Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study aimed to propose a Spatial and Efficient Channel Attention (SECA) mechanism, which enhance the interpretability of smoke detection, e.g., using multi-kernel 1-dimensional (1D) convolutions to capture the diffusivity of smoke.

Abstract:

The research topic is presented in an unambiguous manner. The abstract includes the problem definition and its place in the literature.

It clearly states the suitability of the method for the study's purpose. However, its unique contribution is not clearly emphasized.

The Abstract does not present any quantitative findings. The findings are not supported numerically or data-based.

Introduction

The introduction is sufficient in introducing the topic, summarizing the literature, and presenting the original contribution. The sources are high-quality, sufficient in quantity, and impressive in terms of timeliness.

Some studies are cited only by method name, but critical information such as the dataset on which the method was tested or the metrics on which it performed is not provided. Briefly summarizing the success metrics or key results of some of the sources used in the introduction (e.g., SASC-YOLO, ARGNet) would increase the technical depth of the literature review.

The section titled "Related Works" explains the methodological underpinnings of the study as part of the "Introduction" section. The "Related Works" section explains the study in a systematic, technical, and layered manner. The limitations of both classical and modern approaches are explained in this section.

Method

The study presents the proposed SECA mechanism, an attention module designed for a complex, multi-scale image recognition problem such as forest smoke detection.

The method aims to overcome the limitations of existing attention mechanisms. It also includes important engineering solutions for integration into YOLOv8 and the use of the Haar Wavelet Transform (HWT) to reduce computational overhead.

Results

The results from the study are compared with the results from similar researches. Even though discussion elements are presented in “Results” section, a separate “Discussion” section with critical evaluation should be included in the paper and limitations of the findings should be given. Additionally, statistical significance is not adequately evaluated.

Conclusion

The Conclusion section summarizes the key findings and implications of the study, but it could be strengthened in several ways. The study should more clearly explain not only what was done, but also why it is significant and how the proposed approach can be evaluated or extended in future work.

The structure and presentation of the paper is well organized. The figures and tables are suitable and clearly explain the content.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Overall this is a detailed paper which is for the most part well-written and structured, with some illustrations. The background and context are both good and useful, and the summary and conclusions are appropriate. Personally, whilst the paper is good, I do find that like so many papers there is a strong focus on the technical aspects of the approach without a less technical overview which I think would be beneficial to the reader who may wish to skim read the paper initially without stopping to delve into the fine detail. In this paper, there is clearly a lot of technical content - not only the methodology presented but also the review of existing work - and the diagrams are also quite specific - but without a more basic summary/coverage. It may be that there is not enough space to do this - not maybe does it fit with the Journal's requirements - but it would be helpful from an accessibility point of view. This is not about dumbing down the content - merely about some statements that summarise in a sentence or two or paragraph or two to present the content in a non-technical way. I would also like to have seen - perhaps in the same context - some illustrations/photographs of the problem - there are some later in the paper but these are summary examples as part of a diagram or test example. Personally I think this would enhance the paper somewhat.

Comments on the Quality of English Language

The English is a little ropey in places re: syntax and grammar and perhaps would benefit from a minor edit with the help of a native English speaker and proof reader as well as the use of e.g. Grammarly.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The authors proposed a YOLO based Spatial and Efficient Channel Attention mechanism, termed as SECA, to enhance the interpretability of smoke detection using multi-kernel 1D convolutions to capture the diffusivity of smoke. To speed up, a Dsconv-Haar Wavelet Downsampling technique called DHWD is also provided. Compared to existing methods, their method can achieve a better or at least a comparable performance in smoke detection. Overall, the submission is readable and innovtaive. Still, there are some minor suggestions on the improvements.

1. The proposed attention mechansims are based on YOLO which is very important but not shown in the title. Please show "YOLO" in the title.
2. DWConv1D should be defined after Eq. (3).
3. The proposed attention mechanisms in Fig. 1 should be described using the traditional Q, K, and V symbols. This is to reflect the authors point view (2) from lines 188-191.
4. The input and output in Fig. 1 should be replaced with feature map and attention map, respectively.
5. The symbol like a rectangle box in line 231, 304, and 308 shoudl be changed for clarity.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

You seem to have addressed the reviewers' comments and paid attention to the edits requested to improve the manuscript.

Comments on the Quality of English Language

English is now improved.