This is not an officially supported Google product. The code samples shared here are not formally supported by Google and are provided only as a reference.
This repository currently houses four different video metadata related solutions:
- AI Metadata: Use LLMs to generate metadata and descriptions for video content based on a transcript of the video's audio track.
- Smart Ad Breaks: Identify suitable cue points for VOD content based on shot changes.
- Celebrity Detection: Identify celebrities in video content using GCP's celebrity detection API (soon to be deprecated.)
- Image Metadata: Identify things/objects in video content.
Navigate to their respective directories for more info about each solution.
To manually run tests and the linter:
sh test_and_lint.sh