Bash script to run (recursively) through folders and create a text file for each image from recognized characters

[simonsan]

I create above script that you can find in a gist (for convenience):

https://gist.github.com/simonsan/2b0103b5e49c3e6a3a9d9abc0674edca

It may be nice to add this functionality to ocrs-cli itself though. This is useful, so the OS search engine can index these files and you can search them.

For that, it may be more useful to have an index folder where the text files are created with a path to the file they belong to? Or an option to decide if they should be stored in the same directory or in a central directory?

[robertknight]

Thank-you for the contribution. I think to start with it would probably be better to create a separate project which uses either the Rust library or the CLI. The CLI is easy to use from any language, but the library is slightly more efficient (avoids the need to reload models for each image). I can imagine that there are a lot of options users might want in a CLI tool that does batch indexing (or that the community might think of) and those would add complexity for basic use cases.

[simonsan]

I agree with the extra project character of this. I added the feature to store the text files either next to the image or in a user-defined directory to the script for now. It sounds really interesting, but I'm currently a bit overloaded with projects! I'll update this issue, when I implement a Rust CLI from the script. Could you please let this open, so people can give some feedback on the idea (discovery)? If someone wants to tackle this project to create a nice CLI for, feel free to go on, I may contribute. 🚀 (Don't forget to link here)

[robertknight]

Could you please let this open, so people can give some feedback on the idea (discovery)?

I have converted this issue into a discussion, as that is a better forum for open-ended threads that aren't expected to be "resolved".

[simonsan]

Thanks!