OCR (image to number) with Construct 3 any solutions

0 favourites
  • 5 posts
From the Asset Store
Casino? money? who knows? but the target is the same!
  • HI

    i am trying to get numbers from an image (ocr)

    is there any way i can do this for construct 3

    i know there is some stuff that was done with construct 2 and

    Tesseract,

    c2rexplugins.weebly.com/rex_tesseractjs.html

    but not for construct 3

    thankyou

    also maybe we can upload image and retrieve result back in construct 3

    with this method , but how to do this

    ocr.space/OCRAPI

    or

    antimatter15.com/ocrad.js/demo.html

  • to do it in C3, you may be able to add the tesseract.js scripts from the GitHub reference link given in the Rex plugin page. But that may be a lot of work.

    The ocr.space link looks like it would be fairly easy to use with AJAX calls. If you have a URL to the image, then a GET call would be easiest. Otherwise you would have to convert the image to base64 and do a POST call. The results would be in a json format in the AJAX.LastData

    where are the numbers coming from? there could be easier ways as well...

  • Try Construct 3

    Develop games in your browser. Powerful, performant & highly capable.

    Try Now Construct 3 users don't see these ads
  • HI AllanR

    The Numbers are from a multimeter to test voltage etc here is some examples

    so the android phone would be positioned over lcd display and take a picture

    giving only this image and not the whole multimeter image,

    The Aim is:

    1. To take a camera picture via construct 3

    2. save image to android folder

    3. OCR the image to number

    thankyou

    Hope you can help in some way

  • well, I played around quite a bit...

    here is my file:

    https://www.rieperts.com/games/forum/ocr.c3p

    the bad news is that the ocr.space web site does not detect any text (numbers) in your sample image. I made a test image and it worked great on mine. So, the font on the multi-meter seems to be a problem. I tried setting the language in the request to both Eng or Korean (because they said that might work better for numbers only) and neither worked.

    also, I could only get the ocr.space site to work with the GET method, which requires uploading the image first to a web site and passing the url. Converting the image to base64 and POSTing that to the site ("api.ocr.space/parse/image") only returned errors. I tried from C3 and downloading the Postman app (which gave better error messages). It seems to be a problem with their site - not C3 or me doing something wrong. Posting the base64 gave a 404 "not found error", posting a url to the file gave a filetype not detected, even if I overrode the filetype using the filetype key. The same images passed to "api.ocr.space/parse/imageurl" (which is the GET method) worked fine.

    that is all I can do...

  • Thanks so much.....

    Works really well,

    i think the problem with multimeter image is that there are spaces in each

    character that make up one number, hence the ocr engine wont pick the numbers up from multimeter reading

    but otherwise it picks up normal fonts very well,

    i was looking at this page and it seems this idea would work well

    to identify multimeter readings,

    diga.me.uk/sevenSegDecode.html

Jump to:
Active Users
There are 1 visitors browsing this topic (0 users and 1 guests)