SAR Image Aircraft Target Recognition Based on Improved YOLOv5
Round 1
Reviewer 1 Report
This paper presents an improvement of YOLOv5 for SAR image aircraft target recognition. It is done by proposing a multiscale feature adaptive fusion module to adaptively assign different weights to each scale of the feature layers. This is for extracting richer semantic and textural information. The improvement was demonstrated through an experiment. The results show the increasing mAP of 5.8% than the original YOLOv5.
The presenting work on this paper is very interesting and supported by proper experiment result. However, the description on the Section 3 Material and Method is quite hard to be followed as some notations in the figures and equations are not explained. The authors need to pay attention on it to accommodate readers who are interested in the topic but not familiar in this fields.
It is good enough.
Author Response
Thank you very much for your comments, Based on your suggestion, I have made the following optimizations:
- Page 4 Line 194~203: Optimized the description of this paragraph
- Page 4 Line 219~223: Optimized the description of this paragraph
- Page 4 Line 229: Optimized the formula (6), The parameter in the formula is the same with Figure 1
- Page 8 Line 279: Optimized the description of Ghost-Conv
- Page 9 Line 318~320: Optimized the description of this paragraph (Page 8 Line 306: Add 2 names in Figure 3, It is easier to understand the description of this paragraph)
- Page 10 Line 355, Modify the detector head names in Figure 5, all the names are consistent in this paper.
I already upload the latest manuscript, Please see the attachment. In addition to these modifications, there may still be some other issues. If you have any further questions, please let me know. Thank you again for your suggestion, Have a nice day!
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Development of high-speed algorithms for determining aircraft types and location from synthetic aperture radar data is an urgent task. Therefore, the urgency of the manuscript topic is beyond doubt. The authors describe the essence and results of their study in sufficient detail. I believe that the manuscript corresponds to the subject of the Appl. Sci. journal and can be published after minor revision.
- While reading the manuscript, I had a question: Is this study purely research in nature or is it aimed at a practical result? If the study is purely fundamental research, I have no comments. If the study is aimed at a practical result, then I’d like the authors to supplement the manuscript with material concerning the practical significance of the results obtained. It is stated in the manuscript that the mean accuracy (mAP) of the improved YOLOv5 increased by 5.8% as compared to the original YOLOv5. I’d like the authors to give a condition under which the mean accuracy satisfies practical requirements. If the authors are not ready to answer this question, then I’d recommend them in future publications to be sure to investigate the practical aspects of the accuracy of high-speed algorithms.
- I‘d recommend that the authors read the manuscript more carefully and correct mistakes. For example,
Page 1 Line 40. The authors refer to Hu et al. [1]. The actual reference should be to Fu et al. [1].
Page 2 Line 91. Incorrect aircraft type. Correct A320/321.
Page 10. Lots of empty space.
In Figure 4, the yellow box is barely noticeable and perceived as a vertical line, which makes it difficult to analyze the picture.
Author Response
Point 1: While reading the manuscript, I had a question: Is this study purely research in nature or is it aimed at a practical result? If the study is purely fundamental research, I have no comments. If the study is aimed at a practical result, then I’d like the authors to supplement the manuscript with material concerning the practical significance of the results obtained. It is stated in the manuscript that the mean accuracy (mAP) of the improved YOLOv5 increased by 5.8% as compared to the original YOLOv5. I’d like the authors to give a condition under which the mean accuracy satisfies practical requirements. If the authors are not ready to answer this question, then I’d recommend them in future publications to be sure to investigate the practical aspects of the accuracy of high-speed algorithms.
Response 1: Please provide your response for Point 1. (in red)
Thank you, teacher, for patiently and meticulously providing guidance. Currently, I am conducting purely research in nature. Your suggestions have pointed the way for my future research. I will focus on the accuracy of high-speed algorithms in practical applications in future research. Sincerely thank you.
Point 2: I‘d recommend that the authors read the manuscript more carefully and correct mistakes. For example,
Page 1 Line 40. The authors refer to Hu et al. [1]. The actual reference should be to Fu et al. [1].
Page 2 Line 91. Incorrect aircraft type. Correct A320/321.
Page 10. Lots of empty space.
In Figure 4, the yellow box is barely noticeable and perceived as a vertical line, which makes it difficult to analyze the picture.
Response 2: Please provide your response for Point 2. (in red)
I have made revisions to all the issues you pointed out, and then I carefully read the manuscript and made some improvements. I already upload the latest manuscript, Please see the attachment. In addition to these modifications, there may still be some other issues. If you have any further questions, please let me know. Thank you again for your suggestion, Have a nice day!
Author Response File: Author Response.pdf
