Vehicle Detection in Adverse Weather: A Multi-Head Attention Approach with Multimodal Fusion

Round 1

Reviewer 1 Report

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors,

the paper focuses on improving MVDNet by integrating a multi-head attention layer to improve its performance. This paper confirms the improved performance of MVDNet for multi-head attention based on comprehensive testing on a training dataset obtained from the Oxford Radar Robot. Overall, the work is correct and well organized, although there are some editing errors and some additions are required.

Detailed comments on the manuscript:

1. At the very beginning of the introduction, I propose to add a paragraph regarding the general development of autonomous vehicles. You can add references here, e.g.: 

2. Please consider whether there is a need to use so many keywords, especially since some of them are repetitions of expressions from the title of the manuscript.

3. Chapter 2 provides a good introduction to the subject matter of the article.

4. The authors presented the research methodology used quite well, although Figure 4 is not very legible, it may be worth increasing the font size of the diagram elements.

5. Chapter 4 requires editing corrections and additions. There is no reference to Figures 6 and 7 in the text. It is good if the chapter does not end with a drawing. It may be worth adding a comment or introducing elements of discussion with other research works in this area. This is a weakness of this manuscript.

6. Figure 5 should be moved after the first paragraph of chapter 4.2.

7. It seems that not all references in the reference list follow the template.

Despite these comments, I believe that the work is important and can be printed after corrections.

Thank you!

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Though the research submitted targets an important area of technology today, but I am unable to attribute any of research aspects to the scope of Low Power Electronics and Applications. The paper title is rather misleading by including a phrase “… low power ubiquitous sensing …” because any modeling, assessment or even discussion related to power consumption or efficiency are absent in the paper. As typical nowadays, authors fully dedicate their attention to the quality of machine learning algorithms and their application, which is information technology or math theory field. The subject is data fusion from lidar and radar but any target hardware for low power electronics is not considered at all. Therefore, my suggestion is that the paper should be submitted to other journal with the relevant scope.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Dear Authors,

Thank you for the changes made to the manuscript, the work is definitely better, so I recommend the work for publication.

Regards

Reviewer 3 Report

Authors have addressed the pointed-out issue formally. They have added more description of lidar and radar, as well as future planes to deploy trained models on edge single board computer. Unfortunately, the paper does not focus on power consumption in detail, except the general note that application is dedicated for automotive industry.