用于语义分割的作物、杂草、草和植物茎图像数据集

联合干茎数据集旨在检测和区分双子叶杂草和禾本科杂草。检测需要区分不同的除草方法以及作物茎和双子叶杂草，以便于为植物实施特定的机械除草。发布的数据包含两个子数据集；第一个数据集来自甜菜2016数据集，包括921张1296×966像素的png格式红绿蓝和近红外线图像，第二个数据集由400个512×384像素的png格式红绿蓝图像组成，由无人机拍摄获取。上述数据库图像为语义分割和茎检测任务提供了作物、双子叶杂草、禾本科杂草和背景的像素级注释以及茎位置信息。https://www.ipb.uni-bonn.de/people/lottes/

The Joint Stems Dataset is developed for detecting and differentiating dicotyledonous weeds and gramineous weeds. It aims to distinguish between different weeding methods, crop stems and dicotyledonous weeds, so as to facilitate targeted mechanical weeding for plants. The released dataset includes two subsets: the first subset is derived from the Sugar Beet 2016 Dataset, which contains 921 red-green-blue and near-infrared (RGB-NIR) PNG images with a resolution of 1296×966 pixels. The second subset consists of 400 RGB PNG images with a resolution of 512×384 pixels, captured by unmanned aerial vehicles (UAVs). All images in this dataset provide pixel-level annotations for crops, dicotyledonous weeds, gramineous weeds and the background, as well as stem position information, for semantic segmentation and stem detection tasks. https://www.ipb.uni-bonn.de/people/lottes/