Which software use for qualified OCR scan book on Linux?

[Bink]

For a book, you can export it as an ODT, for editing in LibreOffice.

The export button is here:

[wodsfortdragon]

I converted, but this big files don`t opens in LO
(It just loads pc and logo displays tried to open)

[Bink]

Hmm… Perhaps just convert it to TXT file then, and copy the text into a new LibreOffice Write document?

[wodsfortdragon]

For converting I also need to open it. I cannot open it in LO and WPS. Maybe it just out of my RAM.

[Bink]

@HogRemains

Google Cloud Vision API’ is actually the best OCR tool I’ve used. But its a cloud tool. I’m new to the forum and don’t want to break any rules so I’ll avoid linking to it here. But, I’ve been able to use it extensively without needing to spend money via their developer program.

I’m sorry @HogRemains, I neglected to say hi, and welcome you to the community!

@wodsfortdragon , did you have a look at @HogRemains suggestion? If I’m not mistaken, it is not AI driven, but rather using pre-trained machine learning models much like Tesseract does.

[wodsfortdragon]

I asked in Libreoffice group others to try open it in their pc.
Link is https://cloud.disroot.org/s/53eaxFHrtLtcieJ

[wodsfortdragon]

Didn`t try, but I just want to test this utility(gimagereader with tesseract). If it will be not good to me, I try others.

[wodsfortdragon]

@Bink Maybe you try, please?

[Bink]

I just tried @wodsfortdragon. I can confirm it opens, but it did take a few moments, it’s a big file

My system is a Ryzen 9 5900X with 64GB of RAM.

[Bink]

@wodsfortdragon, are you using the LibreOffice document just as a test for the OCR?

[wodsfortdragon]

Without pictures? Original file: https://nextcloud.documentfoundation.org/s/CiFSWbeiy3ed2mW .

[wodsfortdragon]

Yes)

[wodsfortdragon]

I want to have fully copied file,even with pictures and other details

[wodsfortdragon]

@Bink

My system is a Ryzen 9 5900X with 64GB of RAM.

Cool, I have Intel Core I3-1005G1 with 8GB of RAM

[Bink]

@wodsfortdragon

Without pictures? Original file: https://nextcloud.documentfoundation.org/s/CiFSWbeiy3ed2mW .

That one is a PDF, so you could open it in LibreOffice Draw and retain all those things and have access to the text. OCR isn’t necesarry.

But would I be correct in understanding, this is just a test file and the actual content you’re working with is not a PDF?

[wodsfortdragon]

Yes, it will be photo of paper book. So what use if I just want do photos of pages, upload to pc, do OCR, export/copy images to PDF file, fix 20 mistakes(not 100) and receive 600 pages book?
in this text it is not copied my images

[Bink]

@HogRemains

I understand your hesitation to use an LLM, especially for a large project. But, you won’t achieve perfect OCR that retains formatting without a lot of proofing and manual config, or AI. Personally, I would choose the latter: Give the outputs from Tesseract to an LLM tasked with converting raw text to readable markdown. You’ll lose the original formatting, but I don’t think that’ll meaningfully hurt the material’s readability in the end.

There are going to be tradeoffs to every problem and decision, right? I think using LLM here is going to get you the best end result. I know that’s faux pas but it is what it is.

I think I’d have to agree with @HogRemains with respect to your options. Getting a formatted 600 page document with text and images from OCR, is not at all trivial.

[wodsfortdragon]

How use it?

[HogRemains]

First you’ll have to have a Google account, and a project started in Google Cloud Console. Once you have it, you can follow along with this youtube tutorial. The best part for you will be outsourcing the RAM-intensive tasks to the cloud server, 8gb of RAM is not much headroom for ML tasking.

[cc_spicuous_2]

Tested on 467 scanned pages, offline, on a student laptop, low-power CPU from 2019/20.

The extractor fails to recognize the accents on the letters, which is a minor but noticable error.

https://github.com/hiroi-sora/Umi-OCR („hiroi-sora/Umi-OCR“, hiroi-sora 2026 @ 4. Juli 2026)