A Machine Learning Based Full Duplex System Supporting Multiple Sign Languages for the Deaf and Mute

Round 1

Reviewer 1 Report

The Authors wanted to present the communication system  between sign language speakers and the rest of society they had built. The system presented is very interesting but unfortunately very poorly presented in this manuscript. This work should be completely improved (rebuilt) before it is published.

   1. Section 1: It is unnecessary to divide this section into such short subsections. I think that all these parts could be combined into one whole without losing the quality of this section.

2. Section 2: The Authors have done a thorough literature review of communication systems, datasets and algorithms used in systems for sign language speakers to communicate with other people. Reading this section, one has the impression that it is similar to 'shooting a gun'. Almost every sentence could be bulleted separately. The paragraphs in this section lack an underlying thought.

3. Section 3: The Authors presented a general methodology for conducting scientific research. This section is not needed in this work. The Authors should focus on presenting the methodology of their scientific study, the results of which they present in this manuscript.

4. Section 4: This section should include a more detailed description of the hidden layers, i.e. the types of algorithms that were used in each step and details about the construction of the convolutional neural network.

5. Section 5: Various types of experiments that have been carried out have been described. This description is very laconic. Please provide a more detailed description of these experiments.

6. Section 6: For the number of experiments carried out, the discussion of the results obtained is very poor. At the beginning of this section, conclusions are presented which refer to the literature review. This text should be included in Section 2.

7. Section 7: This chapter should be complemented by the main conclusions of the research carried out.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The authors present a solution for the non-deaf and mute people to converse with D-M people without the need to learn sign language. The idea behind the project is good, but the manuscript needs to be revised to be up to publication standards.

      It needs to be clarified what the differences are between this work and the previous work. 

Saleem MI, Siddiqui A, Noor S, Luque-Nieto MA, Otero P. A Novel Machine Learning Based Two-Way Communication System for Deaf and Mute. Applied Sciences. 2022 Dec 29;13(1):453.

 

Comments on the technical side:

 A. How is overfitting handled?

 B. How does the system avoid detecting gestures on transition? Is this a disadvantage of the system? This issue is vaguely explained in figure 15, but more information is needed.

C. On line 488, if a set of three languages is taken as unique, how can mistranslation be avoided? Is there a language detection stage?

D. On line 489, we expect the software to adjust to the user, not otherwise.

E. On line 509, Is the dataset available online? What about references 46-48?

F. References 41 and 44 are probably suitable for scientific communications.

G. In the conclusion, there is mention of a few modules: image to text recognition modules, etc., but evidence needs to be shown. 

 

The flaws in the writing style are too many to list. Next a summary is listed:

1. The first sentence of the abstract must be omitted: "The use of Machine Learning (ML) is increasing, being used extensively to solve various problems."
2. The authors mention that this work is a continuation of previously published work. Nevertheless, their work needs to be appropriately cited. 
3. The text is full of cliche phrases that are not proper for scientific communications, for example: "The advancement in technology has mostly created a positive impact on society.", "Artificial intelligence (AI) has created a ground-breaking impact on the society by improving the quality of life."
4. On top of the previous comments, some assertions are inaccurate: "The majority of people around the world speak the Chinese language." The world has 8 billion people, and 1 billion speak Mandarin. That is not the majority.
5. Some of the figures are cartoons and informal and do not contribute to a better understanding of the document, for example, figure 1 (right side), 3, and 4.
6. Section 3 is not relevant.
7. The authors do not distinguish between discussion and conclusion. The information is repetitive.
8. Figure 2 can be synthesized.
9. Table 3 can be split. Also, information is needed on the size of the test and train sets.
10. Figures 24 and 25 cannot be read. The colors and font size could be more helpful.

 

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The study is interesting and results are relevant for the journal's aims and scope.

The paper content is unusual for a research article. The writing style and paper content seems more like a presentation of project results then original research. The introduction should be more focused including only enough information to explain the topic and introduce the aims. I think the article can be considered for publication after major text editing according to the typical research article writing standards.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

I am happy to see that the Authors have taken the comments of the Reviewers into account and has significantly improved the manuscript during the revision process. Therefore, I am happy to recommend the manuscript for publication.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The authors have improved the quality of the manuscript.   Despite that, the manuscript is not up to international publication standards.

Professional edition services would be highly beneficial. Training in scientific writing would also be helpful. Not going deep into this issue, the introduction has only one reference. That is highly unusual in a scientific paper.

For example, the introduction mentions that the work is a continuation of previous work. Nevertheless, the authors need to tell the reader how this manuscript differs from the previous one. The first reference needs to be completed; the information about the volume and issue needs to be included.

There is no reason for having a section 1.1 if there is no section 1.2. Please incorporate the content of such a section at the end of the introduction.

The way the literature review is written needs to be better.

Sections 3.1 and 3.2 are irrelevant.

Figures 1, 3, and 4 do not belong to a scientific paper. The contents of figure 1 are not described in the text.

Figures 24 and 25 continue to be low quality. A suggestion is to use a white background and highlight the diagonal using boldface.

The legends in figure 11 cannot be read.

Table 3 has some issues. The headers are too close to each other, which is confusing. Furthermore, the information is not presented properly. For example, it is clear that the columns refer to percentages (it is specified on the header), so there is no reason to include the % sign.

The manuscript mentions that the language is detected in step 4. However, there needs to be an explanation of how this is done. Please clarify this to the readers.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Authors have introduced changes however they are not substantial enough to ensure the quality for publication. The work is unfocused, writing style is unusual. Manuscript is too lengthy. The figures (for example 1 and 3) do not add value to the presentation or clarification of the content. Article is a mix or review and research study which is not necessary and makes reading difficult.

Author Response

Please see the attachment

Author Response File: Author Response.docx