Driver Identification and Detection of Drowsiness while Driving

Round 1

Reviewer 1 Report (New Reviewer)

Comments and Suggestions for Authors

Summary:
In the manuscript the authors present  a cutting-edge approach that combines facial recognition and drowsiness detection technologies with Internet of Things capabilities, including 5G/6G connectivity,  to enhance vehicle security and driver safety. The authors explored various machine learning techniques, with a specific emphasis on video-based identification and analysis for robust drowsiness detection. Finally, the study highlights the potential of these innovations to revolutionize automotive security and accident prevention within the context of Intelligent Transport Systems.

The topic of this manuscript is very interesting and addresses a current topic, however I consider it to have limited contribution in its current state. The manuscript has good structure and organization but requires some actions that must be addressed to improve the work - See comments to the Authors.

Comments and Suggestions for Authors:

- Even if the reader knows the subject, the authors should expand the first occurrence of any the acronyms or initialism, some of them are not expanded as an example GSM, SMS, GPS, etc. 
Even though a list of abbreviations is presented at the end, it is better for the reader to have the acronym expanded at the time of reading.

- References to standards, de facto standard, services, tools, equipment and components must be included, as example Arduino, Python, OpenCV, Docker, Jetson Nano, etc.

-  The authors must include and describe a data flow diagram that describes the end-to-end process. The Client-server scheme in figure 2 is not sufficient to describe the entire process.

- The authors should list the limitations of the system as well as propose alternative solutions or improvements. For example, it is typical that a driver, if he drives on the road with the sun in front of him, wears sunglasses or driving at night.  Any solution or comment about it?

- In section 3. It is understood that for image collection and training, the images are sent to a server in this case using a 5G network. However, once the model is trained, the Jetson Nano is enough to carry out the work autonomously and locally in the car. The authors should include some justification for considering an online system.

- Authors must present response times for detection both to determine the authorized person and to detect an drowsiness event.

- In the manuscript it is mentioned that one of the software tools used is OpenCV Python version. Due to the type of system and application, authors should use the C++ version of OpenCV instead of the Python version.

- The Results and Conclusion sections must be improved by focusing and highlighting the results from a quantitative and not a qualitative point of view; without forgetting that the main topic is the driver identification and detection of drowsiness while driving.

 

 

Author Response

1. The initial occurrences of all acronyms or initialisms have been expanded throughout the document.
2. References to standards, services, tools, equipment, and components have been added throughout the document.
3. The entire process has been more thoroughly explained, particularly in sections 3.3. and 4. Additionally, an extended explanation has been provided in sections 5.1.2., 5.1.3, and 6.
4. System limitations have been elucidated specifically in sections 3.3., 6., and 7.
5. A more detailed rationale for utilizing the server for the conducted analysis is found in sections 3.3, 6., and 7.
6. Section 3.3. presents response times for both authorized person determination and drowsiness event detection.
7. Enhancements have been made to the results and conclusions sections (6. and 7.).
8. English quality throughout the document has been improved.

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

Authors propose a system that focuses on facial recognition to authenticate a driver and  employs continuous eye monitoring features to detect signs of drowsiness.

From my point of view, the work presents several methodological weaknesses:

- It is not clear if the contribution is about real-time implementation or identification strategies using ML.

- What the authors propose is to test different ML models for detection, however there are no clear results on that point. Furthermore, standard metrics are not used to measure the quality of detection.

The contribution with respect to other works in the literature is not clear. The authors simply say that "None of the aforementioned works, while related to the subject matter, aligns entirely with the approach taken in this paper." Because? Justify

The contribution of the work or the originality is not clear.

Perhaps the scope of this magazine is not appropriate.

The wording of the abstract presents inconsistencies:

the sentence “Various machine learning techniques, with a specific emphasis on video-based identification and analysis, are explored for robust productivity detection” and the sentence

 

“The exploration extends to a comprehensive examination of various machine learning techniques dedicated to drowsiness detection, with a pronounced emphasis on video-based identification and analysis methodologies.” They are very similar and do not provide clarity to the abstract. Consider to change that.

 

 

 

 

Comments on the Quality of English Language

there ae some typos and I recommend to review the grammer and the style.

Author Response

1. The contribution of real-time implementation has been clarified throughout the document by addressing inconsistencies found in different sections. Specifically, in the abstract, phrases have been removed to explain it more clearly and accurately.

2. Results have been added in section 6. and response times in section 3.3. to address the measurement times of the two phases, as well as to justify the use of the server.

3. The originality/contribution has been clarified and enhanced in section 2. specifically and throughout sections 3.3., 6., and 7.

Reviewer 3 Report (New Reviewer)

Comments and Suggestions for Authors

This paper presents a two-phase project that integrates facial recognition and drowsiness detection with IoT and advanced connectivity technologies like 5G/6G to improve vehicle security and driver safety. The first phase uses facial recognition for driver authentication, while the second phase monitors for drowsiness with real-time eye tracking and machine learning. However, there are a number of problems with this paper suggesting changes to improve it：

1.      The numbering of the introduction starts from 1 and it is recommended that it not start from 0.

2.      It is recommended that some mathematical expressions be added to the proposed framework to make the proposed method more rational and logical.

 

3.      It is recommended that a number of comparative tests be added to illustrate the performance improvement of the proposed method over other similar methods.

Comments on the Quality of English Language

There exist some minor grammatical errors which should be corrected through careful proofreading.

Author Response

1. The numbering of the introduction section has been modified to start at 1.
2. A more detailed justification for the use of the server for the analysis conducted can be found in sections 3.3, 6., and 7.
3. Section 3.3. provides response times for both determining the authorized person and detecting a drowsiness event.
4. Improvements have been made to sections 6. and 7. concerning results and conclusions.
5. The English quality throughout the document has been enhanced.

Round 2

Reviewer 1 Report (New Reviewer)

Comments and Suggestions for Authors

Summary:
In the manuscript the authors present  a cutting-edge approach that combines facial recognition and drowsiness detection technologies with Internet of Things capabilities, including 5G/6G connectivity,  to enhance vehicle security and driver safety. The authors explored various machine learning techniques, with a specific emphasis on video-based identification and analysis for robust drowsiness detection. Finally, the study highlights the potential of these innovations to revolutionize automotive security and accident prevention within the context of Intelligent Transport Systems.

The topic of this manuscript is very interesting and addresses a current topic and it has good structure and organization and the authors have addressed the reviewer's comments and recommendations.

Author Response

Thank you for reviewing our manuscript. We appreciate your time and effort in evaluating our work and your positive feedback and confirmation that no changes are required.

Reviewer 2 Report (New Reviewer)

Comments and Suggestions for Authors

This is not the way to cite an article:

The article “Avance de las principales cifras de la Siniestralidad Vial”[2] reported by...

use a standard way.

The article after this minor correction can be accepted. 

 

 

Comments on the Quality of English Language

minor revision 

Author Response

We appreciate the reviewer's comments and have diligently incorporated his suggestions into the revised version of our manuscript. We have modified the citation to the article "Avance de las principales cifras de la Siniestralidad Vial"[3] in the first page using a standard form as indicated by the reviewer.

 

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors make an interesting study on the detection of drowsiness in drivers. The methodological approach is clear. It would be interesting to know why CNN is selected and not the other strategies Support Vector Machines (SVM) or Hidden Markov Models (HMM). It is suggested to include other scientific references detailing the other machine learning techniques mentioned to give more formality to the theoretical foundation of the article. It is suggested to include those aspects where the test performed may fail and how it could be detected or have an error rate of the algorithm once the training phase is completed. Finally, it would be good to know if there was express and consented authorization of the person appearing in the photos, otherwise it would be necessary to make sure of this issue for the protection of their personal data. Very good work!

Author Response

First, the English language and style has been reviewed throughout the document.

Second, more relevant references have been added.

Finally, the person appearing in the photographs is me and I sent the consent by the mail requested.

Reviewer 2 Report

Comments and Suggestions for Authors

The paper requires work that is beyond major revision.

The main issues to be fixed for the future submission are as follows:

1. The paper needs to be reformatted and organised to meet the standard paper.

2. It is not a good practice to put code in the paper, please put algorithms or pseudocode.

3. The paper has reported unnecessary figures instead of scientific inputs.

4. The paper lacks enough simulations and studies.

5. The paper should improve the literature study, which could be found at:

Challenges of Driver Drowsiness Prediction: The Remaining Steps to Implementation | IEEE Journals & Magazine | IEEE Xplore

Author Response

First, the English language and style has been reviewed throughout the document.

Second, more relevant references have been added.

Finally, some references to the code have been removed when they were not necessary and only the explanation has been added.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Given that the paper has not been revised based on the suggestions, the reviewer is unfortunately inclined to reject the paper.