ICACT20210194 Question.1
Questioner: baniyab@gram.edu 2021-02-09 오후 2:24:31 |
ICACT20210194 Answer.1
Answer by Auhor satyabrataaich@gmail.com 2021-02-09 오후 2:24:31
Chrome  Click!! |
| Thank you very much for submitting your research article and welcome to IEEE/ ICACT2021.
Can you please explain your contribution in this article? |
Thank you for the question.According to the United States National Highway Traffic Safety Administration (NHTSA), the main cause of road fatalities and injuries is distracted drivers. Driver distraction is a specific sort of driver interference on the roadway. In this case, a deep learning-based system can detect and distinguish the source of distractions in real-time, to avoid traffic accidents and make better transport safety. This research attempts to develop the system using transfer learning and fine-tuning methods with different model architectures. Various pre-trained weights were used with fine-tuning to improve accuracy and implemented using Mobile Net, VGG16, and ResNet50 models. Finally, the results illustrate that transfer learning on Mobile Net with frozen layers is the best model out of three models getting 99% (495 out of 500) of accuracy on the test dataset images. |
ICACT20210194 Question.2
Questioner: joomi@inje.ac.kr 2021-02-09 오후 2:48:42 |
ICACT20210194 Answer.2
Answer by Auhor satyabrataaich@gmail.com 2021-02-09 오후 2:48:42
Chrome Click!! |
| Thank you so much for the slides. How transfer learning is beneficial compared to the other techniques? |
Thank you for the question.Transfer learning is a deep learning technique that stores the knowledge gained while solving one issue and use it to a new but related problem.Most of the model originally trained on the ImageNet database which can classify images into thousands of object categories. These pretrained models helps to provide better accuracy and save computational time when we work on new datasets. |
ICACT20210194 Question.3
Questioner: yfhou@okayama-u.ac.jp 2021-02-10 오전 11:26:21 |
ICACT20210194 Answer.3
Answer by Auhor satyabrataaich@gmail.com 2021-02-10 오전 11:26:21
Chrome Click!! |
| Thank you for your good work. Could tell us how many data samples have been used for traing stage? Thanks |
Thank you for the question. We have used 17,544 Images for training. |
ICACT20210194 Question.4
Questioner: franklinlu888@outlook.com 2021-02-10 오후 12:55:03 |
ICACT20210194 Answer.4
Answer by Auhor satyabrataaich@gmail.com 2021-02-10 오후 12:55:03
Chrome Click!! |
| Is it possible to get better accuracy in your study? |
Thank you for the question. Yes to be honest , there will be always an option for improvement. In our study we have used all the recent techniques and found 99% accuracy. |
ICACT20210194 Question.5
Questioner: fhamza@gatech.edu 2021-02-10 오후 2:24:34 |
ICACT20210194 Answer.5
Answer by Auhor satyabrataaich@gmail.com 2021-02-10 오후 2:24:34
Chrome Click!! |
| Great work! In your assessment or opinion, what types of distracted driving scenarios were under-represented in the images? |
Thank you for the question!In my opinion all the scenarios are well addressed in the dataset. Any scenario other than safe driving is the distracted driving scenario. Although we can add few more distracted driving scenarios as per our way of thinking,however the dataset almost covers most of the distracted driving scenarios. |
ICACT20210194 Question.6
Questioner: mihai.sima@ieee.org 2021-02-10 오후 5:21:17 |
ICACT20210194 Answer.6
Answer by Auhor satyabrataaich@gmail.com 2021-02-10 오후 5:21:17
Chrome Click!! |
| How exactly would you define a "distracted driver?" In Slide 8 you do present a number of examples, but in the end I guess you should refer to the driver response time to road events (long response time would mean that the driver is distracted). Have you considered this? |
Thank you for the question. To be honest we did not think about it. But it is a good suggestion. We will think about it. |
