Design and Implementation of Industrial Accident Detection Model Based on YOLOv4
Round 1
Reviewer 1 Report (Previous Reviewer 2)
Dear authors,
I understand that you have considered all my suggestions, and I am now able to recommend your manuscript for acceptance after polishing its English language.
Moderate editing of English language required
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report (New Reviewer)
The title seems to be more generic. Adding more specific details can make the title more informative. It should highlight the unique contributions.
The introduction should explicitly state the research objectives and questions that the study aims to address. This will help readers understand the purpose of the research.
Ensure all sources are properly cited to support the claims made in the introduction
There is some repetition of information, particularly regarding Korea's accident rate and the industries affected. Please maintain conciseness.
It would be helpful to include a brief mention of the potential limitations of the study
The introduction of various object detection models is informative, but it lacks a clear connection to the research context. Explain how these models relate to the study's goal of industrial accident prevention using CCTV data.
YOLOv4 as the chosen model, but it would be helpful to elaborate on the reasoning behind this choice. What specific criteria or performance metrics led to the selection of YOLOv4 over other models?
Nine objects selected for detection, but it lacks clarity on how these objects are relevant to the risk scenarios and how they contribute to the overall model's effectiveness.
Used CVAT tool for labeling, but it would be helpful to explain why CVAT was chosen over other available options and how it specifically meets the requirements of this study.
it would be beneficial to include some context or explanation of these time measurements. For example, what is considered an acceptable response time and delay time for industrial accident detection systems? How do these times impact the practical use of the model in real-world scenarios?
Try to reduce the conclusion section and add another discussion section to explain the advantages, limitations and future realtime implementations
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
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
1.未考虑光照条件和遮挡对应用效果的影响。
2.场景推导和对象选择的方法相对主观。
3.应用方法比较常规，缺乏创新性。约洛。v5 可能具有更好的性能。
2. The method of scene derivation and object selection is relatively subjective.
3. The application method is conventional and lacks innovation. Yolo. v5 may have better performance.
Reviewer 2 Report
Dear authors, you mentioned a very important theme of preventing accidents at work in industry. However, your manuscript must be revised sufficiently. My principal suggestions are as follows:
1. OSH terminology must be used. For example, workers wear personal protective equipment (PPE), risks are assessed, etc.
2. Both scientific soundness and practical implementation are poor. Today, CCTV is used extensively to prevent workplace accidents.
3. The findings do not provide a clear understanding of how research will help prevent work-related accidents.
4. There are multiple mistakes and negligence throughout the text (for example, Figure 1 DOES NOT DISPLAY 'Dangerous situation determination model process', the titles of Figures 1 and 5 are the same, etc.)
5. Quality of Figures is extremely poor.
Extensive editing of the English language required.
Reviewer 3 Report
After carefully reading the manuscript, I note that the current version may not be suitable for publication due to the following reasons:
1. the writing up and the structure is poor. Lots of the figures are difficult for readers and some sentences are not easy to follow.
2. The latest YOLO updated to v8, while the current one discussed in the manuscript is v4. why not comparing the latest one? In addition, validation should be given in the case study to prove the reliability of the model.
3. overall the novelty of the study is weak as this study only used tradition approaches to detect potential danger scenarios. however, the application is interesting, thereby, I may suggest authors to revise the manuscript and think if the author can do better in the next version.
