InternRobotics/OmniWorld

OmniWorld是一个多领域和多模态的数据集，包含多个不同的子数据集。这个数据集旨在为4D世界建模提供支持，涵盖了文本到视频、图像到视频、图像到3D、机器人技术和其他任务。数据集包括游戏、机器人、人类动作等领域的视频序列，并提供了丰富的模态数据，如RGB、深度、相机姿态、文本描述等。OmniWorld数据集的详细使用说明和文件结构也在README文件中提供。

OmniWorld is a multi-domain and multi-modal dataset consisting of several distinct sub-datasets, designed to support 4D world modeling. It encompasses tasks such as text-to-video, image-to-video, image-to-3D, robotics, and other categories. The dataset includes video sequences from domains like games, robotics, and human actions, providing a rich variety of modalities like RGB, depth, camera poses, and text descriptions. Detailed usage instructions and file structure for the OmniWorld-Game and OmniWorld-Game Benchmark datasets are also provided in the README file.