Review Reports - MobGSim-YOLO: Mobile Device Terminal-Based Crack Hole Detection Model for Aero-Engine Blades

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper proposes a new method for crack detection using deep learning. It introduces several advanced techniques to improve detection accuracy and efficiency. The main innovations include the use of K-means++ to optimize anchor points, the replacement of the YOLOv5s backbone with MobileNetv3 for lightweight design, and the adoption of the SimAM attention mechanism for improved target focusing. Experiments and data analysis show that the improved model significantly reduces model parameters while maintaining high accuracy. The article is innovative and interesting, however, there are some issues that need to be fixed.

1. Some of the pictures are not clear enough and the content of the pictures lacks the necessary description.

2. Pictures from other sources need to be cited and permissions should be obtained.

3. Please improve the quality of the language. Some of the methods used are not very appropriate and some statements are too long and should be streamlined appropriately. There are some writing errors in the text, e.g. line 122.

4. The history of the YOLOv5x in part 2 is too long, it could be streamlined or put into the introduction.

5. The second part of the article should be a sub-section with a separate section on improvements to the YOLOv5s.

6. Abbreviations should be given in full the first time they appear, e.g., FPS, NDT.

7. The authors need to provide comparisons with other methods or models that have been formally established.

8. References are not complete with essential information missing. Moreover, the literature review is inadequate.

9. Can the authors propose any limitations of the current detection method in real applications?

Author Response

Comments 1 : Some of the pictures are not clear enough and the content of the pictures lacks the necessary description.

Response 1:Thank you for pointing this out. We agree with this comment. But I don't know which image is not clear enough, I enlarged all of them. To add to the description, I described in detail the structural diagram of the improved YOLOv5s; added the introduction of traditional convolution and depth-separable convolution; and analyzed and described the training results and test results of the MobGSim-YOLO model.

Comments 2 : Pictures from other sources need to be cited and permissions should be obtained.

Response 2: Thank you for pointing this out. We agree with this comment. I have added pictures of blade cracking, corrosion, tearing and high temperature corrosion in the introduction section and a graph of the test results of the YOLOv5s model at the end of the paper.

Comments 3：Please improve the quality of the language. Some of the methods used are not very appropriate and some statements are too long and should be streamlined appropriately. There are some writing errors in the text, e.g. line 122.

Response 3：Thank you for pointing this out. We agree with this comment. I have corrected the problems in lines 122, 26, 329, and 333, simplifying some sentences accordingly. I am a graduate student in school, the pressure of the end of semester exams, really do not have enough time to modify the paper, if there is still a language problem is willing to re-return to repair, thank you teacher!

Comments 4：The history of the YOLOv5x in part 2 is too long, it could be streamlined or put into the introduction.

Response 4：Thank you for pointing this out. We agree with this comment. I've simplified this part of the history and moved it to the introductory section.

Comments 5：The second part of the article should be a sub-section with a separate section on improvements to the YOLOv5s.

Response 5：Thank you for pointing this out. We agree with this comment. I have placed the description of the algorithm improvement part at the beginning of the second part.

Comments 6：Abbreviations should be given in full the first time they appear, e.g., FPS, ND

Response 6：Thank you for pointing this out. We agree with this comment. Has been revised as requested, please see line 18 and 34 for details, FPS is known as Frames Per Second; NDT is known as Non-Destructive Testing.

Comments 7：The authors need to provide comparisons with other methods or models that have been formally established.

Response 7：Thank you for pointing this out. We agree with this comment. The training results section introduces results from YOLOv3, YOLOv4, and YOLOv7; the testing section performs a comparison between YOLOv5s and MobGSim-YOLO.

Comments 8：References are not complete with essential information missing. Moreover, the literature review is inadequate.

Response 8：Thank you for pointing this out. We agree with this comment. I have made changes to the reference format according to the EndNote format provided on the official website, and I have re-improved the literature review.

Comments 9：Can the authors propose any limitations of the current detection method in real applications?

Response 9：Thank you for pointing this out. We agree with this comment. Current detection methods are unable to weigh the relationship between detection speed, detection accuracy and model size, and can only be embedded into mobile device terminals when all three properties meet the requirements at the same time.

The bolded portions of the text are revised.I have limited ability, if there is a modification is not in place please guide the teacher.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The title of manuscript is very interesting and also the presented method is applicable and practical in industrial, especially for inspection in a dark environment. In addition, the novelty was very clear. It is strongly suggested to publish in the Journal after minor revision as follows:

1- There are some typo, which should be removed. For example, in line 26, the phrase "rotor" is duplicated. In line 329, the phrase "labeling" is duplicated. In line 333, the phrase "labelimg" should change to "labeling".

2- List of abbreviations should be added into the manuscript.

3- Regarding Table 1, what is the meaning of w and h parameters?

4- The proposed methodology should explain more in details.

Comments on the Quality of English Language

The writing language is fluent, but there are some typo that should be corrected.

Author Response

Comments 1 : There are some typo, which should be removed. For example, in line 26, the phrase "rotor" is duplicated. In line 329, the phrase "labeling" is duplicated. In line 333, the phrase "labelimg" should change to "labeling".

Response 1: Thank you for pointing this out. We agree with this comment. I have corrected the problems in lines 122, 26, 329, and 333, simplifying some sentences accordingly. I am a graduate student in school, the pressure of the end of semester exams, really do not have enough time to modify the paper, if there is still a language problem is willing to re-return to repair, thank you teacher!

Comments 2 : List of abbreviations should be added into the manuscript.

Response 2: Thank you for pointing this out. We agree with this comment. Has been revised as requested, please see line 18 and 34 for details, FPS is known as Frames Per Second; NDT is known as Non-Destructive Testing.

Comments 3：Regarding Table 1, what is the meaning of w and h parameters?

Response 3：Thank you for pointing this out. We agree with this comment. w denotes the length of the anchor frame, h denotes the width of the anchor frame

Comments 4：The proposed methodology should explain more in details.

Response 4：Thank you for pointing this out. We agree with this comment. I've simplified this part of the history and moved it to the introductory section. But I don't know where to change this part. Can the teacher guide me?

The bolded portions of the text are revised. I have limited ability, if there is a modification is not in place please guide the teacher.