CelebAMask-HQ is a dataset repository containing 30,000 high-resolution face images selected from CelebA, each paired with 512x512 segmentation masks covering 19 classes including facial components (skin, nose, eyes, mouth, lips) and accessories (hair, hats, eyeglasses, earrings). The dataset is designed for training and evaluating face parsing, face recognition, face hallucination, and GAN-based face generation and editing algorithms. It provides downloadable resources including image data and corresponding attribute masks.