Unity wrapper - Google Tesseract OCR

Hi all,

I've come across quite a number of users here unable to find a quality OCR plug-in for Unity projects.
At DesignSpark (part of engineering distributor - RS Components), an Augmented Reality app (iOS & Android) to discover RS product 3D models was released as a 'closed beta' for testing between September 2017 and September 2018: https://www.rs-online.com/designspark/test-drive-the-designspark-ar-app-iosandroid
This app required the user to point their device's rear camera at a manufacturer part number, which then ran an OCR scan to find the product from the RS catalog and deliver a 3D model along with purchase information.
A key outcome of the app development was a 'first of its kind' wrapper for Google's Tesseract OCR engine for use in Unity C# projects. The wrapper will enable powerful character recognition in apps built for any mobile platform.

The Unity package for this wrapper is downloadable by DesignSpark members here: https://www.rs-online.com/designspark/designspark-ar-app-ocr-module-for-unity

You can seek technical help from the developer of this wrapper: Sam at https://forum.unity.com/members/agentmilkshake1.599188/ or sam@relative-dimensions.com

Cheers!
DJ

 

I downloaded it but there is no demo,Could you give me some help ~?

 

Here is a package to demo the plugin on Android.

1. Set Android as your build target
2. Import the package linked in the first post
3. Import the package attached to this post
4. Build and run the Demo scene to your Android device (it does not work in the Unity editor)

The "TestImage.jpg" is passed into Tesseract and should have it's text printed on the screen. Sorry, I don't have this working for iOS yet (not sure if my project will get far enough for me to care) but the demo should at least get you started.

 

Thanks for your sharing.
But some errors occurred.

 

Sounds like you need to do step 2 in my post above. Did you import the package that the original poster linked to?

 

Hi, I made it work, it's great, thanks for sharing!

Go to File > Build Settings..., switch to Android. There is a preprocessor that limits it to Android and iOS.

Also, you need to build it and run it on an actual mobile device, it's not working on a desktop. So it's not working with Unity Remote, you really need to copy the apk and install, or setup Unity to deploy on mobile.

 

Thank you for the example
If I were to add my own list of words, where would I add them? I need to add error codes that should be read from an image

Thank you

 

Hi there, I am unable to build your demo scene to an iOS device due to a number of linker issues related to the TesseractOCR library. Do you have any idea how I might resolve these?

 

Same for me on iOS, still digging around but no luck.

Edit: Just add libz.tbd under "Linked Frameworks and Libraries" in xcode and it should work.

 

Changes to the config file "StreamingAssets\tessdata/tessconfigs/recognitionConfigs.txt" does not seem to work. im trying to whitelist and black list certain characters and add a few patterns.
Has anyone had luck trying to do this?

 

I need to configure Tesseract to work with the pattern B102 so Character/Number/Number/Number/Number.
My problem now is that tesseract expects the B to be an 8 because its nexto to a series of numbers.
As I see it I have several parameters to tweak.

• Whitelist and or Blacklisting characters: this would help produce more reliable results, although probably slow things down. I think i will set a whitelist and not blacklist anything. its easier to clean up the OCR result afterwards.
  
• Patterns: ive tried defining patterns but the seem to have no effect. ( does anyone now how to debug teseract to see if its actually using the patterns.
  
• The distance between the letters ( no idea where to change it). If i change this setting then i could get tesseract to detect individual letters and not try to make sense of it as a word.
• turning off dictionary: tried this in the config file but seemed to have no effect.
• retraining tesseract for my purpose.... i have no idea how and its likely not to work since changing the config file has no effect.

So in order for me to move on I need help figuring out how to debug tesseract so that i can see if my settings indeed are loaded from the configfile.

 

I am trying to get a demo built on an IOS device (My project would require both platforms to work). I have successfully built it to the iPhone and am given the error code (on screen) "Error: Unable to load DLL "libtess.so", the specific module could not be found" has anyone experienced a similar situation and know how I could resolve this?
Thank You for any help

 

What should I have to do if I want to take the input image from the camera in my app?

 

I am very new to unity. Can you give me any hint or idea about how can i make an ocr like the one design park team built for their android app(where user uses the camera to scan the word and then get processed by the ocr) using the TesseractWrapper.unitypackage?

 

how can i add OCR for Hindi language in plugin given by design spark ??

 

Hello,
I could install and try the plugin on Android.
However I'm experiencing issues with OCR. I'm trying to scan business cards with my device camera and I have a lot of errors in recognition : bad characters, false positives, bad whitespaces and I'm sure it's not beacause of the photo.

Has anyone faced this issue and could config, somehow, the algorithm to make it work better ? Since there is no documentation, I don't which files to edit.

 

How to add another language ??

 

Can try this https://medium.com/xrpractices/offline-ocr-using-tesseract-in-unity-part-1-b9a717ac7bcb
It guides you on how to set up a standalone OCR for Unity using Tesseract

 

There is nothing to download on the page, has the download link been removed?

 

Is it works on IOS.?

 

I think these packages need a 2021 update.
They're chock-full of errors and having separate packages strewn about forums isn't very practical?