Hyperparameter Tuned Deep Autoencoder Model for Road Classification Model in Intelligent Transportation Systems

Round 1

Reviewer 1 Report

In this paper, a novel MODAE-RCM algorithm was proposed and automated road classification. This study is interesting, but needs to be further revised before publication. 

(1) Why use NASNet as the feature extraction network? As far as I know, there are other well-known image feature extraction networks (e.g., VGG, Resnet, etc.)

(2) In the method part, the author explains the principle of DAE. However, it is necessary to further clarify the role of DAE in this paper, and what is the input and output of DAE.

(3) The author should add the calculation formula of performance evaluation indicator in the method part.

(4) This paper seems to have no contribution from the UAV field. It is suggested to modify the keywords.

(5) According to the publication standard of MDPI, ‘Fig. #’ in the manuscript should be modified to ‘Figure #’.

Author Response

In this paper, a novel MODAE-RCM algorithm was proposed and automated road classification. This study is interesting, but needs to be further revised before publication.

(1) Why use NASNet as the feature extraction network? As far as I know, there are other well-known image feature extraction networks (e.g., VGG, Resnet, etc.)

Thank you for the comment. We have chosen NASNet feature extractor over other DL architectures due to the following reasons:

• Neural Architecture Search (NAS) automates network architecture engineering. Basically the idea was to search the best combination of parameters of the given search space of filter sizes, output channels, strides, number of layers, etc.
• It searches for the best algorithm to achieve the best performance on a certain task.
• Commercial services such as Google’s AutoML and open-source libraries such as Auto-Keras make NAS accessible to the broader machine learning environment.
• Gains interest due to better, faster, and more cost-efficient NAS methods.

 (2) In the method part, the author explains the principle of DAE. However, it is necessary to further clarify the role of DAE in this paper, and what is the input and output of DAE.

As per the reviewer comment, necessary information related to the role of DAE is given in the revised manuscript. Kindly refer Page 7, Section 3.2, Lines 167-170.

 (3) The author should add the calculation formula of performance evaluation indicator in the method part.

As per the reviewer comment, the experimental details are provided in the revised manuscript. Kindly refer Page 11, Lines 244-258.

 (4) This paper seems to have no contribution from the UAV field. It is suggested to modify the keywords.

Based on the reviewer comment, we have provided the contribution in the revised manuscript and keywords are properly given in the revised manuscript. Kindly refer Pages 1 and 2.

 (5) According to the publication standard of MDPI, ‘Fig. #’ in the manuscript should be modified to ‘Figure #’.

Based on the reviewer comment, we have changed Fig. into Figure in the revised manuscript.

Reviewer 2 Report

the paper is well written and well presented, the idea, the contribution, and the quality is good, however minor corrections are needed as follow:

1- state clearly the contributions of the study in introduction section as contribution 1 , contribution 2 , and so on 

2-it is recommended to write the algorithm in algorithmic manner, ( use intended )

3-enrich the paper by citing recent published work in  the filed of road detection 

Murad, N. M., Rejeb, L. ., & Ben Said, L. . (2022). The Use of DCNN for Road Path Detection and Segmentation. Iraqi Journal For Computer Science and Mathematics, 3(2), 119–127. https://doi.org/10.52866/ijcsm.2022.02.01.013

4- add numerical results to conclusion 

Author Response

the paper is well written and well presented, the idea, the contribution, and the quality is good, however minor corrections are needed as follow:

1- state clearly the contributions of the study in introduction section as contribution 1 , contribution 2 , and so on

Based on the reviewer comment, we have provided the contribution in the revised manuscript. Kindly refer Page 3, Paragraph 2, Lines 67-80.

2-it is recommended to write the algorithm in algorithmic manner, ( use intended )

Based on the reviewer comment, we have clearly described the algorithm in the revised manuscript.

3-enrich the paper by citing recent published work in  the filed of road detection

Murad, N. M., Rejeb, L. ., & Ben Said, L. . (2022). The Use of DCNN for Road Path Detection and Segmentation. Iraqi Journal For Computer Science and Mathematics, 3(2), 119–127. https://doi.org/10.52866/ijcsm.2022.02.01.013

Based on the reviewer comment, the recent references are included in the revised manuscript. Kindly refer Page 23, Reference 10.

4- add numerical results to conclusion

Based on the reviewer comment, the experimental values are included in the conclusion section.  Kindly refer Page 22, Section 5, Lines 345-347.