Font recognition

[PankajWalia1]

BP presents many options on selecting the closest font to the image in the region.. When I look at the rendered image it seems to be a close fit to what I see on the screen, but it still doesn't recognize the text.. I have tried with many different fonts, expanded search list and also tried manual option but still not able to capture the text.. can you share your experience on how to pick the right one/eliminate various other font options? Also, if the rendered image seems close shouldn't that be enough to use that for getting text? look forward to getting a response..

[keri_z_volans]

Fonts can be a pain. I'm pretty sure it's a very crude recognition method, literally comparing pixels. So similar isn't good enough, unfortunately. If you've chosen a font you think should work, you can get more info by clicking on the ""show character"" button in region editor. You can see which characters aren't matching. Make sure that the region is the correct size, and the vertical positioning is consistent. Worst comes to worst you'll have to merge characters into an empty/similar font. If you can find a decent third party font analyzer that would be a big help. (Post it on here if you do) If you figure out that it's a font that isn't included by default, you can import it through ""System Manager""->""Fonts"" Good luck!

[PankajWalia1]

Thank you. That worked with creating a new font, still not able to use any of the existing font options that BP presented. quick question. In the new font I just created I need to add some more characters so I associate the region with that font. When I do show characters it doesn't give me an option anymore to verify the character.. Not able to scroll at the bottom to left and right.. Has anyone seen that issue?

[Charanjiv_Singh]

Hi, I had the same issue when I used the element data as RECOGNISE TEXT, after struggling to figure out different ways to have the font read properly. Then I changed the element data to READ TEXT WITH OCR and I was able to read the text from image. Hope this is the solution you are looking for.

[BartWille]

Recognize text will look for a perfect match. The hitrate is lower but there are no false positives. OCR (Tesseract) will adjust the image (eg. straighting lines) and try to find a match. This increases the hitrate but has the downside of false positives. False positives = Reading a character wrongly

[TetsujiJunicho]

For Recognize Text, it's tricky, but it worked for me by modifying fonts. https://portal.blueprism.com/system/files/Surface%20Automation%20-%20Gu