FR: Tesseract for open-source OCR

zsviczian / obsidian-excalidraw-plugin

A plugin to edit and view Excalidraw drawings in Obsidian

4.27k stars 233 forks source link

FR: Tesseract for open-source OCR #897

Open Comprehensive-Jason opened 2 years ago

Comprehensive-Jason commented 2 years ago

Is your feature request related to a problem? Please describe. Thanks for adding OCR via Taskbone to Excalidraw! I find it disconcerting for the Excalidraw plugin to rely on a closed-source external service for OCR capabilities.

Describe the solution you'd like Could you add the option to use Tesseract for OCR? Tesseract is an open source OCR engine. I know that Omnisearch uses tesseract.js to extract text from images.

zsviczian commented 2 years ago

I fully understand that for some, using the cloud is a dealbreaker. In that case I recommend not turning on the feature.

I tested Tesseract with very poor results. Practically no support for handwritten text and not very reliable results for photo recognition.

For now, I am going to leave the scanning bit at this solution. Maybe in the future I'll revisit it and give Tesseract another chance.

DavidFarago commented 6 months ago

How is the OCR performance (accuracy) of Tesseract vs. Taskbone? Can you train Taskbone on your own handwriting?

zsviczian commented 6 months ago

I've tried Tesseract. It's performance is beyond very poor. In my experience completely useless.