Young Labeled Faces in the Wild (YLFW)

YLFW数据集是由科英布拉大学系统与机器人研究所开发的，专注于儿童面部识别的标准化基准。该数据集包含约10k图像，涵盖约3k不同身份的儿童面部图像，分为YLFW-Benchmark和YLFW-Dev两个部分，分别用于测试和训练。YLFW旨在解决现有面部识别数据集中对成人偏见的问题，特别是在寻找失踪或被绑架儿童以及防止儿童被剥削的场景中具有重要应用。数据集通过网络爬虫收集，确保了非洲、亚洲、高加索和印度的种族平衡，适用于多种面部识别算法的开发和评估。

The YLFW dataset is a standardized benchmark for child face recognition, developed by the Institute of Systems and Robotics of the University of Coimbra. It contains approximately 10,000 images of children’s faces from roughly 3,000 distinct identities, and is divided into two subsets: YLFW-Benchmark and YLFW-Dev, which are dedicated to testing and training respectively. YLFW is designed to mitigate the adult bias issue existing in mainstream face recognition datasets, and holds significant practical value in scenarios such as locating missing or abducted children and preventing child exploitation. Collected via web crawlers, the dataset ensures racial balance across African, Asian, Caucasian, and Indian populations, and is applicable to the development and evaluation of various face recognition algorithms.