Causal3DIdent

该数据集名为Causal3DIdent，主要用于基于三维图像的表征学习，特别关注因果关系。此外，该数据集还用于展示具有拓扑约束的潜在ANM模型的有效性，其任务属于无监督表征学习。This dataset, named Causal3DIdent, is primarily used for 3D image-based representation learning, with a particular focus on causal relationships. Furthermore, it is also employed to demonstrate the effectiveness of latent ANM models with topological constraints, and the tasks related to this dataset fall under the category of unsupervised representation learning.

搜集汇总

背景与挑战

背景概述

Causal3DIdent是一个基于因果图结构的3D视觉数据集，作为3DIdent的更新版本，它引入了六个额外的对象类别（如野兔、龙、牛等），并在潜在变量上施加了因果图以支持因果推理研究。数据集包含训练集和测试集，总大小约为9.1 GB，适用于机器学习和计算机视觉任务。