DriveNet

The DriveNet dataset comprises 11,160 images representing four distinct driving postures. The images were extracted from video recordings of three individuals to ensure robustness and diversity within the dataset. All videos were captured using a Samsung S24 Ultra under identical conditions. The images are stored in .jpg format with a resolution of 1920 × 1440 pixels. The dataset is categorized into four classes, each containing 2,790 images, representing the following driving postures: lean backward, move backward, sit straight, and right position.