im2txt but for video #34
Description
What is the problem that is being solved with the new feature?
I would like to extract more metadata from videos, objective data that can be used to cluster similar videos together, preferably offline and output <1kb per row per column.
Enumerate an unordered list of alternatives that you've thought about
- extract frames and run im2txt
- foundation model
- use subtitles or generate captions from audio
If applicable, state your a preferred solution
It would be nice if there was an existing C, Rust, or Python application or library that can do this already