AgroBench (Agronomist AI Benchmark)
收藏资源简介:
AgroBench是一个用于评估视觉语言模型(VLM)在农业领域应用的数据集,由日本国立先进工业科学技术研究院(AIST)等机构创建。该数据集包括203种作物类型、682种疾病类别、134种害虫类别、108种杂草类别,以及与作物管理相关的98种机械类别和77种传统管理方法。数据集包含专家标注的问答对,涵盖了农作物病虫害识别、作物管理、传统农业方法等多个领域。
AgroBench is a dataset designed to evaluate the application of vision-language models (VLMs) in the agricultural field, created by the National Institute of Advanced Industrial Science and Technology (AIST) of Japan and other institutions. This dataset covers 203 crop types, 682 disease categories, 134 pest categories, 108 weed categories, as well as 98 machinery categories and 77 traditional management methods related to crop management. The dataset contains expert-annotated question-answer pairs, covering multiple areas such as crop disease and pest identification, crop management, and traditional agricultural practices.




