User-Centered Design of a Computer Vision System for Monitoring PPE Compliance in Manufacturing

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article presents a valuable and practical approach to monitoring Personal Protective Equipment (PPE) compliance in manufacturing using computer vision technology. The article makes a significant contribution to industrial safety and the application of AI in production environments.

Some suggestions to consider by the Authors:

The methods are thoroughly explained, including empathy mapping, architectural design (C4, physical/logical diagrams), and usability testing. More details on YOLOv8 training (e.g., dataset specifics) would be beneficial.

Additional tests under challenging conditions (e.g., low light, rapid movement) would strengthen the model’s validation.

The focus is on prototype usability testing, but results from actual industrial deployments are not presented.

Figures (e.g., empathy maps, architecture diagrams) and tables (e.g., product backlog) are clear but could benefit from higher resolution or simplified labels for broader readability.

The language is academic and fluent, with minor grammatical errors (e.g., occasional awkward phrasing). A final proofreading would enhance clarity.

 

Author Response

Comment 1: The methods are thoroughly explained, including empathy mapping, architectural design (C4, physical/logical diagrams), and usability testing. More details on YOLOv8 training (e.g., dataset specifics) would be beneficial.

Response 1: Thank you for pointing this out. We have carefully reviewed the manuscript to ensure that this issue is adequately addressed. In particular we explain in detail the dataset we will use in the application, with the number of images and which ones will be used for training, validation and testing. As well as talk a little about the expectation of the mAP of our application. This change is in the paragraph located between line 332 to 339.

Comment 2: Additional tests under challenging conditions (e.g., low light, rapid movement) would strengthen the model’s validation.

Response 2: We appreciate the comment. This suggestion was taken into account, and we now include a mention of the model’s training under challenging conditions (such as low light and rapid movement) in lines 99 to 102 of the revised manuscript. This helps clarify the robustness of the model in more complex scenarios.

Comment 3: The focus is on prototype usability testing, but results from actual industrial deployments are not presented.

Response 3: We thank the reviewer for this valuable observation. In response, we clarified in the revised manuscript that the individuals who participated in the evaluation are professionals actively working in the manufacturing industry. This information, now included in lines 471 to 473, reinforces the practical relevance of the usability testing and strengthens the context of the prototype's application.

Comment 4: Figures (e.g., empathy maps, architecture diagrams) and tables (e.g., product backlog) are clear but could benefit from higher resolution or simplified labels for broader readability.

Response 4: Gracias por su comentario constructivo. Hemos tomado en cuenta su observación y realizamos una revisión del manuscrito con el objetivo de corregir este detalle. Es por ello que realizamos la revisión de todas las figuras y tablas para agregarle una mayor resolución y que se puedan entender al detalle.

Comment 5: The language is academic and fluent, with minor grammatical errors (e.g., occasional awkward phrasing). A final proofreading would enhance clarity.

Response 5: We appreciate the reviewer’s suggestion. A careful proofreading of the manuscript was carried out, and the minor grammatical issues and awkward phrasings we were able to identify have been corrected to improve clarity and readability.

Reviewer 2 Report

Comments and Suggestions for Authors

1- The full spelling of the abbreviation should be provided where it is first used in the manuscript. such as lean UX (line 71),  EPP (line 90), BiFPN (line 92), R-CNN (line 130), MEAG-YOLO (line 147),   MSCA (line 149), ADHD (line 162),  PPEYE (line 301.

2- The contribution of the article to the literature should be rewritten by indicating the difference from the existing ones.

3- Table-2, table-5 and table-6 should be rearranged on the page.

4- The methodology is described in detail in section 3.

5- Experimental application and results are shared in detail.

6- The positive outputs of the developed application are given in detail, but the negative outputs are not shared. The authors should also share the negative outputs and discuss the discussion section.

Author Response

Comment 1: 1– The full spelling of the abbreviation should be provided where it is first used in the manuscript, such as lean UX (line 71), EPP (line 90), BiFPN (line 92), R-CNN (line 130), MEAG-YOLO (line 147), MSCA (line 149), ADHD (line 162), PPEYE (line 301).

Response 1: Thank you for your valuable comment. We agree with this observation. Consequently, we have revised the manuscript to address these problems. Specifically, we have corrected the error of using abbreviations without first mentioning them, specified the abbreviations as YOLO (line 37), lean UX (line 64), BiFPN (line 91), R-CNNN (line 142), MEAG-YOLO (line 163), MSCA (line 166), ADHD (line 186), in addition to specifying that PPEYE is the name of the application we propose and corrected the abbreviation of EPP.

Comment 2: The contribution of the article to the literature should be rewritten by indicating the difference from the existing ones.

Response 2: We appreciate the reviewer’s constructive comment. In response, we have clarified the distinctive contributions of our work in comparison to existing studies. These differences are now explicitly discussed in the revised manuscript in lines 97 to 102, 124 to 129, and 170 to 175, helping to better position our approach within the current literature.

Comment 3: Table-2, table-5 and table-6 should be rearranged on the page.

Response 3: We sincerely appreciate your comment. We agree with the point raised and consequently, we have made the appropriate modifications in the manuscript to correct this aspect. Specifically, we have reorganized all the tables and figures in the document so that it is understandable and does not have page breaks. These changes are shown in each image and table.

Comment 4: The methodology is described in detail in section 3.

Response 4: We appreciate your valuable comment and appreciate highlighting that point. We did a detailed approach and research to describe in detail the methodology and it gives us happiness to know that we did a good job describing in detail.

Comment 5: Experimental application and results are shared in detail.

Response 5: We sincerely appreciate your appreciation for the experimental application and presentation of the results. We are pleased to know that the information provided was clear and detailed. This type of feedback motivates us to continue to strengthen the quality of our work.

Comment 6: The positive outputs of the developed application are given in detail, but the negative outputs are not shared. The authors should also share the negative outputs and discuss the discussion section.

Response 6: Thank you for pointing this out. We have carefully reviewed the manuscript to ensure that this problem is adequately addressed. We have added in the discussion part some negative aspects of our results, which will serve to take into consideration and improve these aspects in our application. This change can be found on page 18, paragraph 2, from line 502 to 508.

Reviewer 3 Report

Comments and Suggestions for Authors

This paper presents a well-structured design and usability evaluation of a computer vision desktop application to monitor PPE usage in manufacturing environments. The focus on a lightweight, locally operable YOLOv8-based system is practical, particularly for small and medium-sized companies.

Strengths:

• The use of Lean UX methodology throughout the development process is clearly described.

• SUS testing with 50 users is a solid way to validate the usability claims.

• Figures and system diagrams are well-organized and help readers understand the architecture.

Suggestions for improvement:

1. English Language: While the paper is readable, there are several awkward phrases and grammatical errors throughout. A language review is recommended to improve clarity and fluency.

2. Methods Section: Consider expanding the explanation of how the YOLOv8 model was trained. Details about the dataset used (e.g., train/test split, annotations) and performance metrics (e.g., accuracy, mAP) would help readers better evaluate the technical strength.

3. Limitations: The discussion briefly mentions future improvements. It would be beneficial to add a more explicit limitations section to address aspects like generalizability, false positives, or dependency on camera quality.

4. References: The literature review is comprehensive and up-to-date. However, some references are preprints (e.g., arXiv) and could be replaced with peer-reviewed alternatives if available.

Author Response

Comment 1: While the paper is readable, there are several awkward phrases and grammatical errors throughout. A language review is recommended to improve clarity and fluency.

Response 1: We appreciate the reviewer’s suggestion. A careful proofreading of the manuscript was carried out, and the minor grammatical issues and awkward phrasings we were able to identify have been corrected to improve clarity and readability.

Comment 2: Consider expanding the explanation of how the YOLOv8 model was trained. Details about the dataset used (e.g., train/test split, annotations) and performance metrics (e.g., accuracy, mAP) would help readers better evaluate the technical strength.

Response 2: Thank you for pointing this out. We have carefully reviewed the manuscript to ensure that this issue is adequately addressed. In particular we explain in detail the dataset we will use in the application, with the number of images and which ones will be used for training, validation and testing. As well as talk a little about the expectation of the mAP of our application. This change is in the paragraph located between line 332 to 339.

Comment 3: The discussion briefly mentions future improvements. It would be beneficial to add a more explicit limitations section to address aspects like generalizability, false positives, or dependency on camera quality.

Response 3: We thank you for your input. We agree with the observation and have worked on updating the manuscript accordingly. So we talked about the limitations that our application will have such as the quality of the cameras or the minimum processing that a team must have for the fully efficient operation of the functionalities. The changed section is in lines 502 to 508.

Comment 4: The literature review is comprehensive and up-to-date. However, some references are preprints (e.g., arXiv) and could be replaced with peer-reviewed alternatives if available.

Response 4: We thank the reviewer for this helpful suggestion. In response, we have replaced the preprint source with a peer-reviewed publication, where available. In addition, the corresponding section of the manuscript, from lines 79 to 82, was updated to reflect this change.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for your efforts to improve the quality of the manuscript.