ScanObjectNN 3D 点云数据集

ScanObjectNN 是一个真实的 3D 点云分类数据集，益于背景、缺失部分及变形等特点，该数据集极具挑战性。它由从真实世界室内扫描中提取的遮挡对象组成，包含来自 15 个类别的 2,902 个 3D 物体。

ScanObjectNN is a real-world 3D point cloud classification dataset. Featuring characteristics such as background clutter, partial incompleteness and geometric deformations, this dataset is highly challenging. It consists of occluded objects extracted from real-world indoor scans, and contains 2,902 3D objects belonging to 15 categories.

搜集汇总

背景与挑战

背景概述

ScanObjectNN是一个具有挑战性的真实3D点云分类数据集，包含来自15个类别的2,902个3D物体，数据总量达208.45GB。其特点在于包含真实世界室内扫描的遮挡对象，具有背景干扰、缺失部分及变形等复杂情况，适合用于3D物体识别研究。