On-Road Driver Emotion Recognition Using Facial Expression

Round 1

Reviewer 1 Report

In my opinion, the work can be published as presented. The paper details the background to the study and the methodology and results. The authors have indicated their contribution to the research and have provided a discussion and indicated limitations.  A three-sentence commentary on how the presented method could be used to improve driving safety would be helpful.

The following are comments that, if taken into account, may improve the quality of the paper.
1. Comment on how the presented method can be used to improve driving safety.
2. Please describe in more detail the weaknesses and limitations of the model, e.g., if how the issue of driver's feelings towards the photographs was verified.  It seems that driver attitude and its reflection in facial expression may be an individual issue.
3. Justify your choice of model configuration in more detail and provide more information on the algorithm used and critically compare it with those used in other studies.
4. Provide a comparison of night and day images, was the recognition of the driver's attitude at a similar level?

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This paper presents a model of neural network used to detect faces in the image to recognize the emotions during drive.

1. Present better novelty of your idea. We have many research on deep learning and driving models so discussion on your advances in the field of driving autonomous vehicle and deep learning would be important. Explain also your novelty in relation to other research.
2. Related idea: Deep learning based crowd counting model for drone assisted systems, Lightweight CNN model for human face detection in risk situations, Self-attention negative feedback network for real-time image super-resolution.
3. How was your model in fig. 1 developed? Did you test also other configurations? What type of filtering and dropout was used in research? Can you show results of other configurations to justify your choice?
4. How was your model trained? Which algorithm was used? Make presentation.
5. What is optimal distance of the driver form your camera? How is the lightening influencing result of processing? What are weak points of your model? Does it work good in any situation? Results from night drive would be also very important.
6. Your model need verification on other data collections to show advances of your idea. Make comparisons and discuss results.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

I think it is revised so i suggest to accept