Dataset · netflix
Netflix layered video dataset for video editing
netflix/Vera-Layered-Video-Dataset
Dataset for the Vera model (Netflix and Caltech) for layered video editing with background preservation.
What's inside
A video dataset containing 10k to 100k examples, prepared by researchers from Caltech and Netflix for the Vera project. The dataset is designed for training and validating diffusion models that edit videos by layers, generating a modification layer and alpha channel without distorting the original content.
What you can do with it
- Training diffusion models for layered video editing (text-to-video and video-to-video)
- Training systems for alpha mask generation and local object replacement in video clips
- Benchmarking algorithms for preserving the original background when modifying a video scene
- Tasks
- Text → video
- Size
- 10K–100K rows
- Downloads this month
- 23.1K
- Likes
- 61
How to load it
from datasets import load_dataset
ds = load_dataset("netflix/Vera-Layered-Video-Dataset")