PlantStereo

PlantStereo数据集是由浙江大学生物系统工程与食品科学学院创建的一个大规模立体匹配数据集，专门用于植物表面密集重建。该数据集包含812对高精度的左右图像，覆盖了菠菜、番茄、辣椒和南瓜等多种植物。数据集的创建过程采用半自动方法，通过相机校准和图像注册从深度图像中获取高精度的视差图像。PlantStereo数据集的应用领域主要集中在植物表型分析和自动化育种，旨在通过高精度的深度估计提高食品生产效率并解决全球饥饿问题。

The PlantStereo dataset is a large-scale stereo matching dataset created by the College of Biosystems Engineering and Food Science, Zhejiang University, which is specifically designed for dense surface reconstruction of plants. It contains 812 pairs of high-precision left and right images covering various plant species including spinach, tomato, pepper and pumpkin. The dataset was constructed via a semi-automatic approach, where high-precision disparity images were obtained from depth images through camera calibration and image registration. The primary application fields of the PlantStereo dataset focus on plant phenotyping and automated breeding, aiming to improve food production efficiency and address global hunger through high-precision depth estimation.