Dataset · yutongli2024
EgoExo-Next benchmark for egocentric actions
yutongli2024/EgoExo-Next
Visual benchmark for egocentric-exocentric action understanding and future state prediction.
What's inside
The dataset contains between 10K and 100K English images for visual question answering and image-to-text tasks. Prediction focuses on selecting the target state among four candidates, including distractors.
What you can do with it
- Training and evaluating Vision-Language Models for action understanding
- Testing models on predicting future visual states
- Tasks
- Visual question answering, Image → text
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 20.8K
- Likes
- 0
How to load it
from datasets import load_dataset
ds = load_dataset("yutongli2024/EgoExo-Next")