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.
The authors seem satisfied with the requested changes and therefore I can recommend this article for acceptance.
[# PeerJ Staff Note - this decision was reviewed and approved by Xiangjie Kong, a PeerJ Section Editor covering this Section #]
no comment
no comment
no comment
no comment
paper is written according to standard
experiment is given properly
results are given properly
authors add most there papers in references for self citations
Good
Good
Good
I have no further questions, and all previous inquiries have been addressed. Consequently, I would like to accept the manuscript for further processing. Thank you.
**PeerJ Staff Note:** Please ensure that all review, editorial, and staff 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.
**PeerJ Staff Note:** PeerJ's policy is that any additional references suggested during peer review should only be included if the authors find them relevant and useful.
**Language Note:** When preparing your next revision, please ensure that your manuscript is reviewed either by a colleague who is proficient in English and familiar with the subject matter, or by a professional editing service. PeerJ offers language editing services; if you are interested, you may contact us at [email protected] for pricing details. Kindly include your manuscript number and title in your inquiry. – PeerJ Staff
1. Ensure the abstract clearly states the problem, the limitations of existing methods (especially DETR-based ones in small object detection), and how PF-DETR addresses those.
2. Highlight the significance of the problem in real-world aerial applications (e.g., surveillance, urban planning, disaster monitoring).
1. The idea of Progressive Fusion is interesting. Make sure the paper clearly defines what "progressive fusion" entails and how it differs from existing fusion or refinement mechanisms in DETR-like models.
2. Emphasize what components of PF-DETR are novel
1. Clarify how the fusion is progressively applied — is it in the encoder, decoder, or both?
2 .Use diverse aerial datasets (e.g., DOTA, VisDrone, xView) to validate generalization.
This is novel research work
Experiment details are given properly
Results are discussed in detail
Overall, the paper is organized properly; the concept and future research directions are extensively explained. So, the paper is accepted after following minor changes:
1. The problem of paper and motivation is not clear in the introduction
2. Comparison of current research is not given with previous work
3. Algorithm is discussed, but no pseudo-code is given for the readers' ease
4. Figure 2 colors are light and not clear, so add dark colors for a clear view
5. The paper contains a few grammar mistakes, which will be corrected in the final version.
6. Only a few references are added in the paper, but more than 50 references are added to attract readers, and add a few latest references related to this paper
The article is generally well written, although minor editing is recommended to improve fluency in some long and densely packed sentences. The manuscript is structured according to academic norms, and the figures, tables, and raw data are appropriate and well presented.
The background is extensive but overly citation-driven. Many methods are listed with minimal synthesis or critical comparison. The Related Work section could benefit from clearer differentiation between existing models and the proposed PF-DETR. I recommend citing Gui et al. (2024) (https://doi.org/10.3390/rs16020327) to provide a more recent and integrated context for object detection methods in remote sensing and show where this work fits in the broader development of deep learning approaches.
The manuscript includes all necessary formal components. However, some terms, such as those introduced in custom modules (e.g., S2-CCFF, CSPOK-Fusion, PPA), should be more clearly defined when first introduced.
The presentation of results would benefit from deeper interpretability analysis, particularly in the visualizations of detection and attention maps, to support the claims about effectiveness in small object detection.
The research question addresses well-known challenges in small object detection in aerial images. The proposed method, PF-DETR, builds upon RT-DETR and introduces several architectural enhancements targeted at this problem.
The selection of RT-DETR as the sole baseline backbone is not sufficiently justified, and the exclusion of more recent detectors like RTMDet or YOLOv8 weakens the experimental rigor. A brief rationale for this design decision should be added.
The justification for the combination of modules (PPA, S2-CCFF, CSPOK-Fusion, and NWD loss) should be strengthened. Currently, each module is introduced independently, but the manuscript does not explain why they are used together, nor does it evaluate whether a simpler configuration might be sufficient (more ablation studies, please). The model complexity and increased FLOPs are notable, and their implications for deployment in real-world aerial scenarios should be discussed.
The performance gain is relatively limited (e.g., 3.1% mAP improvement), especially considering the increased model complexity. The cost-benefit analysis in terms of detection performance versus computational overhead is not discussed in depth and should be added to validate the applicability of the proposed method.
The NWD loss is properly used and correctly referenced in its mathematical formulation, but the manuscript should clarify that this loss is not novel and is integrated from previous work. Claims should be adjusted accordingly.
The conclusions are generally well stated and tied to the research questions. The limitations of the method and possible future directions are briefly mentioned, but should be discussed more transparently.
The architectural design needs to be better justified as a coherent whole. The Related Work section should be revised to include recent high-level reviews, and the discussion of feature fusion should be expanded. I suggest reading works like Albanwan et al. (2024) (https://doi.org/10.14358/PERS.24-00110R1) to support the fusion-based design rationale.
It would be helpful to provide class-specific examples or failure cases to show where PF-DETR succeeds and where it struggles. This will also help improve the interpretability of the results.
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.