A Large Dataset of Object Scans

对象扫描大数据集是一个包含一万多个真实对象 3D 扫描的数据集。为了创建数据集，作者招募了 70 名操作员，为他们配备了消费级移动 3D 扫描设置，并付钱让他们扫描环境中的物体。操作员在实验室外扫描他们选择的对象，没有计算机视觉专业人员的直接监督。结果是大量多样的对象扫描集合：从鞋子、杯子和玩具到三角钢琴、工程车辆和大型户外雕塑。作者与律师合作，确保数据采集不违反隐私限制。获取的数据被放置在公共领域并且可以免费获得。

The ObjectScan Large Dataset is a collection of over 10,000 3D scans of real-world objects. To develop this dataset, the authors recruited 70 operators, equipped them with consumer-grade mobile 3D scanning setups, and compensated them for scanning objects in their surrounding environments. The operators scanned self-selected objects outside of laboratory settings, without direct supervision from computer vision professionals. The resulting collection constitutes a highly diverse set of object scans, ranging from everyday items such as shoes, cups, and toys to large objects including grand pianos, engineering vehicles, and large outdoor sculptures. The authors collaborated with legal professionals to ensure that the data collection process did not violate privacy regulations. All acquired data has been released into the public domain and is freely available.

搜集汇总

背景与挑战

背景概述

该数据集包含超过一万个真实物体的3D扫描数据，由70名操作员使用消费级移动设备在非实验室环境中自主采集，涵盖了从日常物品到大型雕塑的广泛对象。所有数据经过法律合规审查后置于公共领域，可供免费获取和使用。