All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
Both reviewers suggest the acceptance of the manuscript.
[# PeerJ Staff Note - this decision was reviewed and approved by Mehmet Cunkas, a PeerJ Section Editor covering this Section #]
Authors have made a considerable effort to take my comments into account. I acknowledge this particularly for the case of the updates made in the state-of-the-art review.
Results also seem to be adequate to the aim to be solved.
Clearly stated. I agree with it.
Useful and worth of publication findings
no comment
no comment
no comment
no comment
Authors need to address the comments from the reviewers, paying attention to highlighting the novelty of their proposal, giving more details about the proposal and the experimental evaluation to ensure reproducibility and, finally, provide a more extensive comparison with the current state-of-the-art.
Finally, as an editor's request, I would like the authors to make their software available for scientific community in line with open science principles.
**PeerJ Staff Note:** Please ensure that all review and editorial comments are addressed in a response letter and that any edits or clarifications mentioned in the letter are also inserted into the revised manuscript where appropriate.
**Language Note:** The review process has identified that the English language must be improved. PeerJ can provide language editing services - please contact us at [email protected] for pricing (be sure to provide your manuscript number and title). Alternatively, you should make your own arrangements to improve the language quality and provide details in your response letter. – PeerJ Staff
This paper proposes a pipeline for video compression based on the selection of relevant frames. The topic of this paper is still relevant nowadays, even though it is a research area intensively researched, for many reasons (video surveillance for crome or terrorist reason, etc) It had a maximum peak of research in the early 2000 (after the september 11th terrorist attacks) and has continued on an alomst constant basis.
My main comments about the paper are as follows:
1) The English level is not bad. There are a few typos that the authors should revise. For instance, in the "abstract", third line: "contri-buting...", or the first work in the Introduction: "Now a days".
2) There is a problem with the reference to the Sections in lines 103-18, page 3/22.
3) I miss a clear and profound framework for the state-of-the-art explanation. Authors can not simply put one reference in lines 57-62, page 2/22. The same for lines 64-68.
4) Lines 193-203, page 5/22. References about video compression are simply too old. Lines 217-220. Just one reference?
5) Section 2.3. Lines 237-239. One reference is simply not enough. There are thousands, just in "2023-":
https://scholar.google.es/scholar?q=video+compression&hl=es&as_sdt=0%2C5&as_ylo=2023&as_yhi=
6) Section 3.1.1. What are the conditions that the 15 selected videos must meet?
7) Section 3.2. Lines 264-266. I miss an explanation about the relevance concept and how this relevance is better than other methods to detect important objects (2023-).
https://scholar.google.es/scholar?hl=es&as_sdt=0%2C5&as_ylo=2023&q=relevant+object+video+processing&btnG=
8) Figures 3 and 4. I do not see them necessary. This is a very well-known architecture.
9) Section 3.5. From my point of view, this does not have any original idea to dedicate 10 lines of explanation.
10) Table 8. What is the numerical comparison here in terms of performance?
Even though I see an effort in the research made by the authors, I miss a more profound comparative analysis with other methods that may do the same. Table 8, in particular, would necessitate more quantitative/numerical analysis in terms of performance comparison.
I consider that there is a real need to develop a competitive pipeline in terms of compression and elimination of video frames for security and surveillance purposes, but this research line is clear, and many people have had to work on this for many years. I do not clearly see how authors have made their best to compare themselves against the state-of-the-art.
I would suggest the authors make a deeper effort to compare their method against others, and in particular, to complete table 8 witn a numerical/performance comparison.
The topic "FRVC: Frame Relevance-Based Video Compression for Surveillance Videos Using Deep Learning Methods" is interesting and highly relevant. However, it would be beneficial to provide more context regarding the significance of this approach.
-
Consider the robustness of the proposed model against adversarial attacks. Given the importance of security in surveillance systems, this would add significant value to the overall contribution.
It would be helpful to mention the accuracy of other existing methods in the abstract, not just the proposed YOLOv9 model, to give a more complete comparison of the improvements your model offers.
The abstract could more explicitly state the novelty of your work, particularly focusing on the unique aspects of your frame relevance-based video compression method, rather than just highlighting the application of YOLOv9.
The article is generally well-crafted and effectively conveys the topic.
Nonetheless, there are several areas that could be enhanced:
1-The introduction would benefit from additional background information to better situate the study within the larger framework of existing research.
It is recommended to include recent developments in video compression and deep learning that are relevant to this work.
2-Certain technical terms in the manuscript lack clear definitions.
For example, terms such as "frame relevance classification" and "compression efficiency" should be explicitly defined for readers who may not be acquainted with them.
3-Although the structure largely adheres to PeerJ standards, the authors should ensure that each section transitions smoothly into the next, thereby improving overall clarity.
Experimental design is a vital component of this study, and several aspects require attention:
1-While the article presents a relevant methodology, it lacks sufficient detail for replication purposes.
It is essential to explicitly specify the datasets utilized, the computing infrastructure employed, and the precise steps undertaken during the experiment.
2-The discussion regarding data preprocessing is insufficiently addressed.
Providing a more comprehensive account of the preprocessing steps implemented prior to the application of the YOLOv9 model would significantly enhance the methodology section.
3-Although some evaluation methods are referenced, they are not described in sufficient detail.
A thorough explanation of the assessment metrics used to evaluate compression performance, as well as the criteria for model selection, should be included to improve transparency and facilitate replicability.
The study's findings are promising; however, there are specific areas that require improvement for a more compelling presentation:
1-The article lacks a clear evaluation of the impact and originality of the proposed FRVC algorithm when compared to existing methodologies.
A comprehensive comparative analysis would strengthen the claims regarding its effectiveness.
2-While the conclusions are generally well-articulated, they need to be more closely tied to the results presented.
It is essential for the authors to explicitly demonstrate how their findings support the conclusions and to acknowledge any limitations that may affect the applicability of their results.
3-The manuscript should more explicitly address any unresolved questions or potential future research directions to provide a clearer framework for ongoing studies in the field.
The article introduces an innovative method for video compression through deep learning, which could significantly benefit surveillance systems and data management.
Nevertheless, improving clarity and providing more detail in certain critical sections would enhance the manuscript.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.