Feature request - Specific Font OCR about ocrad.js

[antimatter15]

I think building a font-specific OCR system would be really useful, but
it's a bit outside of what Ocrad.js is meant to do. While working on
Naptha, I've played with building my own text recognition system which was
dependent on knowledge of the font a priori (another interesting problem to
solve is to figure out what font some text is written in before recognizing
its letters, which serves as the basis of an interesting chicken-and-egg
problem).

I'd really love to see something like this exist, and I'm not sure I'll
have the time to make something like this (and I'm not convinced I know
enough about OCR to do it well). But I too would love to see something like
this exist.

On Sat, May 3, 2014 at 5:33 AM, adam80 [email protected] wrote:

THis is a great tool and I have been playing around with it for the last
couple of days. Is there anyway to use an specific font for the basis of
OCR, say Calibri? This means that if the user know what the base font of
the text they are scanning is there would be a higher chance of conversion?
That is what I am hoping for.

Please let me know of there is a way to do this if you can.

Thanks!

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/8
.

[bradisbell]

Is there any library you would suggest that can be configured for a particular font or be trained in some way? I'm looking for a way to recognize scoreboard data, ideally in client-side JS, and am curious where to begin if not this library. Thank you!

[antimatter15]

Have you tried using tesseractJS instead? It should be generally more accurate

…

On Mon, Dec 3, 2018 at 9:44 AM Brad Isbell ***@***.***> wrote: Is there any library you would suggest that can be configured for a particular font or be trained in some way? I'm looking for a way to recognize scoreboard data, ideally in client-side JS, and am curious where to begin if not this library. Thank you! [image: scoreboard-guests] <https://user-images.githubusercontent.com/1097509/49391241-bee40a80-f6f0-11e8-9099-05b29a4ae61e.png> [image: scoreboard-10] <https://user-images.githubusercontent.com/1097509/49391242-bee40a80-f6f0-11e8-91a4-a2880f18f468.png> [image: scoreboard-68] <https://user-images.githubusercontent.com/1097509/49391243-bee40a80-f6f0-11e8-8a43-b4916f335911.png> — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#8 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAB1ZsrdVt5vJZvzFiGXXUumCC6HbMbrks5u1WMagaJpZM4B3rA3> .

[bradisbell]

@antimatter15 Thanks, yes I've tried it but it's mis-identifying the numbers since they do look quite a bit like other characters. 8, for example, appears often as a B. If I could re-train the analysis, I think there'd be a lot more accurate detection.

I'm just getting started with this, so I'm not sure how to make my own dataset for Tesseract, but I'll continue looking into it. Thank you for your advice!