Next Article in Journal
On the Integrity of Large-Scale Direct-Drive Wind Turbine Electrical Generator Structures: An Integrated Design Methodology for Optimisation, Considering Thermal Loads and Novel Techniques
Previous Article in Journal
Anti-Offset Multicoil Underwater Wireless Power Transfer Based on a BP Neural Network
Previous Article in Special Issue
Improved Support Vector Machine for Voiceprint Diagnosis of Typical Faults in Power Transformers
 
 
Article
Peer-Review Record

A Real-Time Dual-Task Defect Segmentation Network for Grinding Wheels with Coordinate Attentioned-ASP and Masked Autoencoder

Machines 2024, 12(4), 276; https://doi.org/10.3390/machines12040276
by Yifan Li 1,2, Chuanbao Li 1,2, Ping Zhang 1,2 and Han Wang 1,2,*
Reviewer 1:
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Machines 2024, 12(4), 276; https://doi.org/10.3390/machines12040276
Submission received: 23 March 2024 / Revised: 18 April 2024 / Accepted: 18 April 2024 / Published: 21 April 2024
(This article belongs to the Special Issue Application of Deep Learning in Fault Diagnosis)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This work dealt with a novel light-weight CA-ASP module to the DeeplabV3+ network based on the discovery of similarities between CA and ASPP, which outperforms the original ASPP baseline with less calculation. The results obtained appear reasonable and can attract numerous readers. However, some suggestions and revisions to improve the quality of the manuscript should be considered before acceptance:

1) Graphical abstract and highlights should be added to the manuscript.
2) The manuscript should be revised from an English language perspective.
3) Add in the abstract the most quantitatively significant results obtained. It seems that the abstract is not well presented.
4) The main novelty of this research should be added in the last part of the introduction. Additionally, how does this research address the main question?
5) How does it expand the subject area compared to other published materials?
6) What specific improvements should the authors consider in terms of methodology? What other controls should be considered?
7) Some articles should be added to the introduction and/or results. It seems that the following papers may be useful to complete the introduction section:
https://doi.org/10.1016/j.jclepro.2023.139341
https://doi.org/10.1016/j.aei.2023.102254
https://doi.org/10.3390/su16020833
https://doi.org/10.1016/j.ymssp.2023.110826
https://doi.org/10.1016/j.triboint.2024.109502
https://doi.org/10.1016/j.ins.2022.11.035
https://doi.org/10.1016/j.vehcom.2023.100725
https://doi.org/10.1007/s11265-023-01908-1
https://doi.org/10.1016/j.engappai.2023.107458
8) Results and discussion should be separated.

Comments on the Quality of English Language

Moderate editing of English language is required.

Author Response

Dear Reviewer:

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The research proposed a series of improvement strategies targeted at high computational efficiency based on the DeeplabV3+ network by analyzing patterns in the Grinding Wheel datasets. Using a simple but efficient Coordinate Attentioned - Atrous Spatial Pyramid (CA-ASP) module, which has achieved higher segmentation accuracy compared to ASPP with a smaller size of parameters to aim at boosting segmentation performance of the DeeplabV3+ without additional calculation growth on the dual-side defects semantic segmentation tasks of grinding wheels. This research has certain application value and research value, but there are some small problems.

1.Figure 19 shows that the accuracy of MobileNetV3_large+DeeplabV3+is better than that of this paper. What is the research significance and innovation of this method?

2. How does this article address the natural phenomenon of data imbalance in defect datasets?

3. How is the recognition error between the feature image in Figure 5 and the original image determined?

4. The theoretical depth of this article is insufficient.

5. How does this article demonstrate the stability of the model?

6. Some clarification of the new technical contributions of the proposed method should be described in the manuscript.

Author Response

Dear Reviewer:

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The paper is solid and interesting. A few things should be improved to have an excellent paper:

1. If we compare Table 5 and 6 implications are interesting. You could address this issue in more details and try to explain this results (add to the text of the manuscript). 

2. Table 5: It seems that some SOTAs are better than the proposed according to some criteria. You should use multicriteria optimization to provide realistic comparison what is the better algorithm. Than, from the total ranking, you should conclude about the choice of the best algorithm.

Comments on the Quality of English Language

1. Figure 2: check if the term "Reverse side" is appropriate for this context. 

2. Missing spaces between  numbers in many cases, such as [34,37,38], which should be [34, 37, 38].

Author Response

Dear Reviewer:

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

 

The revised manuscript reflects the reviewers' comments well. Therefore, the manuscript can be accepted.

Comments on the Quality of English Language

Minor editing of English language is required.

Author Response

Dear Reviewer:

Thank you very much for your affirmative comments on our work and the previous kind opinions. 

Best regards.

Reviewer 3 Report

Comments and Suggestions for Authors

Scientific contributions are weakly explained. Application contributions are fine.  You should shorten the Introduction, but add some improved expanation of scientific contributions.

Author Response

Dear Reviewer:

Please see the attachment.

Author Response File: Author Response.docx

Back to TopTop