In: Electrical Engineering
What is the Conclusion of Drowsiness Detection for Driver State Monitoring Using Deep Neural Networks ?
By seeing and evaluating the different scenarios ,we saw that eyes are critical for drowsiness classification in any circumstances.Mainly the efficiency of model drops in the cases of wearing sunglasses because algorithm is not able to detect the drivers eyes.
The results also shown that other important factors is luminosity beacuse the error rate rise by 6% with rise in the luminosity.
By using the deep neural networks it has been found that the accuracy results is almost equal to 81%.
When we see the model size,it's complexity and storage ,there is some reduction when face landmarks coordination is used to detect the drivers drowsiness.
Category:1- with glasses ,the accuracy is 84.848.
2- Night without glasses,the accuracy is 81.40
3- Night with glasses ,the accuracy is 76.142
4- Without glasses,the accuracy is 88.362
5-With glasses,the accuracy is 76.30
Overall the accuracy was 89.67%.