RobustPointSet
收藏资源简介:
RobustPointSet是由Autodesk AI Lab创建的一个用于评估点云分类模型鲁棒性的数据集,基于ModelNet40数据集,包含12,308个CAD模型,分为40个类别。该数据集在原始训练和测试集的基础上,增加了6个经过不同变换的测试集,以模拟未在训练中见过的变换情况。创建过程中,对点云应用了噪声、缺失部分、遮挡、稀疏、旋转和翻译等变换。该数据集旨在解决点云分类模型在面对未见过的数据变换时的性能评估问题,推动模型在实际应用中的鲁棒性研究。
RobustPointSet is a dataset developed by Autodesk AI Lab for evaluating the robustness of point cloud classification models, built on the ModelNet40 dataset. It comprises 12,308 CAD models grouped into 40 categories. Building upon the original training and test splits, this dataset adds six transformed test sets to simulate transformation scenarios unseen during model training. During its construction, various transformations including noise, partial missing, occlusion, sparsification, rotation, and translation are applied to the point clouds. This dataset aims to address the challenge of evaluating the performance of point cloud classification models when faced with unseen data transformations, and to promote research on model robustness in real-world applications.

- 1RobustPointSet: A Dataset for Benchmarking Robustness of Point Cloud ClassifiersAutodesk AI Lab · 2021年



