Distracted Driver Detection using Stacking Ensemble

Distracted driving is one of the primary causes of car crashes. While driving the vehicle, drivers frequently perform secondary activities that distract driving. A decrease in driver distraction is a critical aspect of the smart transportation system. To decrease accidents and improve safety, this paper proposes a distracted driver detection system that classifies various types of distracted activities using ensemble techniques. Different convolutional networks had been trained on images by eliminating the final layer to get there feature vectors. By using the stacking ensemble technique, we stack all the feature vectors to train it on a convolutional network. This stacking technique, which is used to detect the distracted driver posture, achieves 97% accuracy. The study shows how models predict the desired classes. The model proposed in this paper can be used in a real- time environment to detect activities done by the driver.