2016年德国甜菜图像数据集

甜菜2016数据集中的图片使用配备多个传感器的田间机器人获取。图片可用于杂草控制以及定位和导航。该数据集是一次使用机器人获取此类图片的初步探索。最终形成了规模较大的数据集。使用近红外线机器人在受控的照明下，在甜菜农场获取三个月内四通道红绿蓝和近红外线图像，收集图像为png格式，1296×966像素。除了用于导航的数据外，该数据集还包括283个像素级别的多类（即甜菜和九种不同类型的杂草）注释图像，以及更大的一组约12,340个三类（即，作物、杂草和背景）像素级注释。该数据集已被广泛使用用于开发机器人作物和杂草检测算法。https://www.ipb.uni-bonn.de/data/sugarbeets2016/

The images in the Sugar Beets 2016 dataset were captured by a field robot equipped with multiple sensors. These images can be applied to weed control, robot localization and navigation tasks. This dataset is an initial exploration of acquiring such imagery via robotic platforms, which eventually evolved into a large-scale dataset. It was collected over a three-month period at a commercial sugar beet farm under controlled lighting conditions using a near-infrared-equipped robot, encompassing four-channel red, green, blue, and near-infrared (RGB-NIR) images. All collected images are saved in PNG format with a resolution of 1296 × 966 pixels. Beyond navigation-related data, the dataset includes 283 pixel-level multi-class annotated images, with the classes being sugar beets and nine distinct weed species, as well as a larger corpus of approximately 12,340 pixel-level annotations for three classes: crop, weed, and background. This dataset has been widely employed for developing robotic crop and weed detection algorithms. For additional details, please refer to https://www.ipb.uni-bonn.de/data/sugarbeets2016/

搜集汇总

背景与挑战

背景概述

2016年德国甜菜图像数据集是一个由田间机器人获取的大规模数据集，包含四通道红绿蓝和近红外线图像，主要用于杂草控制、定位和导航。数据集包含283个像素级多类注释图像和约12,340个三类像素级注释图像，广泛应用于机器人作物和杂草检测算法的开发。