Turn raw videos into reasoned training questions.
What it does
Creates video-understanding training datasets containing captions, structured descriptions, and question-answer pairs with reasoning traces.
When to use it
Use when the user wants to auto-label videos, generate video QA datasets, or create video training data with reasoning traces.
How to use it
Give it videos; it asks about their domain, endpoint access, and run scope, then generates annotation datasets in a results directory.
What you provide
- Your files
Uses
Latest review
What you need · 9
Requires Docker for the container-based pipeline.
Requires NVIDIA Container Toolkit alongside Docker.
Access · 2
This skill
Gemini
Write
Sends annotation inputs to Gemini
ffmpeg
Execute
Installs ffmpeg when missing