LaPa

该数据集名为LaPa，是一个大规模的人脸解析数据集，包含了超过22,000张面部图像，并且每张图像都附有像素级的标签图和106个关键点标记。这些图像均提供了包含11个类别的像素级标签图，使得这项数据集在人脸解析任务中具有显著的大规模优势。

This dataset, named LaPa, is a large-scale face parsing dataset. It contains over 22,000 facial images, each accompanied by a pixel-level label map and 106 keypoint annotations. The pixel-level label maps cover 11 categories, which endows this dataset with significant scale advantages in the face parsing task.