Horama/wow_scraped

Wildlife Scraped数据集是一个多源野生动物图像数据集，涵盖122个标签（物种、亚种和一些家养类别）。该数据集包含来自iNaturalist、Wikimedia Commons、Wikipedia、Flickr和DuckDuckGo的139,231张图像，分为训练集（70%）、验证集（15%）和测试集（15%）。每张图像都附有元数据，如标签、物种、来源、许可证和作者。数据集结构便于加载和过滤，适用于图像分类任务。The Wildlife Scraped dataset is a multi-source wildlife image dataset covering 122 labels (species, sub-species and a few domestic categories). It includes 139,231 images from iNaturalist, Wikimedia Commons, Wikipedia, Flickr, and DuckDuckGo, split into training (70%), validation (15%), and test (15%) sets. Each image is associated with metadata such as label, specie, source, licence, and author. The dataset is structured to facilitate easy loading and filtering and is designed for image classification tasks.