Objectron

Objectron数据集由谷歌创建，包含14,819个短视频，涵盖9种常见物体的3D边界框标注。数据集收集自全球10个国家，包含400万张标注图像，旨在推动3D物体检测技术的发展。每个视频捕捉物体从不同角度的视图，并附有增强现实(AR)会话元数据，如相机姿态、稀疏点云和表面平面。Objectron数据集的应用领域包括3D物体跟踪、视图合成和3D形状重建，旨在解决真实世界场景中的3D物体理解问题。

Objectron Dataset was created by Google, containing 14,819 short videos with 3D bounding box annotations for 9 common object categories. Collected across 10 countries worldwide, the dataset includes a total of 4 million annotated images, and is designed to advance the development of 3D object detection technologies. Each video captures the target object from various angles, and is accompanied by augmented reality (AR) session metadata including camera poses, sparse point clouds and surface planes. Application scenarios of the Objectron Dataset include 3D object tracking, view synthesis and 3D shape reconstruction, aiming to address 3D object understanding tasks in real-world scenarios.

搜集汇总

背景与挑战

背景概述

Objectron数据集由谷歌创建，包含14,819个短视频和400万张标注图像，涵盖9种常见物体的3D边界框标注，收集自全球10个国家。该数据集旨在推动3D物体检测技术，提供增强现实(AR)元数据如相机姿态和稀疏点云，应用领域包括3D物体跟踪、视图合成和3D形状重建，专注于解决真实世界场景中的3D物体理解问题。