Improvement and Optimization of Underwater Image Target Detection Accuracy Based on YOLOv8
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsDear Authors,
I congratulate you on your work titled "Improvement and optimization of underwater image target 2 detection accuracy based on YOLOv8". The paper states that existing underwater target detection methods are inadequate due to poor image quality and low recognition accuracy. In order to solve this problem, YOLOv8 object detection model is used. Attached, my comments are given about the manuscript.
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe manuscript proposes an enhanced version of YOLOv8s for underwater object detection, introducing three key modifications: replacing a convolutional layer with Deformable Convolutional Network v4 (DCNv4), incorporating the Triplet Attention mechanism, and substituting CIoU with Shape IoU as the loss function. The improvements are evaluated on a relevant underwater dataset and show performance gains in mAP and FPS. The paper addresses an important and technically challenging application and is generally well-structured and technically sound. On the other hand, I have the following questions/concerns:
- In the introduction and conclusion sections, please clearly state the practical motivation for choosing only YOLOv8s, as opposed to YOLOv8m/l/x, beyond stating that it is “compact.” Emphasize the trade-off between real-time performance and accuracy.
- To enhance the completeness of the related work and demonstrate awareness of the most recent advances, please include and cite the following references.
A) Zhang, M., Wang, Z., Song, W., Zhao, D. and Zhao, H., 2024. Efficient small-object detection in underwater images using the enhanced yolov8 network. Applied Sciences, 14(3), p.1095.
Motivation: This work proposes an enhanced YOLOv8 tailored for detecting small objects in challenging underwater environments, which directly aligns with the authors’ goal of addressing occlusions and multi-scale target detection.
B) Song, G., Chen, W., Zhou, Q. and Guo, C., 2024. Underwater Robot Target Detection Algorithm Based on YOLOv8. Electronics, 13(17), p.3374.
Motivation: This paper presents a YOLOv8-based underwater robot detection algorithm with application-specific optimizations. It supports the authors’ claims regarding deployment feasibility in robotic platforms.
C) Alhulayil, M., Aqoulah, M.A., López-Benítez, M., Al-Mistarihi, M.F., Alammar, M. and Al Ayidh, A., 2025. Integrated THz/mmWave Transmission Method for Enhanced URLLC Communications. IEEE Access.
Motivation: While not directly about underwater images, this paper introduces integrated THz/mmWave transmission methods for ultra-reliable low-latency communication (URLLC), which may become increasingly important for real-time control of underwater robots. It supports the broader systems perspective.
D) Varghese, R. and Sambath, M., 2024, April. Yolov8: A novel object detection algorithm with enhanced performance and robustness. In 2024 International Conference on Advances in Data Engineering and Intelligent Computing Systems (ADICS) (pp. 1-6). IEEE.
Motivation: This work offers a benchmark for YOLOv8’s performance enhancements and robustness. It should be cited when introducing YOLOv8 or justifying its architectural choices.
- Figures and Captions
- Figures 1 and 3: Please revise the captions to better reflect the experimental setup and significance of each figure rather than repeating generic labels (e.g., “Physical image of ROV”).
- Style and Grammar
- Minor grammatical issues exist throughout the paper. For example:
- Line 24: “thereby markedly enhancing” → consider “resulting in a marked enhancement”.
- Line 49: “dedicated substantial efforts to this field” → “dedicated substantial efforts to advancing this field”.
- Consistency: Use either “YOLOv8s” or “YOLOv8-S” uniformly throughout.
- Methodology Clarification
- In Section 2.2.2, the motivation for selecting the ninth convolutional layer specifically for DCNv4 replacement should be explained more clearly.
Minor edit is needed.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsIn the manuscript entitled “Improvement and Optimization of Underwater Image Target Detection Accuracy Based on YOLOv8,” the authors present relevant results. Despite the manuscript being well prepared, there are some points that must be addressed before recommending it for publication:
- The figures lack detailed descriptions. Therefore, I recommend enhancing the figure captions to provide clear context for each image and help the reader better understand their relevance within the study.
- I suggest adding a section explicitly highlighting the novelty of the work compared to previously published studies. This section should be highly descriptive to clearly demonstrate the scientific contribution of the manuscript.
- Several images could potentially be merged. I recommend carefully reviewing all figures to determine which ones can be combined for clarity and conciseness.
- If the robot employed is not commercially available, it would be helpful to include the models of the components used (including the electronic boards) to allow for the reproducibility of the results.
- The discussion section lacks references. It is important to include citations of relevant research to support the discussion—such as comparisons with other studies. I suggest rewriting this section to strengthen the analysis and improve its scientific rigor.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for AuthorsPlease see the attached file.
Comments for author File: Comments.pdf
Please see the attached file.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsDear Editor and Author(s),
The similarity ratio should be reduced. Other correction requests have been fulfilled. Otherwise, it should not be considered suitable.
Author Response
Comments 1: [The similarity ratio should be reduced. Other correction requests have been fulfilled. Otherwise, it should not be considered suitable.]
Response 1: Thank you for your suggestion.The similarity ratio has been reduced to 9%, please refer to the supplementary materials uploaded for the similarity_report.
Reviewer 3 Report
Comments and Suggestions for AuthorsThe authors have properly addressed the comments and suggestions. Therefore, I recommend accepting the manuscript for publication
Author Response
Thank you for your suggestion.
