| الملخص: | Malaysia has one of the highest traffic fatality rates in the world. The main cause towards the increment of annual rate on traffic accident in Malaysia is due to distracted driver on wheel. Due to the advancement and integration of technologies within the society, drivers tend to get distracted either by their devices or infotainment that is build-in with the vehicles. This paper presents the application of deep learning to classify driver's distracted behavior behind the wheel. This paper implements deep convolution neural network to classify driver's distracted behavior behind the wheel. The experiment was conducted to classify drowsiness dataset of 10 classes from State Farm and 2 classes from National Tsing Hua University (NTHU). Fast and accurate models for driving behavior classification are desired for real-world deployment and application in vehicle system. The results of this investigation show that MobileNetV2 outperforms other models, presenting a good balance between accuracy and processing runtime for real-world deployment. © 2022 IEEE.
|
|---|