zeroframe/sample_3d_objects

Urban3D数据集样本子集是一个包含超过10,000个跨城市类别的多视角视频和3D重建的大规模收集，用于机器学习研究，特别是在自动驾驶系统、智能基础设施和城市场景理解方面。数据集中的每个对象都是从单个视频序列中提取的，并且经过了自定义的COLMAP流程处理，提供了高分辨率的RGB帧、相机姿态、特征匹配数据库、稀疏3D点云等。

The Urban3D dataset sample subset is a large-scale collection of over 10,000 multiview videos and 3D reconstructions across urban categories, used for machine learning research, especially in autonomous systems, smart infrastructure, and urban scene understanding. Each object in the dataset is derived from individual video sequences and processed through a custom COLMAP pipeline, providing high-resolution RGB frames, camera poses, feature match databases, sparse 3D point clouds, and more.