Detecting Distracted Drivers using Transfer Learning and Image Classification Models
Yigao Jin · 2023
Distracted driving is one of the major causes of car accidents worldwide, with specific behaviors such as talking on the phone or texting while driving. In the future, as human-machine co-driving becomes mainstream, monitoring drivers becomes crucial. This involves switching between different driving modes and taking measures when drivers are distracted. Deep learning techniques can effectively accomplish this task. This study focuses on the transfer learning of four image classification models and their application in detecting distracted drivers. These models can predict with varying degrees of accuracy whether a given image displays a distracted driver while driving. The models used in this study are Resnet, Nasnet, Dense, and VGG_with_MobileNet. Multiple optimized models were trained using different adjustment methods, and they all achieved an accuracy of over 93% on the AUC dataset.