Driver Behavior Image Dataset

High-resolution images in the dataset were collected across different lighting conditions and vehicle types, representing diverse driving situations. The images, presented in raw, unannotated form, allow for flexible pre-processing by machine learning practitioners. The dataset is organized into five behavioral classes:

Safe Driving: Images show drivers fully attentive to the road, either with both hands on the steering wheel or one hand on the wheel and the other on the gear stick, exemplifying distraction-free driving.

Turning: Photos capture drivers as they turn, moving their heads or bodies to indicate a change in direction. This behavior reflects the driver’s engagement in typical actions like steering.

Texting on Phone: Images of drivers using mobile phones to type messages or interact with screens. This class addresses one of the primary causes of distracted driving.

Talking on Phone: Drivers are depicted holding their phones to their ears or speaking on the phone while driving. This class helps to identify the distractions associated with phone conversations.

Others: Includes images of behaviors that compromise safe driving, such as drinking water, conversing with passengers in the back seat, or showing signs of drowsiness.

With its wide-ranging real-life conditions, this dataset offers a rich resource for developing systems to detect and classify risky driving behaviors, supporting efforts to improve road safety.