Vehicle Re-Identification Method Based on Multi-Task Learning in Foggy Scenarios

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper proposes a multi-task learning framework that integrates defogging and re-identification tasks to improve accuracy under foggy conditions. A phase attention mechanism is used to extract weighted feature map information and then combine it with the original image to get vehicle detailed information. It also introduces a multi-scale fusion module for fog removal. 

 - In the abstract, decreased accuracy due to fog removal may be rewritten. McF acronym can be put before the word defogging. Similarly, the full form of mAP?

- Grammatical/typo errors are to be checked in the whole document.

- Vehicle re-identification: Importance, when it is required, and how needs to be highlighted in general at the start of the Introduction. The general introduction needs to be increased so understand the context of the research undertaken.

- At the end of related work, the research gap analysis needs to be highlighted and how this work tackles it.

- Some basic info on ResNet in brief needs to be added in Sec 3.

- The second part of Fig. 2 is not readable.

- Is there is inbuilt mechanism for combining both outputs?

- The algorithm needs to be discussed by using pseudo code/flowchart for a better understanding

- The challenges that may be faced or realistic implementation can be highlighted.

 

Comments on the Quality of English Language

Moderate

Author Response

Dear Reviewers,

Thank you for your insightful comments and suggestions. We appreciate your time and effort in reviewing our manuscript, and we have addressed each issue to improve the clarity and quality of our work.

We have revised the Abstract to more clearly reflect the impact of dehazing on accuracy. The McF (Multi-Scale Fusion) acronym and mAP (Mean Average Accuracy) are now clearly defined at the first appearance to avoid any ambiguity.

We have thoroughly reviewed the document to correct all grammatical and typographical errors.

We have expanded the Introduction to emphasize the importance of vehicle re-identification in modern traffic systems and the challenges it faces in harsh environmental conditions.

At the end of the Related Work section, we provide a detailed analysis of the research gaps and how our proposed framework addresses these gaps.

We have added a brief overview of ResNet in Section 3, explaining its relevance and usefulness in our research to ensure that readers are fully aware of the technical background.

Figure 2 Clarity and Output Mechanism: We have revised Figure 2 to improve readability and included a description of how the outputs of the dehazing and re-identification modules are combined in the text.

To enhance understanding, we present a detailed description of the flowchart in Figure 1. This supplement aims to clarify the operational details of our multi-task learning framework. We discuss potential challenges and practical implementation issues in the conclusion, highlighting how our framework can be adapted for real-world applications. We believe these revisions comprehensively address your concerns and enhance the contribution of the manuscript to the field. We look forward to your feedback and hope that our manuscript will now meet the journal's publication standards. Thank you again for your constructive feedback.

Reviewer 2 Report

Comments and Suggestions for Authors

This paper introduces a multi-task re-identification framework that concurrently performs defogging and vehicle re-identification tasks. The paper is well-structured and logically coherent. The specific review comments are as follows:

1. Classify the vehicle re-identification methods under different scenarios and provide more literature regarding vehicle re-identification under foggy conditions.

2. In Figure 6, the abbreviations of various methods need to be explained in detail when they first appear.

3. Supplement the verification of the proposed method’s transferability, i.e., its performance on other datasets.

4. In the conclusion section, add the limitations of this study and suggestions for future work.

Comments on the Quality of English Language

Good

Author Response

Dear Reviewers,

Thank you for your insightful comments and suggestions. We have carefully addressed each point raised in your feedback to enhance the clarity and quality of our manuscript.

Regarding the transferability of our method to other datasets, we acknowledge that extensive validation across diverse datasets has not been conducted due to the scarcity of vehicle re-identification datasets specific to foggy conditions. This limitation has been explicitly mentioned in the manuscript to ensure transparency about the scope of our testing and the generalizability of our results.

We have incorporated recent studies into our literature review to reflect the latest advancements in the field, ensuring that our references are up-to-date and relevant to the current state of research in vehicle re-identification.

Additionally, we have clarified the abbreviations used throughout the manuscript, including those in Figure 6, to enhance the readability and accessibility of our content for all readers.

Finally, we have refined the conclusion section to more effectively summarize our findings and outline the implications of our research. This revision includes a better synthesis of our results and a clearer presentation of future research directions.

We hope that these revisions adequately address your concerns and that our manuscript now meets the publication standards of your esteemed journal. We appreciate the opportunity to improve our work based on your feedback and look forward to any further suggestions you might have.

Thank you for your consideration.

Best regards,

Reviewer 3 Report

Comments and Suggestions for Authors

• The abstract includes the various artificial intelligence methods adopted in this research work and a description of the data.
• Discuss/describe the advanced algorithms and sophisticated image processing techniques in line number 39 and 41
• In Figure 1, the direction flow work is not mentioned, including the direction arrow from one process to another process.
• Inference from the literature review has to include the end of the related work.
• Font style must be uniform throughout the manuscript, including Figures and Tables.
• Along with existing vocabulary is required for various notation used in the manuscript, for example H, C, W, etc
• Figure 3, each subfigure, what it means include as subheadings.
• ResNet50 network is used in this research; explain the benefits of using ResNet50
• Authors are required to justify for why they did not use real-time datasets. At least for validation of the model, a few images were collected and utilized.
•  Include discussion with numerical input and output values in the conclusion part of the manuscript. Discuss the social relevance of the research in the concluding part.
• Include the latest research works carried out in 2024

Author Response

Dear Editor and Reviewer,

Thank you very much for your valuable comments on our manuscript. Based on your suggestions, we have made the following revisions to the manuscript:

Abstract and Methods Description: We have expanded the Abstract section to describe the AI ​​methods and image processing techniques used in detail.
We have added a specific description to Figure 1 to clarify the process direction.
We have expanded the literature review section to cover the latest research up to 2024 and unified the font style throughout the text.
Clear subtitles and descriptions have been added to each subfigure in Figure 3 to ensure transparency and easy understanding of the results.
We have added a section on the selection of the ResNet50 network, explaining its advantages and its applicability.
We have added a detailed explanation of why no real-time datasets were used.
The article has added a detailed discussion of the numerical inputs and outputs used in the experiments, emphasizing the relevance and significance of these data in the research results.
We have discussed the social relevance of this study in the conclusion and added the latest research in 2024 to ensure that our manuscript reflects the latest trends and progress in the current field.
We believe that these revisions will significantly improve the quality and impact of the manuscript. Thank you again for your valuable suggestions and support.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have extensively revised the paper. However, the second (lower) part of Figure 2 is still not readable. That needs to be changed. 

All my other suggestions are incorporated into the revised version. 

 

Comments on the Quality of English Language

Moderate 

Author Response

Dear Reviewer,

Thank you for your insightful comments and suggestions on our manuscript. In response to your concerns, we have further elaborated on the details of the images as per your suggestions in the latest revision of our manuscript. We have ensured that  image is described with greater accuracy and detail, enhancing the clarity and depth of our presentation.

We appreciate your careful scrutiny and believe that these enhancements have significantly improved the quality of our submission. We hope that our revisions meet your expectations and look forward to your feedback.

Best regards

Reviewer 2 Report

Comments and Suggestions for Authors

None

Comments on the Quality of English Language

Good

Author Response

Dear Reviewer,

Thank you for your valuable feedback and suggestions regarding our manuscript. We have carefully considered and addressed your previous comments in our revised submission. We are pleased to note that there are no further concerns at this time. We appreciate your thorough review and guidance, which have undoubtedly improved the quality of our work. We look forward to your final decision.

Best regards