Deep CNN models for Driver Activity Recognition for Intelligent Vehicles

2020 International Journal of Emerging Trends in Engineering Research  
This paper aims to ensure safety while driving driver choices and driver decisions are fundamental variables that can affect safe driving. A recognition system for driver operation is designed to recognize driver behaviors that are based on profoundly convolutionary neural networks (CNN). Typical driving habits, such as normal driving, right-mirror testing, rear-mirror checking, left-mirror verifying, media change, passenger speaking, text and cell phone responding, swapping signs, smoking,
more » ... -up, etc. The first four actions are usual driving actions and the remaining behavior is driving diversion. The Gaussian Mixture Model (GMM) will be used as an input to the proposed model in handling the images like segmentation.CNN models are prepared for the function of binary detection and determine whether or not the driver is being disturbed. Additionally, we propose a deep learning-based accuracy Achieved by the binary detection rate of 91.4 percent.
doi:10.30534/ijeter/2020/828102020 fatcat:z4pwmyv7kzdlxkuiye7w4khyce