Skip to content

Dataset · rghermi

Short-films 20K video dataset for story understanding

rghermi/sf20k

A large open video dataset of 20,143 amateur films with a total duration of 3,582 hours.

What's inside

Contains 20,143 English short films with an average duration of 11 minutes and a total volume of 3,582 hours. Includes a training split with synthetic questions and a test benchmark with human-crafted questions.

What you can do with it

  • Training video analysis and video understanding systems
  • Developing visual question answering models
Tasks
Visual question answering
Languages
English
Size
100K–1M rows
Downloads this month
634
Likes
15

How to load it

from datasets import load_dataset

ds = load_dataset("rghermi/sf20k")