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 paper may be accepted.
[# PeerJ Staff Note - this decision was reviewed and approved by Xiangjie Kong, a PeerJ Section Editor covering this Section #]
-
-
-
Please respond to the comments from the reviewers
This paper reports a gas plume detection and segmentation method for multibeam water column image. The experiments verify the effectiveness of the proposed method. The following points could be considered.
In the abstract and Introduction, the motivation and novelty of the paper should be improved.
More detailed explanation of the C-BiFormer module architecture and implementation should be provided. The current description lacks sufficient technical details for reproducibility.
The ablation studies should be expanded to analyze the individual contribution of each key component in the proposed YOLOv7-CB model. Current experiments do not fully justify design choices. Additional ablation experiments are needed to validate the effectiveness of different modules.
Detailed description of data collection process, data preprocessing steps, train/validation/test split ratios should be provided.
More state-of-the-art methods should be compared in the experiments. Deeper analysis of when/why the proposed method fails is needed.
Hardware specifications, training protocols, hyperparameter selection, and runtime performance metrics should be included.
Additional evaluation metrics, such as inference time analysis, model size, and memory usage should be included.
The resolution of the experimental results is low. More visual results are needed.
1.1 In the introduction, the author introduced in detail a large number of related works in the field of generating high-resolution water column images, especially the work related to object detection based on YOLO. However, as a method that combines CNN and transformer, it is necessary to introduce representative transformer-based related works in recent years, and the references of YOLO-related methods mostly introduce articles in the past two years, which helps to improve the persuasiveness of the method.
1.2 The related work section introduces the relevant principles of YOLOv7 and Transformer, but the introduction to the limitations of these methods is slightly insufficient. Introducing the problems that still exist in the current work in combination with the difficulty of the gas plume target task is more conducive to the introduction of this article's methods and motivations.
1.3 The model and experimental images shown in this article are too low in definition, especially the experimental images. It is impossible to compare the specific morphology of the target detected by the proposed method through the local magnified images. Please replace them with the original high-resolution versions of various images.
2.1 The experiment used 320 water column images collected by the Kongsberg EM710 multibeam bathymetry system. The number of data sets may not be enough to fully demonstrate the performance of the method (the method may be designed only for this data set). If possible, please introduce more data sets into the training and testing experiments.
2.2 In the module ablation experiment and the comparative experiment of different methods, the author focused on the changes of related indicators, but lacked further analysis. It is recommended to supplement the analysis of the impact of different variants of each component in the model on the final result (such as the possible reasons for the performance change caused by using different numbers of ELAN modules)
3.1 In the comparative experiment, the author only showed the detection and segmentation results of the baseline and the proposed method YOLOv7-CBI. Please add more detection results of the comparative methods to facilitate a more intuitive comparison of the detection and segmentation performance of different methods under water column images.
3.2 It is recommended that the authors point out the limitations of the current method in the summary chapter and describe the future work plan in more detail, which will help inspire researchers in related fields to conduct further exploration and application.
NA
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.