FL-YOLOv7: A Lightweight Small Object Detection Algorithm in Forest Fire Detection

Round 1

Reviewer 1 Report

FL-YOLOv7: A Lightweight Small Object Detection Algorithm

in Forest Fire Detection.

 

1. The abstract is not correlated with the title of the paper. The abstract of the manuscript should be revised as the new contributions and the explanation are not clear.

2. Please check the manuscript carefully, for examples, what do some abbreviations state for the first time; such as FL-YOLOv7, ASFF, Wise-IoU in the Abstract and etc.

3. The authors claimed that the proposed method uses an improved FL-YOLOv7 for smoke an fire recognition, but I haven't seen detailed information about how they improved the existing YOLOv7 approach. They should clearly write in the abstract that they used the already improved YOLOvt method or they improved the traditional YOLOv7 artichecture by themselves during developing the forest smoke (fire) detection method?

4. Please make a Related Work section with recently published fire detection and recognition papers such as Machine/Deep Learning and other famous approaches. I reviewed from the internet several following forest fire detection papers that are expected to help more ideas to readers with previously achieved works. Try to discuss these papers outputs and limitations in the Discussion section.

https://doi.org/10.3390/fi15020061

5. Figure 1 should be cited. 

6. Dataset. Please make a Table to describe the number of used fire images dataset and other important information.In the dataset section, please make the tables by showing the number of images for daytime and nighttime retrievals. 

7. Can the authors comment on the minimal requirement for the training data set (e.g., the number of images required)? Are you using the augmentation technique to increase the training dataset artificially?

8. Any way to reduce the computational cost compared to other approaches, please discuss.

9. More information is required about the method followed in the so-called subjective evaluation. I mean about the procedure and environment (the information provided to the subjects.

10. Achieved results caused by overfitting? Please discuss in experiment section.

11. The authors didn’t mention any limitations of the proposed method, they should. Proposed method how to distinguish fire-like scenarios like sunrise or sunset? 

12. Using this system can we recognize artificial fires or smoke? What I mean is sometimes fire detection methods detected the fire even in non-fire pictures because fire-like scenarios (environments) were incorrectly classified as fires and warnings. Human eyes easily distinguish artificial fires, but computers sometimes incorrectly classify neon signs, streetlights, and the headlights of vehicles as real fires because they have similar brightness, and reflection.

13. Once fire accident is detected by your approach and how the notifications are sent to the fire department or other arbitrators?

14. Conclusion part is needed to be written with major finding of this article. And future research direction on how to fix that challenge.

Moderate editing of English language required

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

See attached file. 

Comments for author File: Comments.pdf

Many repetitions are present. Punctuation is sometime misused.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

1) The authors tried to address all my comments to improve the paper's quality and readability but I suggest to the authors that they use the text of the answer offered to my comment to include in their paper not only in the cover letter. That text is much better written and is more understandable.

 

2) They should use references to support their claims. 

3) Suggested to include the following paper in the related work section and discuss output results in the experimental results or discussion section. It was recommended in previous comments but it is not visible in a cover letter or revised version of the manuscript.

Moderate editing of English language required

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper notably improved after its revision, and I have no further comments. The only minor point is that charts of Figure 10 have poor resolution. Therefore, I suggest the Authors to improve them in order to not hinder their readability.

Author Response

Thank you very much for your suggestion, we have adjusted the clarity of figure 10