Practical Study of Recurrent Neural Networks for Efficient Real-Time Drone Sound Detection: A Review

Round 1

Reviewer 1 Report

Following are the main observations:- 

The prediction metrics for the simple RNN achieves approximately 100% results as compared to the GRU 64 cell, what is the computational cost? 

There is a requirement to present the complexity of the all the suggested algorithm with their mathematical proof. 

The selection of the performance metrics needs to be explained? Why only these metrics have been considered? 

In figure 11 and 12 of the manuscript, there are two lines in each plot, please elaborate the difference between the dotted line and line? 

Why the accuracy of the CNN is less as compared to the other methods? 

There is a need of comprehensive comparison with the existing techniques? please add in the revised version. 

The contribution and organization of the article is missing, please add. 

Need to add more literature review ? 

Author Response

Thanks for the reviews and the responses have been attached below as a pdf file. Other additional modifications have been provided as a result of other reviews.

Author Response File: Author Response.pdf

Reviewer 2 Report

Section 1 must be improved.

-       Authors should emphasize contribution and novelty, the introduction needs to clarify the motivation, challenges, contribution, objectives, and significance/implication. 

-       You should introduce the problem in more detail so that the reader is immediately clear about the purpose of your study.

-       Specify better the essential elements of the problem. You should add more information in the introductory part, you should add other works that have also addressed the problem.

-       Add references to works that have already dealt with the topic (UAV detection), for example:” Acoustical unmanned aerial vehicle detection in indoor scenarios using logistic regression model”, “Audio based drone detection and identification using deep learning”,” Development of an acoustic system for UAV detection”

-       You must properly introduce your work, specify well what were the goals you set yourself and how you approached the problem.

-       At the end of the section, add an outline of the rest of the paper, in this way the reader will be introduced to the content of the following sections.

 

Section 3 must be improved.

-       Introduce adequately the RNN technology: Explain how do Recurrent vs. Feed-Forward Neural Networks Work. In this way, it will be possible for the non-expert reader to understand the potential of RNNs.

-       explains in detail the role of each layer

-       “vanishing and exploding gradient problems” Introduce adequately the topic

-       You must properly introduce the equation, list in detail the variables contained in it with a concise description of the meaning. To make them more readable show them in a bulleted list. In this way the reader will be able to understand the contribution of each variable.

-       Try to enrich the captions of the figures, the reader should be able to read the figure without the need to retrieve the information in the paper. Try to summarize the essential parts of the Figure and what you want to explain with it. For example for Figure 1,2,3,4.

-        

Section 4 must be improved.

-       The section relating to the methodologies based on Machine Learning must be enriched. You must summarize the essential characteristics of the methods you have used and justify your choices. Try to summarize what are the strengths and weaknesses of the methods, in this way you can make the reader understand why you have chosen these methodologies.

-       Describe in detail the equipment used to make the sound recording. Extract this data from the datasheet of the instrumentation manufacturer. To make reading the specifications of the instruments more immediate, you can insert them in a table, listing the instruments used and the specific characteristics for each.

-       Introduce adequately the melspectogram

-       Furthermore, a description of the hardware and software used for data processing is completely missing. Describe in detail the hardware used:  Extract this data from the datasheet of the hardware manufacturer. To make reading the specifications of the hardware more immediate, you can insert them in a table, listing the instruments used and the specific characteristics for each.

-       Also, you should describe in detail the software platform you used.

-       Also describe the machine learning-based libraries you used.

-       I could not find a detailed description of the evaluation metrics you have adopted. How will you measure your model's performance? This section is essential in order to demonstrate the effectiveness of your methodology. Furthermore, only by adopting adequate metrics will it be possible to compare your results with those obtained by other researchers.

 

Section 5 must be improved.

-       A detailed discussion of the results obtained is missing. Try to summarize what was obtained and try to extract useful information from the work carried out. Also add bibliographic references to support your conclusions, to give more weight to your statements.

 

Section 6 must be improved.

-        Paragraphs are missing where the possible practical applications of the results of this study are reported. What these results can serve the people, it is necessary to insert possible uses of this study that justify their publication.

 

73) Add subsection (ML methods), I have seen that you often use this format, so I will not repeat this advice again, it also applies to the other occurrences.

73) “ML” Do not use acronyms until you have presented the full definition, I will not repeat this advice again, it also applies to the other occurrences.

131) Explain what is intended for unfolded form

209)” According to previous studies” What previous studies?

224-225) Explain what you mean with: "Unloaded","Loaded" and "Background

313-316) Feature extraction?

319) In Figure 7 the axis labels are missing. Add label to subplot

327) In Figure 8 add label to subplot, I will not repeat this advice again, it also applies to the other occurrences.

 510) Add authors contribution

514) Remove Appendix A section

Author Response

All answers for comments were uploaded via the pdf file below.

P/S: The meaning of "Loaded and Unloaded" is modified in Data Preparation Part. 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors responded almost all the raised questions. 

 

Reviewer 2 Report

The authors addressed the reviewer's comments with attention and modified the paper with the suggestions provided. The new version of the paper has improved both in the presentation and in the contents .