OCR and AI Summarization via External API

[eifrig]

Hi everyone,

I've been using Photo Supreme for several years now and I'm really happy with it – kudos to the developers and the community!

I have a specific use case and would love to hear if anyone has already implemented something similar or has ideas for a possible solution:
I regularly scan a large number of newspaper articles and would like to make them searchable via OCR. Ideally, I’d also like to automatically generate a summary of each article using AI (e.g., a short abstract or description of the content).

My idea is to send the scanned image (JPG or TIFF) from Photo Supreme to an external (http-post) API (which I would build myself). This API would perform the OCR and AI-based summarization. The response – such as the recognized text or summary – should then be written back into a specific field in Photo Supreme (e.g., a custom field or the description field).

Has anyone here done something similar or found a way to implement such a workflow?
Are there ways to trigger external tools from within Photo Supreme (via scripting, context menus, or special labels, for example)? Or do you have other workflows that could be adapted to this use case?

Any suggestions, tips, or pointers would be greatly appreciated!

[andrew.heard]

I use VueScan for scanning paper to PDF with OCR enabled. It works brilliantly.

[G8DHE]

If you want to trigger external programs/Apps/routines I just use the External applications bar (top right) to run dos batch files only requirement is to have at least one image selected to pass to the external routine. You setup the program from the normal Preferences Menu|Other settings.

[Mke]

I use VueScan for scanning paper to PDF with OCR enabled. It works brilliantly.

I do the same with the OCR option on my Epson Perfection scanner, which works well. I also use Pdfforge's PDF Creator (https://www.pdfforge.org/pdfcreator) to archive digital documents with OCR.

However I only have around 90 PDFs logged within PSU; the rest of them (around 18,000) are just in my file system, which allows me to do a full text search.