Herbarium Half-Earth dataset

Herbarium Half-Earth数据集是由苏黎世联邦理工学院等机构创建，包含超过250万张植物标本图像，涵盖近64,500个分类群，主要来自美洲和太平洋地区。该数据集旨在通过高分辨率图像自动识别植物分类群，解决现有数据集规模小、多样性不足的问题。数据集内容丰富，包括从种级到科级的分类标签，支持多级分类研究。创建过程中，数据集通过标准化标签和图像预处理，确保数据质量和可用性。该数据集广泛应用于植物学研究，特别是在生物多样性监测、气候变化影响评估等领域。

The Herbarium Half-Earth Dataset was developed by institutions including ETH Zurich and other collaborating organizations. It contains over 2.5 million high-resolution plant specimen images, covering nearly 64,500 plant taxa, with most specimens sourced from the Americas and the Pacific region. This dataset is designed to enable automatic plant taxon identification using high-resolution imagery, addressing the limitations of small scale and inadequate taxonomic diversity in existing datasets. It features comprehensive annotation labels spanning from species to family taxonomic ranks, supporting multi-level taxonomic research. During the dataset construction process, standardized labeling and image preprocessing workflows were implemented to guarantee data quality and usability. The Herbarium Half-Earth Dataset has been widely adopted in botanical research, especially in areas such as biodiversity monitoring and climate change impact assessment.