Skip to content

Dataset · DAGroup-PKU

Robotic videos for embodied generation

DAGroup-PKU/RoVid-X

4M robotic clips with RGB, depth, and optical flow for image-to-video.

What's inside

The RoVid-X dataset contains 4M robotic video clips (over 10K hours) for large-scale video generation training. It features 1,300+ fine-grained robotic skills and multimodal physical annotations: RGB, depth, and optical flow. The data covers various robots and tasks; language is English, modality is video.

What you can do with it

  • Training image-to-video models on robotic scenes
  • Evaluating video generation for embodied AI
  • Fine-tuning video models with depth and optical flow
Tasks
Image → video
Languages
English
Size
over 1T rows
Downloads this month
13.2K
Likes
74

How to load it

from datasets import load_dataset

ds = load_dataset("DAGroup-PKU/RoVid-X")