Dataset · yyd7
Source Videos for OneStreamer-1M Dataset
yyd7/OneStreamer-Core-SourceVideo
English and Chinese video files for Visual Question Answering tasks from the OneStreamer-Core project.
What's inside
The repository contains source video materials associated with annotations from the primary MCG-NJU/OneStreamer-1M dataset. The data is partitioned by source and packaged into part-XXXXX.tar archives along with JSON manifests. The materials are presented in English and Chinese and are subject to the licenses of their original sets.
What you can do with it
- Training Video-LLMs to answer questions based on video content
- Developing streaming content analysis and comprehension systems
- Tasks
- Visual question answering
- Languages
- English, Chinese
- Downloads this month
- 70.3K
- Likes
- 5
How to load it
from datasets import load_dataset
ds = load_dataset("yyd7/OneStreamer-Core-SourceVideo")