iNatAg
收藏资源简介:
iNatAg是一个大规模的图像数据集,包含超过470万张图片,涵盖2959种不同的农作物和杂草。该数据集由iNaturalist数据库中的公开数据经过精心筛选而来,旨在为农业应用服务。数据集包含精确的物种分类注释,以及从二元农作物/杂草标签到具体物种标签的税务分类层次。该数据集具有广泛的地理覆盖范围,反映了自然图像捕捉和环境变化的多样性,可为构建健壮的、具有地理位置意识的农业分类系统提供新的基础。
iNatAg is a large-scale image dataset containing over 4.7 million images covering 2959 distinct crop and weed species. This dataset is curated from publicly available data sourced from the iNaturalist database, and is developed specifically for agricultural applications. It features accurate species classification annotations, as well as a taxonomic hierarchy ranging from binary crop/weed labels to fine-grained species-level labels. Boasting extensive geographic coverage, it reflects the diversity of natural image capture scenarios and environmental variations, providing a novel foundation for building robust, geospatially-aware agricultural classification systems.




