggestions.Based on these comments and suggestions, we have made careful modification on the original manuscript and marked the changes in red. Here are our responses to your comments.
Comments 1:Regarding the results section it was demonstrated that the proposed approach outperforms other classical model but when comparison with other works in the literature it is difficult to see the real performance, all works uses the same dataset? All the works were tested using the same parts of the dataset? Discussion about this could be a nice complement of the current manuscript.
Response 1:We feel great thanks for your professional review work on our article.We have thoroughly discussed and studied the comparisons with other literature in the Results section, using the same dataset for these comparisons. The research listed in the discussion helps to understand the recent research on tomato pests and diseases, and some classic models compared on the same dataset are included in the results section. The same dataset is used for the accuracy comparison of the models. We apologize for not expressing ourselves clearly enough, and we have already provided a clarification in the text. And based on your feedback, we have provided a more detailed and convincing description of the experimental comparison section, which has been highlighted in red in the text.We acknowledge your comments and constructive suggestions very much, which are valuable in improving the quality of our manuscript.
Comments 2: Regarding the processing speed and computational resources usage, there is a lack of discussion about this, in my opinion the interment in accuracy is a little bit low comparing with previous works and if this increment of accuracy affects other issues such as the mention before, the accuracy increment could be irrelevant.
Response 2: We sincerely appreciate your valuable feedback. We think this is a great suggestion. Based on your suggestion, we have added a discussion on the necessity of improving processing speed, computing resources, and accuracy. And during the accuracy comparison, three types of models were compared simultaneously: CNN class, Transformer class, and their combination class, fully demonstrating the effectiveness of the model in improving accuracy. Based on your suggestions, we have made revisions to the manuscript and highlighted it in red. Thank you for your reminder.
Comments 3: There are some minor grammatical and style errors. I suggest a detailed revision of the English language.
Response 3: Thanks for your suggestion. We have tried our bestto polish the language in the revised manuscript. And we hope the revised manuscript could be acceptable for you.
Thank you for providing us with useful feedback. We hope this revision will meet your satisfaction.
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsAuthors have properly address the reviewers comments.