SB20

该数据集是2020年的甜菜数据集，包含了实例级别的手动标注数据，主要用于训练和推断，重点关注杂草种类。该数据集的训练集中包含1424个植物实例，图像分辨率为640×480。规模上，数据集分为71张训练图像，37张验证图像和35张评估图像，任务类型为实例分割。This is a 2020 sugar beet dataset featuring manually annotated instance-level data, which is mainly used for model training and inference, with a core focus on weed species. The training split of this dataset contains 1,424 plant instances, and all images have a resolution of 640×480. The dataset is partitioned into 71 training images, 37 validation images, and 35 evaluation images in terms of scale, and the corresponding task is instance segmentation.

搜集汇总

背景与挑战

背景概述

SB20是一个甜菜数据集，由波恩大学使用Intel RealSense D435i传感器在Sagarobot上采集，包含RGB-D图像，覆盖甜菜和八种不同杂草，涉及多种生长阶段、自然光照条件和遮挡情况，并提供多类别标注。