HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object Detection

Round 1

Reviewer 1 Report

The work proposed A Transformer-Based Hybrid Network for Real-Time Underwater Feeble and Small Object Detection. The idea is interesting and the reviewer only has few comments.

1.       Some important works in deep learning or AI models in object detection using remote sensing images may be further supplemented, e.g., ORSIm detector, UIU-Net, etc.

2.       What is the main difference between the proposed method and existing methods. The newly-added values can be further clarified.

3.       How about the computational complexity? How to show the real time property?

4.       How to define the small objects? What are the main challenges for this kind of objects? How to solve them accordingly.

5.       Some future directions should be pointed out in the conclusion.

Author Response

Author Response File: Author Response.docx

Reviewer 2 Report

This paper proposes A Transformer-Based Hybrid Network for Real-Time Underwater Feeble and Small Object Detection. It looks interesting and innovative. However, there are some minor comments and suggestions.

1.       The title is complex for the reader, it is suggested the title should be revised/changed.

2.       The abstract section of the paper needs to be rewritten, as the abstract is very difficult to understand. There are more abbreviations used in this section which is difficult for the reader to understand.

3.       figure 1 should be moved into a suitable place in another section.

4.       please kindly request to reduce the plagiarism report by up to 15% the report is attached here.

5.       Many grammatical mistakes are found throughout the paper, correct them.

6.       So many abbreviations are used in the entire paper that, again, it is very difficult to understand. It should be double-checked.

7.       The equations need to be properly discussed and numbered

8.       The future work are missing at should be add after conclusion a few lines

9.       The format of the literature citation is not uniform and needs to be adjusted.

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

The paper proposes a framework for underwater object detection based on a CNN-transformer hybrid network. 

The writing is well structured, and the research topic could interest researchers.

Here are some suggestions for improving the article:

-Some words are misspelled; consider checking the spelling in detail (i.e.,  hybird in the abstract and other words in the text)

-Description of the variables used in the paper ( i.e., C in line 183)

- 32x32 ? pixels? (line 30)

-line 63 - replace a inductive...  by an inductive...

-Include an analysis of time concerning other methods.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

Proposed schematic diagram can be included.

Title is generic, can be included appropriate to your methodology.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The authors have well addressed the reviewer's concerns. No more comments.