Next Article in Journal
Artificial Intelligence for Quality Defects in the Automotive Industry: A Systemic Review
Previous Article in Journal
Evaluation of Pulse Oximetry Accuracy in a Commercial Smartphone and Smartwatch Device During Human Hypoxia Laboratory Testing
 
 
Article
Peer-Review Record

Deep Learning-Based Algorithm for Road Defect Detection

Sensors 2025, 25(5), 1287; https://doi.org/10.3390/s25051287
by Shaoxiang Li and Dexiang Zhang *
Reviewer 1: Anonymous
Reviewer 2:
Reviewer 3: Anonymous
Sensors 2025, 25(5), 1287; https://doi.org/10.3390/s25051287
Submission received: 14 January 2025 / Revised: 18 February 2025 / Accepted: 18 February 2025 / Published: 20 February 2025
(This article belongs to the Section Fault Diagnosis & Sensors)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

After carefully reviewing the paper titled "RepGD-YOLOV8W: An Improved Model for Road Defect Detection," I have identified the following issues regarding writing, content, and logic that warrant rejection of the manuscript.


1. The paper contains numerous instances where the writing is unclear and imprecise, making it difficult for the reader to understand the authors' intended meaning. For example, in the Introduction section, the paper mentions "challenges remain, such as insufficient detection precision for small targets and issues of missed or false detections in complex backgrounds," but it fails to provide a clear and concise explanation of why these challenges are significant or how they affect the field of road defect detection. This lack of clarity undermines the paper's overall readability and comprehension.

2. The authors propose an improved YOLOv8-based model called RepGD-YOLOV8W, but they do not adequately justify their methodological choices. For instance, the paper introduces the GD mechanism and the Rep-C2f module without explaining why these components were selected or how they are superior to alternative approaches. This lack of justification weakens the paper's argument and raises questions about the validity of the proposed method.

3. The paper uses inconsistent and confusing terminology throughout, which hinders the reader's ability to follow the authors' reasoning. Terms such as "Rep-GD module," "C2f-RepViTBlock," and "Wise-IoU loss function" are introduced without clear definitions or explanations, and they are often used interchangeably in ways that are not always clear. This terminological confusion undermines the paper's coherence and readability.
4. The paper presents experimental results to evaluate the performance of the proposed model, but the evaluation is inadequate and the comparison with other algorithms is not comprehensive. The authors compare RepGD-YOLOV8W with mainstream target detection algorithms such as Faster R-CNN, SSD, and various YOLO versions, but they do not provide a detailed analysis of the strengths and weaknesses of these algorithms in the context of road defect detection. Additionally, the paper fails to discuss potential limitations or biases in the experimental results, which raises questions about the generalizability and reliability of the findings.

5. The paper contains logical inconsistencies and gaps in reasoning that undermine the authors' arguments. For example, the paper claims that the Rep-GD module enhances the feature extraction capability of the model while avoiding a significant increase in the number of parameters, but it does not provide a clear explanation of how this is achieved or why it is important. Similarly, the paper introduces the Wise-IoU loss function to improve the model's sensitivity to edges and fine targets, but it does not adequately demonstrate how this loss function addresses the problems of insufficient gradient and low-quality anchor frames in road defect detection tasks. These logical inconsistencies and gaps in reasoning weaken the paper's overall persuasive power and credibility.

Comments on the Quality of English Language

The paper contains numerous instances where the writing is unclear and imprecise, making it difficult for the reader to understand the authors' intended meaning. For example, in the Introduction section, the paper mentions "challenges remain, such as insufficient detection precision for small targets and issues of missed or false detections in complex backgrounds," but it fails to provide a clear and concise explanation of why these challenges are significant or how they affect the field of road defect detection. This lack of clarity undermines the paper's overall readability and comprehension.

Author Response

Please see the attachment. 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have proposed a deep learning-based algorithm for road-defect detection. Overall, the work is good. However, the paper needs thorough English correction and proof reading.

A few suggestions to the authors:

1)     All the acronyms must be expanded before they’re first used in the text.

2)     Nothing is clear from Figures 1 and 2.

3)     More Figures should be provided to show the efficiency of the proposed algorithm.

4)     Performance of the algorithm should be compared with the state-of-the-art methods.

Comments on the Quality of English Language

The authors have proposed a deep learning-based algorithm for road-defect detection. Overall, the work is good. However, the paper needs thorough English correction and proof reading.

A few suggestions to the authors:

1)     All the acronyms must be expanded before they’re first used in the text.

2)     Nothing is clear from Figures 1 and 2.

3)     More Figures should be provided to show the efficiency of the proposed algorithm.

4)     Performance of the algorithm should be compared with the state-of-the-art methods.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Comments to the Author(s)

This manuscript presents

 Deep Learning Based Algorithm for Road Defect Detection. This paper contains a good effort related to the monitoring of highway infrastructures. The topic is original. The methodology of the study is well explained. In general, the manuscript is organized well. The study offers reliable findings and is supported by sufficient proof.  It could be accepted for publication if the authors resolve the following issues. All the answers should be included in the manuscript.

1.      Please put all the reference numbers in quotation signs (e.g., [1]).

2.      Add at least 10 recent relevant references to the introduction.

3.      Add a Figure to show the common highway diseases from the used RDD2022 dataset.

4.      Add more details to the captions of Fig.1 and Fig. 2.

5.      Add some references to the equations of section 2.1.1 and 2.1.2.

6.      Section 3, the data was divided into a training set and validation set according to the ratio of 7:3. How did you test the system without a test dataset?

7.      Figure 5. Can the RepGD-YOLOV8W label of highway diseases? Please clarify that.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors
  1.  The paper mentions that the processed dataset is divided into training, validation, and test sets. However, it would be helpful to explicitly state the exact number of images in each set or the proportion used for validation and testing. This information is crucial for reproducibility and understanding the robustness of the model evaluation.
  2. While the paper includes visual results of the model's performance, it would be beneficial to add more detailed visualizations, such as precision-recall curves or confusion matrices, to provide a more comprehensive view of the model's performance across different classes of road defects. This would help readers better understand the model's strengths and areas for improvement.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Back to TopTop