Skip to content

Dataset · junchaoh-cs

Video Indexes and Annotations for World Models

junchaoh-cs/SolarWM-Data

Metadata, indexes, and camera-aligned video reconstruction packages for training world models.

What's inside

The main repository contains between 1 million and 10 million English records: test indexes, format examples, licenses, and an annotation reconstruction package for video data with camera parameters. The original video files and latent representations are distributed separately due to their large size and licensing terms.

What you can do with it

  • Research and development of world models with camera control
  • Standardization and testing of video data reconstruction pipelines
Languages
English
Size
1M–10M rows
Downloads this month
11.2K
Likes
56
Access
on request on Hugging Face

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("junchaoh-cs/SolarWM-Data")