Reading text with OCR with dark background

[ArshuHansda]

Hi, I am trying OCR on a red background image having text in black, but it doesn't give any value, that is the value from read stage is empty. Experimenting with the Scale value didn't help. I have read the OCR pdf from the portal. Can anyone help me with this and tell me where I am wrong? Steps I followed - 1)Opened the image with image loader. 2) Spied the region with region mode 3)Used read stage - Read text with OCR , Scale value changed from 10 to 100 4) Got empty value   I tried inverting the image and then used OCR, this also didn't work.   Attached is the image.

[ashish.easow]

Hi, Are you on version 6.x or earlier. I tried using v6, but it was detected fine. its a simple object with the read - read text with OCR stage. attached results there is a red gradient in the background, so that may cause problems, and I even expected it to get a few alphabets wrong because of that, but it seems to be fine.  Check if gets solved with v6, if you are on an earlier version, maybe you can try updating the tesseract pack.   regards Ashish  

[ArshuHansda]

Yes, I am using v6, tried it multiple colors, only this red image was not able give the output.  I am also using a simple read stage with ""Read text with OCR"" with scale. I guess using the Image loader made the difference, I will try with Internet Explorer. Will try with updating the tesseract too.   Thanks, Arshu

[John__Carter]

You must be doing something wrong. I tried your image and it worked first time with the default inputs - all I did was make a coordinate region to use in a read stage.

[ArshuHansda]

I did the same thing actually, using the Image loader may be the problem. Attached screenshot.

[DhruvinPatel]

Did you have font smoothing settings disabled? If not, that should be first step. and if you have that already, try to get the overlaid on new screenshot and try. It should work.