AFLW2000-3D

该数据集名为ALFW2000，是从ALFW数据集中选取的前2000张图片，并标注了真实的三维人脸以及相应的68个关键点。该数据集包含了光照和遮挡条件存在较大变化的样本，为头部姿态估计任务提供了丰富的变体。规模上，该数据集包含了2000张图片。其所涉及的任务主要是头部姿态估计。

This dataset, named ALFW2000, is selected from the original ALFW dataset by taking the first 2000 images, and is annotated with ground-truth 3D facial models and the corresponding 68 facial landmarks. It contains samples with significant variations in illumination and occlusion conditions, providing rich diversities for the head pose estimation task. In terms of scale, this dataset includes a total of 2000 images. The primary task associated with this dataset is head pose estimation.