DrowsyFace Profiles

The dataset includes three distinct subsets of facial images: (1) grayscale images, (2) lightly processed images, and (3) images that have only undergone region of interest (ROI) extraction. In all cases, facial regions were isolated to ensure a consistent focus on relevant features. The dataset includes facial representations in different profiles, capturing visual indicators associated with drowsiness, as well as control samples that show no signs of drowsiness. Furthermore, each subset is accompanied by annotations in the Pascal VOC format, which provides structured metadata for object detection and localization tasks. This dataset is intended for research in computer vision, machine learning, and human fatigue detection.本数据集包含三类互不相同的人脸图像子集：(1) 灰度图像；(2) 轻度处理图像；(3) 仅经过感兴趣区域（Region of Interest，ROI）提取的图像。所有样本均已完成人脸区域提取，以确保统一聚焦于相关视觉特征。本数据集涵盖多姿态人脸图像，包含与困倦状态相关的视觉特征样本，同时配备无困倦迹象的对照样本。此外，各子集均附带Pascal VOC格式的标注文件，可为目标检测与定位任务提供结构化元数据。本数据集旨在支撑计算机视觉、机器学习及人类疲劳检测领域的相关研究。