FFYOLO: A Lightweight Forest Fire Detection Model Based on YOLOv8

Round 1

Reviewer 1 Report

This manuscript, titled “FFYOLO: A Lightweight Forest Fire Detection Model Based on YOLOv8” delves into the research background, objectives, methods, processes, and results of a forest fire detection. The main improvements are as follows: a channel prior dilatation attention module (CPDA) is proposed in order to better extract the features of fire and smoke; the mixed-classification detection head (MCDH), a new detectionhead is designed; the MPDIoU is introduced to enhance the regression and classification accuracy of the model; in Neck section, a lightweight GSConv module is applied to reduce parameters while maintaining model accuracy; Finally, knowledge distillation strategy is used duringtraining stage to enhance the generalization ability of the model and reduce the false detection.

The structure of the main text is clear and its logic is rigorous. The authors provided a detailed introduction to the research background and explained the importance of this study. In the methods section, the authors put their methods in detail, enabling readers to understand the entire implementation process of the study. In the experiments and analysis section, the authors presented the experimental results in detail and conducted in-depth analysis. In the discussion section, the authorssummarized the entire text and pointed out the limitations of the research and future research directions.

The reference section lists the literature materials that the author referred to during the research process, covering relevant knowledge from multiple fields, and has good reference value.

In addition, the data and charts in this manuscript are very detailed, supporting the main points and conclusions. The data presented in this study are available on request from the corresponding author to help readers better understand the content of the paper.

In summary, this manuscript meets the publication requirements, and I suggest accepting its publication.

Author Response

Thank you very much for taking the time to review this manuscript, your recognition of this work is a great encouragement and motivation for our future work. Please find the detailed revisions in the re-submitted files.

Reviewer 2 Report

The article “FFYOLO: A Lightweight Forest Fire Detection Model Based on YOLOv8” proposes a new method for early detection of forest fires using deep learning. The manuscript exhibits a high level of writing proficiency and presents a novel approach in its methodology.

The abstract is concise and informative.

The introduction is well prepared using relevant references. However, the research hypotheses are not clearly stated. By clearly specifying the hypotheses, the article provides an obvious framework for the research. I suggest adding the research hypotheses to provide a logical and structured discussion of the research objectives.

The methodology has been stated clearly.

The results have been presented very well.

The discussion can be improved considering the research hypotheses.

Author Response

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files.

Comments 1: The discussion can be improved considering the research hypotheses.

Response 1:  Thank you for pointing this out. We agree with this comment. Therefore, we have revised the content of the third and fourth paragraph to better reflect our research hypotheses. We move the summary of the experimental results of the fourth paragraph to the third paragraph of the original text. In the fourth paragraph, we summarize the method proposed in this paper and correspond it to the issues mentioned in the introduction.

(please refer to paragraph 3&4 for details in Section 4: Conclusions)

Reviewer 3 Report

Dear Authors,

The subject of the paper is a current and very important topic since it is a contribution to the study of forest fire detection methods. In my opinion, the subject fits the scope of the Fire Journal.

The paper proposes a lightweight forest fire detection model based on YOLOv8.

 Main Remarks:

- The document is well written and clearly, but the article does not have a conclusions section.

-The discussion of results is presented throughout section “3. Experiments and Analysis.”

- I propose to change the name of section “4. Discussion” to “4. Conclusions”.

- Line 150: “And YOLOv7 [23] proposed a…”. Remove “And”

- Lines 155-156: “…task of class prediction By appropriately…”. Replace by“…task of class prediction by appropriately…”

Author Response

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files.

Comments 1: I propose to change the name of section “4. Discussion” to “4. Conclusions”.

Response 1:  Thank you for pointing this out. We agree with this comment. We have changed the name of section "4. Discussion" to "4. Conclusions".

Comments 2: Line 150: “And YOLOv7 [23] proposed a…”. Remove “And”

Response 2:  We have removed the word "And".

Comments 3: Lines 155-156: “…task of class prediction By appropriately…”. Replace by“…task of class prediction by appropriately…”

Response 3:  We have changed "By" to "by".