FFHQ-Wrinkle

该数据集名为FFHQ-Wrinkle，包含了人脸图像及其相应的皱纹遮罩对。其中，包括了1,000张由人工标注的图像和50,000张无需人工干预生成的弱标注图像。该数据集涵盖了各种肤质条件的多样化个体，非常适合用于训练能够应对广泛临床场景的模型。规模上，数据集包含了1,000张人工标注图像和50,000张弱标注图像，其任务是对面部皱纹进行分割。

This dataset is named FFHQ-Wrinkle, which contains pairs of facial images and their corresponding wrinkle masks. It includes 1,000 manually annotated images and 50,000 weakly labeled images generated without manual intervention. This dataset covers diverse individuals with various skin conditions, making it highly suitable for training models that can handle a wide range of clinical scenarios. In terms of scale, the dataset contains 1,000 manually annotated images and 50,000 weakly labeled images, and its task is facial wrinkle segmentation.

搜集汇总

背景与挑战

背景概述

FFHQ-Wrinkle是FFHQ数据集的扩展，专注于面部皱纹特征，包含1,000张手动标注的皱纹掩码和50,000张自动生成的弱标签皱纹掩码，所有图像分辨率为1024x1024。该数据集旨在支持面部识别、老化模拟等研究，并提供了预训练模型和推理工具。