A Lightweight Crop Pest Detection Method Based on Convolutional Neural Networks

Round 1

Reviewer 1 Report

1. The Equation (8) shows an equation to get an average over number W but the range of i is incorrect. What I understand is that the range of i should go from 0 to W-1.

2. The Equation (8) shows an equation to get an average over number H but the range of j is incorrect. What I understand is that the range of j should go from 0 to H-1.

3. If the computation are according to the equation (8) and (9) then the author is required to recompute the results based the corrected versions of (8) and (9).

4. Referring to Table 3, the difference between Ave_IoU values of K-means (71.01) and K-means++ (71.67) is not seem significant. (71.67-71-01)/71.67 x 100 =  0.92%. The author is suggested to improve this values if possible or otherwise mention the reason/arguments in favor of no more improvement.

5. The author have chosen different sets of coordinates for K-means and K-means++ in Table 3. Is it possible to provide same sets of coordinates for both methods? Or otherwise mention clearly if not possible based technical backgrounds of the two methods used (if any).

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The author's purpose "A Lightweight Crop Pest Detection Method Based on Convolutional Neural Networks". The contributions of this study are good and clear. Overall, the paper is an interesting one with reasonable solutions and promising experimental results. However, there are quite a few issues to be resolved before being considered for publications

- The abstract needs to present the context of achieving the main results of 82.9 % and also discuss the core findings. I understand that the abstract is quite long; however, it could be restructured in order to present numeric results and findings.

- The introduction needs to explain more clearly the aims of the paper, and the exact way that it could be assistive to the related research community. In addition, I feel that it is important to highlight the importance of the four contributions of the work. Also please explain how the present authors' work in could be assistive in the formulation of systems and tools. Please point out the novelty of the paper too.

_ In the introduction part explain the Yolo model much related to your work  “Abdullah et al , “Real-Time Hand Gesture Recognition Based on Deep Learning YOLOv3 Model. Applied Sciences”.  https://doi.org/10.3390/app11094164” and other segmentation models like U-net for this are included in the introduction https://doi.org/10.3390/s22041552

- The paper also needs to present additional related works and methods. In addition, it needs to explain in better detail for each paper, the methodology of it as well as the findings. Please explain how it builds on and extends previously conducted studies in the literature.

- Section 2 needs to present in better detail core aspects of the author's work. The characteristics of the dataset need to be presented as well as the exact training procedure of the paper.

-The actual architecture need to be presented in better detail and the layers need to be substantiated.

-Section 3 needs to present in better detail the aims of the experimental study and the results.

 

- The results need a deeper and more complete discussion and also the possible connection with existing works and results in the literature.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors done all corrections which i mentioned