This repo presents the code for distracted driver detection and classification with deep Convolutional Neural Network (CNN) network on AUC dataset. Distracted driver detetion is one of the safety measures in Advanced Driver Assistance Systems (ADAS) to take countermeasures and enable safe driving. In realtime applications like ADAS, the algorithm not only has to be accurate but also efficient in terms of memory and speed. Hence, we focused on developing computationally efficient CNN while maintaining good accuracy. The proposed mobileVGG architecture has only 2.2 Million parameters.
The American University in Cairo (AUC) distracted driver detection dataset defines ten postures of the driver to detect: Safe driving and nine distracted behaviors i.e. talking on a cellphone with the left or right hand, texting using cellphone with the left or right hand, eating or drinking, reaching behind, hair and makeup, adjusting the radio and talking to passenger.
If you find this research interesting, please cite our paper(s):
[1] B. Baheti, S. Talbar and S. Gajre, "Towards Computationally Efficient and Realtime Distracted Driver Detection With MobileVGG Network," in IEEE Transactions on Intelligent Vehicles, vol. 5, no. 4, pp. 565-574, Dec. 2020, doi: 10.1109/TIV.2020.2995555.
[2] B. Baheti, S. Gajre and S. Talbar, "Detection of Distracted Driver Using Convolutional Neural Network," 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2018, pp. 1145-11456, doi: 10.1109/CVPRW.2018.00150.