Image2Biomass Pasture Innovation Challenge
收藏资源简介:
该数据集包含了从澳大利亚19个地点收集的1162张标注过的牧草俯视图,这些图像跨越多个季节,包括一系列温带牧草种类。每张图像捕捉了一个70cm × 30cm的正方形区域,并与地面测量数据配对,包括按成分分类的生物量(绿色、死亡和豆科植物分数)、植被高度以及来自主动光学传感器(AOS)的归一化植被指数(NDVI)。数据的多维特性结合了视觉、光谱和结构信息,为推进精准放牧管理技术的应用开辟了新的可能性。该数据集在Kaggle竞赛中发布和托管,挑战国际机器学习界进行牧草生物量估计的任务。
This dataset contains 1,162 labeled top-down images of pasture grass collected from 19 locations across Australia, spanning multiple growing seasons and encompassing a diverse array of temperate pasture species. Each image captures a 70 cm × 30 cm square area and is paired with ground-based measurements, including biomass categorized by component (green, dead, and legume fractions), vegetation height, and Normalized Difference Vegetation Index (NDVI) derived from an Active Optical Sensor (AOS). The multi-dimensional dataset combines visual, spectral, and structural information, creating new opportunities to advance the deployment of precision grazing management technologies. This dataset was published and hosted as part of a Kaggle competition, challenging the global machine learning community to address the task of pasture biomass estimation.




