AFLW 人脸数据库

AFLW Dataset 是一个包含多姿态、多视角的大规模人脸数据库，其中每张人脸都被标注了 21 个特征点，其包含不同姿态、表情、光照和种族等因素影响的图片，该数据库拥有约 25000 万手工标注的人脸图片（59% 为女性，41% 为男性），其中大部分为彩色图片，也有少部分灰色图片，其适用于人脸识别、人脸检测和人脸对齐等研究，具有很高的研究价值。

The AFLW Dataset is a large-scale face database encompassing multiple poses and viewpoints. Each face contained within it is annotated with 21 facial landmarks. The database comprises images influenced by various factors including diverse poses, expressions, lighting conditions, and ethnicities. It holds approximately 25,000 manually annotated face images, with 59% being female faces and 41% male faces. The majority of these images are color photographs, while a small subset consists of grayscale images. This dataset is suitable for research areas such as face recognition, face detection, and face alignment, and possesses significant research value.

搜集汇总

背景与挑战

背景概述

AFLW人脸数据库是一个包含约25000万手工标注人脸图片的大规模数据集，每张图片标注了21个特征点，涵盖多姿态、多视角、不同表情、光照和种族，适用于人脸识别、检测和对齐等研究。该数据集由格拉茨科技大学于2011年发布，具有重要的研究价值。