2018年用于智能农业的密集语义杂草分类多光谱图像和MAV数据集

weedNet数据集来源于一个受控的甜菜田间实验中收集，该实验包含三个田间实验点，分别用于作物、作物和杂草的混合物，以及纯杂草。控制配备2m高四通道多光谱相机的无人机进行图像采集。生成的数据集包括375张训练图像，其中包括132张作物图像和243张杂草图像，以及90张作物和杂草混合测试图像，所有图像均为png格式。训练或测试图像均由三组单色图像组成，即红色、近红外线和NDVI（归一化差异植被指数，源自红色和近红外线图像）。图像在作物、杂草和背景的像素级别上进行了注释。https://github.com/inkyusa/weedNet

The weedNet dataset is collected from a controlled sugar beet field experiment, which includes three field plots for crops, crop-weed mixtures, and pure weeds respectively. A drone equipped with a 2-meter-tall four-channel multispectral camera was used for image acquisition. The generated dataset contains 375 training images (132 crop images and 243 weed images) and 90 test images of crop-weed mixtures. All images are saved in PNG format. Each training or test image consists of three monochromatic modalities: red, near-infrared, and NDVI (Normalized Difference Vegetation Index, calculated from the red and near-infrared images). All images are annotated at the pixel level for crops, weeds, and background. https://github.com/inkyusa/weedNet