You can deploy this model and start earning money today!

interlive/
scenewalk

video-text-to-text

visual-question-answering

Views

No views yet

	SceneWalk Dataset




	
	Description

SceneWalk is a large-scale video dataset for long-form video understanding with detailed scene-level annotations and conversation data.

	Dataset Structure




	
	Data Fields

id: Video ID video: Path to video file total_time: Total video duration in seconds clip_num: Number of clips/scenes scenes: Scene segmentation with timestamps conversations: QA pairs for video understanding

Files

scenewalk/ ├── data/ │ └──… See the full description on the dataset page: https://huggingface.co/datasets/interlive/scenewalk.