Railway Obstacle Intrusion Detection Based on Convolution Neural Network Multitask Learning

Round 1

Reviewer 1 Report

I have read manuscript with great attention and interest. The article deals with an interesting topic of use of deep learing in transport. This topic is very needed. The application is interesting and I like your article. In general, you have a proper academic way of referring and a good language. I really like the introduction and state of the art sections.

Congratulations to the authors of the work.

Comments and suggestions:

1. It is not clear to me how the system would be used in practice. Would it alert the driver? Is it realistic that this will reduce the accident rate and the train will be able to brake in time?

2. Could you provide brief introduction to the Convolution Neural Networks? I think that this can help to the readers.

3. What are the limitations of your system?

4. Do you think that this method can be used in teaching of CNN? I think that this application can be atrractive for students.

Author Response

We gratefully thanks for the precious time the reviewer spent making constructive remarks.  In the attachment , we summarize our responses to each comment from the reviewer. Thanks again for your review.

Author Response File: Author Response.pdf

Reviewer 2 Report

Railway Obstacle Intrusion Detection Based on Convolution Neural Network Multi-Task Learning

I would like to highlight the following major suggestions.

1. Starting from the introduction part, unable to see any intext citations in the complete introduction section?

2. Authors are suggested to make tables of findings in each sections 2.1, 2.2 and also 2.3. A comparison of the proposed approach with the available literature is suggested.

3. In line 104, see [6-10], it should be illustrated individually in place of collectively. Again line 131 see [16-19] & [20-23] should be highlighted individually.

4. Section 3 methodology section should have flow chart.

5. figure 1 needs revision.

6.  Equations need upgradation with the help of mathematical equation editors.

7. Figure A1 is confusing and needs extensive revision.

8. Complexity analysis is required for comparing the proposed framework with existing approaches.

9. Result section needs more performance parameters.

Author Response

We gratefully thanks for the precious time the reviewer spent making constructive remarks.  In the attachment , we summarize our responses to each comment from the reviewer. Thanks again for your review.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Railway Obstacle Intrusion Detection Based on Convolution Neural Network Multi-Task Learning

Now the manuscript is accessible, I still believe some enhancement in the existing work.

I would like to suggest following enhancement in the manuscript.

1. Methodology section must have flow chart with proper notations for better clarity to the readers.

2. In the related work section-2. Please add  table and point wise description for each heading such as  Track Obstacle Detection, Lane Detection, Multi-task Learning etc. 

3. Please improve in-text citation for example in introduction part.

 

Author Response

Thank you so much for your careful check. In the attachment , we summarize our responses to each comment from the reviewer. Thanks again for your review.

Author Response File: Author Response.pdf