WeedsGalore
收藏资源简介:
WeedsGalore数据集是由德国地球科学研究中心创建的多光谱、多时相无人机(UAV)影像数据集,包含RGB、红边、近红外五个波段的数据,针对玉米田中的作物和杂草进行像素级别的语义和实例分割标注。该数据集在德国波茨坦的Marquardt地区采集,包含不同生长阶段的玉米和四种杂草类别,是目前首个公开的多光谱无人机杂草监测数据集,提供了比常规RGB数据集多两个波段的信息,具有前所未有的杂草密度。数据集旨在推进细粒度杂草识别方法和运营系统的研究,增强无人机杂草管理的鲁棒性和适用性。
The WeedsGalore dataset is a multi-spectral and multi-temporal unmanned aerial vehicle (UAV) imagery dataset developed by the German Research Centre for Geosciences (GFZ). It includes five spectral bands, namely RGB, red edge, and near-infrared (NIR), with pixel-level semantic and instance segmentation annotations for crops and weeds in maize fields. Collected in the Marquardt region near Potsdam, Germany, the dataset covers maize at different growth stages and four weed categories, and is the first publicly available multi-spectral UAV dataset for weed monitoring. It provides two additional spectral bands compared to conventional RGB-only datasets and features unprecedented weed density. This dataset aims to advance research on fine-grained weed recognition methods and operational systems, as well as improve the robustness and applicability of UAV-based weed management.




