CelebAMask-HQ provides the parsing map of images in CelebA-HQ down-sampled to 512 × 512, where pixel-level annotation of 19 classes, including facial components and accessories, are provided.
BibTex:
Before browse our site, please accept our cookies policy
