Automatic Metadata Creation -- Description, Subject, Title, and Keywords

[Tach]

A little script that uses a local open source vision language model to automatically create descriptive metadata for images. Requires either an nVidia graphics card or a decently beefy computer, but it runs everything on your computer without sending any data out. Let's you see what it does before it makes changes (enable Dry Run box) and is completely open source. Might take a little above average skill to operate but for the bold, it can free up a whole lot of grunt work, and is actually pretty good at what it does.

https://github.com/jabberjabberjabber/LLavaImageTagger