Classification of Distracted Driving Using Transfer Learning and Deep Neural Network

Distracted driving is a significant contributor to traffic accidents.In order to improve road safety, it is critical to not only detect instances of driver distraction but also to identify the core causes of these distractions.In this study, description about a complete technique to classifying instances of distracted driving that fully incorporates transfer learning technology.Our process is predicated on a precisely produced dataset that has been thoroughly annotated to assure ac-curacy.This dataset is augmented further, and feature extraction is carried out using a wide selection of transfer learning models, including VGG16, Resnet50, Inception, Densenet, and Xception.Following that, the collected features are fed into a DDD classifier which classifies and identifies distraction types.Our experimental results indisputably show that the DNN model, after feature extraction via the Resnet50 transfer learning model, outperforms all other models in the context of distracted driving classification.