Lightweight Multiscale CNN Model for Wheat Disease Detection

Round 1

Reviewer 1 Report

This paper proposes a lightweight multiscale CNN model for detecting wheat diseases, which is crucial for improving wheat yield and quality. The paper discusses the difficulties involved in detecting wheat diseases in the field due to the different growth cycles of wheat and the various types of diseases that can develop. The proposed model includes three Inception modules to enhance the depth and width of the network and reduce computational complexity. Additionally, a residual module with two hybrid attention mechanisms is built to focus on disease features and reduce the influence of complex backgrounds in images. Finally, the module structure is modified to constitute a hybrid attention mechanism multi-scale CNN model. The proposed method achieves an average accuracy of 98.7% on the test dataset, outperforming classical convolutional neural networks and classical lightweight networks. The model has strong generalization performance and can efficiently and rapidly identify wheat disease. Overall, this paper provides a promising solution to the challenge of detecting wheat diseases in a complex context.

It seems that the paper presents a solid and interesting approach. However, I believe that there are some minor issues that need to be addressed before publication. Please find them below:

·        -As the paper presents important process, it is required that the discussion section should be presented in a clear and separate section to convince readers that the proposed method is more effective than existing studies.

·        -Please note that your manuscript requires English corrections.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The paper provides an extensive and interesting evaluation of Wheat disease detection using  multi-scale CNN. It would have been useful to also include a secondary recurrent model, or a respective ensemble structure to account for temporal variations in Wheat disease and evaluate how different models compare in terms of efficiency, and time complexity.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

This paper mainly focused on building t Inception-ResNet-CE model for the problem of auto- 105 matic wheat disease identification.

Firstly, we thank the authors for their work

But we have few comments that need to be handled before publication.

1.    The results section needs to be extended by exploring the effect of different optimizers, compare between them and record the notes.

2.    Authors need to draw a clear figure for the proposed framework with clear justification.

3.    Comparison with literature need to be added.

4.    The model performance seems good, but about the generalization ability, please clarify.

Authors need to add more 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 5 Report

 

The authors proposed lightweight multiscale CNN model for detecting wheat diseases in this paper. There are some suggestions that should be incorporated before acceptance as below:

1. Firstly, the article lacks clarity in terms of the research contribution and the novelty of the proposed model. The authors need to provide more details on how their proposed model is different from the existing model and how it contributes to the field of agriculture.
2. Secondly, the methodology section is not well-explained, and the experimental analysis is not thorough enough. The authors need to provide more details on the data preprocessing, feature extraction, and model training process. The article lacks information on the selection of performance metrics, and the statistical analysis is not presented in a clear and concise manner.
3. Thirdly, the article lacks a comprehensive discussion on the limitations and implications of the research findings. The authors need to provide more insights into the practical implications of their research for the field of smart faring and the potential limitations of their proposed model.
4. Finally, the language and structure of the article need improvement. The article is poorly structured, with several sections lacking coherence and flow. The language used is not clear and concise, with several grammatical errors and awkward sentences.
5. The figure 7 is not clear.
6. The abstract and conclusion are not perfect. Rewrite it to be consistent with the objectives and methodology.
7. Put a new section "2. Background" to illustrate the idea's significate concepts, and definitions, and preliminary to include all the definitions and concepts.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 4 Report

This model We propose an inception- ResNet-CE model for the automatic identification of wheat diseases, unless the research has a promising quality, it needs o some enhancements to improve paper quality.

1-      The model trust could not be ensured with the used dataset, so try to test your model with the unseen dataset to ensure model generalizability.

2-      Please compare your work and the literature

3-      Provide the lineation of your work and clarify your future.

4-      Enhance English grammar and solve writing mistakes.

 

5-      Add more references to summarize the state of art.  

Author Response

Please see the attachment

Author Response File: Author Response.docx