CellTypeGraph

CellTypeGraph是一个专为植物发育生物学中细胞分类问题设计的新型几何计算机视觉基准数据集。该数据集由德国海德堡大学人类与环境中心创建，包含95,757个节点，用于节点分类任务。数据集基于拟南芥胚珠的3D共焦显微图像，通过细胞实例分割和组织注释生成。数据集旨在通过机器学习和计算机视觉技术，解决植物器官复杂细胞结构和形状的自动注释问题，特别是在拟南芥胚珠中的应用。

CellTypeGraph is a novel geometric computer vision benchmark dataset tailored for cell classification tasks in plant developmental biology. It was created by the Centre for Human and Environmental Sciences, Heidelberg University, Germany, and contains 95,757 nodes for node classification tasks. The dataset is generated from 3D confocal microscopy images of Arabidopsis thaliana ovules through cell instance segmentation and tissue annotation. It aims to address the automatic annotation problem of complex cell structures and shapes in plant organs using machine learning and computer vision technologies, particularly for applications in Arabidopsis thaliana ovules.