Distinctive Human Appearances (DHA) dataset

Distinctive Human Appearances (DHA) 数据集由史蒂文斯理工学院商学院创建，包含1000张具有独特外观的人脸图像，旨在评估人脸检测算法在处理多样性外观时的性能。数据集涵盖了多种外观特征，如遗传疾病、严重事故受害者、独特时尚趋势追随者等。创建过程中，研究者通过网络搜索和文献调研收集了这些低频外观的图像。DHA数据集的应用领域包括人脸检测技术的公平性和准确性评估，旨在解决现有数据集在外观多样性上的不足。

The Distinctive Human Appearances (DHA) dataset was created by the School of Business at Stevens Institute of Technology. It comprises 1,000 facial images with unique human appearances, and is designed to evaluate the performance of face detection algorithms when handling diverse appearance variations. The dataset covers a wide range of appearance traits, including individuals with genetic disorders, victims of severe accidents, and followers of distinctive fashion trends, among others. During the dataset creation process, researchers collected these low-prevalence appearance images through web searches and literature reviews. Application fields of the DHA dataset include fairness and accuracy assessments of face detection technologies, aiming to address the shortage of appearance diversity in existing datasets.