GAZEONROAD Dataset part 1.zip

This dataset contains high-resolution images of 15 male drivers aged 25-40, captured in indoor car scenes. It is designed for training and validating models for face detection, head pose detection, and gaze estimation. The dataset includes variations in facial appearance such as facial hair, glasses, and caps. With 80% of the images allocated for training and 20% for testing, totaling 87,025 images, it also features seven classes of driver head poses and images of eyes at different orientations to enhance accuracy.

搜集汇总

背景与挑战

背景概述

该数据集包含15名25-40岁男性驾驶员在室内车场景中的高分辨率图像，用于面部检测、头部姿势检测和视线估计模型的训练与验证。数据集包含87,025张图像，分为80%训练集和20%测试集，涵盖不同面部特征（如胡须、眼镜、帽子）和七种头部姿势类别。