The result is really not good about tesseract-ocr-ios HOT 14 CLOSED

u must configure tesseract for best results. all helps are available at www.google.com :)

from tesseract-ocr-ios.

I'm having the same issue. I tested the template framework project on my iPad, and I put this image (http://www.myfico.com/Images/sample_overlay.gif) on my computer and I used the camera and the app won't detect :c I did something wrong? What I will need optimize? Thanks for you work!

from tesseract-ocr-ios.

https://imgur.com/a/MN0av is the result. I only import UIImage+Filters and downloaded and imported the last English Tessdata from the Tesseract Google Site. I tested too only with UIImage+Filters imported, and I have every time the same results.

from tesseract-ocr-ios.

Have you tried different images as source?

Sent from my iPhone

On 18 Aug 2014, at 06:06, Jakuya [email protected] wrote:

https://imgur.com/a/MN0av is the result. I only import UIImage+Filters and downloaded and imported the last English Tessdata from the Tesseract Google Site. I tested too only with UIImage+Filters imported, and I have every time the same results.

—
Reply to this email directly or view it on GitHub.

from tesseract-ocr-ios.

Yes, I made a few tests and I never had a good result :/

from tesseract-ocr-ios.

Actually, tesseract is only OCR, nothing more.
And it supports only fonts, which it supports, nothing more.

First, you need to preprocess your images to allow tesseract to work correctly, as it's written here:
https://code.google.com/p/tesseract-ocr/wiki/ImproveQuality.
That means you need to remove a noise from your images, rotate images to make the text horizontal, and so on.
Second, there is an issue with fonts. I'm not sure the tesseract can recognized outlined fonts. So you need to train tesseract to support such fonts. "How to" is written here: https://code.google.com/p/tesseract-ocr/wiki/TrainingTesseract3
Or, which is easier, use some filters, to make your text solid. Actually, you could achive this just using adaptive binarization.

So, the issue have nothing related to the project, it's related to image processing basics.

from tesseract-ocr-ios.

I guess, it could be closed.

from tesseract-ocr-ios.

I've moved the information provided here to the Wiki: https://github.com/gali8/Tesseract-OCR-iOS/wiki/Tips-for-Improving-OCR-Results

I've also added some more notes about using GPUImage. I'll close this issue, but any other ideas about improving OCR results for Tesseract should be added to the Wiki page.

from tesseract-ocr-ios.

Good job!

Should we explain about skipping the internal thresholding somewhere? or at least mention about this feature on that page?

from tesseract-ocr-ios.

Yeah I think that'd be good to add. Want to add another section (one "#" in markdown) to explain about it?

from tesseract-ocr-ios.

How can I do this? Do I have a write access to the Wiki?
Wow! I do! :)
Let me try.

from tesseract-ocr-ios.

No. actually I don't. I can only edit the Sidebar content. I cannot find how to edit the wiki pages.

from tesseract-ocr-ios.

I think you should be able to because @bamx was able to edit one of the Wiki pages and he doesn't have collaborator status. Here's where I see it:

from tesseract-ocr-ios.

Ah! Thanks a lot for the help.
Seems ok now.

from tesseract-ocr-ios.