Employing FGP-3D, a Fully Gated and Anchored Methodology, to Identify Skeleton-Based Action Recognition

Round 1

Reviewer 1 Report

In this paper, the authors develop a novel generalizable skeleton-based action recognition technique named as FGP-3D. Compared to state-of-the-art techniques, the proposed method show much better predictive performance by some real data applications. The paper is well written and well organized. I have a few questions as follows.

 

Question 1: In section 4, the real data application section, is the accuracy are predictive accuracy in Table 1 to 4? If yes, could the author clarify more about how to define the training and testing data (within UTD-MHAD dataset, MSR dataset, KARD dataset, SBU dataset respectively)? Is it by cross validation? If it is by cross validation, could the author provide how many replicates they do and the 95\% confidence interval for the accuracy? Is it possible to provide the accuracy for both the training and testing data?

 

Question 2: In introduction, could the author provide some additional illustration about how to link the action recognition technique to clinical outcome in some disease area? It would help to highlight the importance of human action recognition as it may help disease diagnosis.

Author Response

Thank you very much for your email about our manuscript (applsci-2310896). We would like to thank you the reviewer for your careful review and helpful comments. Your comments and suggestions are undoubtedly essential to help us improve the quality of this manuscript. The manuscript has been revised carefully according to the Reviewer's opinions, and corrections have significantly improved the manuscript. We hope the revised manuscript will meet your objections and requirements.

Author Response File: Author Response.docx

Reviewer 2 Report

The authors try to propose a unified skeleton-based framework for action recognition called FGP-3D.

The authors fail in the definition of the model and in the computational mathematical approach to the calculation of the parameters.

The definition of several of the model parameters is not clear and the reader must guess and assume several points in the proposed methodology.

 Specific comments:

Page 4, Eq1 and Eq.2.- What the parameters gamma_1 and gamma_2 refer to are not defined. It is advisable to define and show in some of Figure 1 that represent the parameters to be defined, such as the acromioclavicular joints and the anchored points of the model, among others.

Page 6, Line 241 and 242.- “For now remember the formula for ?_?.” No equation for z_t is found within the paper. Nor is it clear what use is given to this parameter within the model. All punctuation marks are missing at the end of all equations.

Page 7, Line 294.- “Fig. 3 depicts the finished model architecture” The reference should be to Figure 2 and not to Figure 2. The authors did not take the slightest care to do a minimal review of the figures that are cited in the paper.

Page 7, Eq.18.- The mathematical or computational definition of this equation is not clear. The way to perform this operation must be explained within the text and the parameters involved must be defined [ ℎ_? ℎ_? ℎ_? ?_??]

 

Author Response

Thank you very much for your email about our manuscript (applsci-2310896). We would like to thank you the reviewer for your careful review and helpful comments. Your comments and suggestions are undoubtedly essential to help us improve the quality of this manuscript. The manuscript has been revised carefully according to the Reviewer's opinions, and corrections have significantly improved the manuscript. We hope the revised manuscript will meet your objections and requirements.

Author Response File: Author Response.docx

Reviewer 3 Report

This manuscript has a detailed introduction on the background and significance of skeleton-based action recognition. However, the paper organization and presentation, especially Section 3 and 4 MUST BE IMPROVED for better readability. These two sections are relatively long, having a few repetition, with low explicity. The decision is rejection, as the authors need to polish the presentation for next submission.

Comments for author File: Comments.pdf

Author Response

Thank you very much for your email about our manuscript (applsci-2310896). We would like to thank you the reviewer for your careful review and helpful comments. Your comments and suggestions are undoubtedly essential to help us improve the quality of this manuscript. The manuscript has been revised carefully according to the Reviewer's opinions, and corrections have significantly improved the manuscript. We hope the revised manuscript will meet your objections and requirements.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

 

The authors have considered the observations made in the review in a general way, although some considerations to be improved are still presented.

Specific comments:

Page 6, lines 264-266.- The variables must be z,r,h,W,U, b,i,j,k and others must be defined using italic font.

Page 8, line 383.- Section names should not appear as the last line of a page.

Page 9, line 420, Table 2.- The tables should not appear truncated but complete on the page.

 

 

 

Author Response

Dear Reviewer of Journal, Applied Sciences:

  Thank you very much for your email about our manuscript (applsci-2310896). We would like to thank you the reviewer for your careful review and helpful comments. Your comments and suggestions are undoubtedly essential to help us improve the quality of this manuscript. The manuscript has been re-revised carefully according to the Reviewer's opinions, and corrections have significantly improved the manuscript.

  We hope the re-revised manuscript will meet the Reviewer’s objections and requirements and be considered for publication in Applied Sciences.

  Thank you and the reviewer again for your arduous work on our manuscript.

Author Response File: Author Response.docx

Reviewer 3 Report

Mostly, the authors have revised their manuscript per the reviewer's suggestion. In particular, the presentation of methodology is re-organized in a clear way, an important improvement over the old version. Some of the languages are polished. Overall, it can be accepted for this form.

Author Response

Dear Reviewer of Journal, Applied Sciences:

  Thank you very much for your email about our manuscript (applsci-2310896). We would like to thank you the reviewer for your careful review and helpful comments. Your comments and suggestions are undoubtedly essential to help us improve the quality of this manuscript. The manuscript has been re-revised carefully according to the Reviewer's opinions, and corrections have significantly improved the manuscript.

  We hope the re-revised manuscript will meet the Reviewer’s objections and requirements and be considered for publication in Applied Sciences.

  Thank you and the reviewer again for your arduous work on our manuscript.

Author Response File: Author Response.docx