Zheng0309/DomainNet

DomainNet是一个大规模图像分类数据集，包含345个不同类别的物体和场景，如飞机、汽车、动物等。数据集覆盖六个域：剪贴画、信息图、绘画、快速绘图、真实照片和素描，旨在支持多域学习和零样本图像分类任务。它分为训练集（409,832个示例）和测试集（176,743个示例），总大小约1.57GB，下载大小约18.5GB。该数据集常用于计算机视觉研究，特别是在领域泛化和迁移学习方面。DomainNet is a large-scale image classification dataset containing 345 distinct categories of objects and scenes, such as aircraft, cars, and animals. The dataset spans six domains: clipart, infograph, painting, quickdraw, real photos, and sketches, designed to support multi-domain learning and zero-shot image classification tasks. It is divided into a training set (409,832 examples) and a test set (176,743 examples), with a total size of approximately 1.57GB and a download size of about 18.5GB. This dataset is commonly used in computer vision research, particularly in domain generalization and transfer learning.