Re: [orca-list] Speech-dispatcher library for C#?

[Janina Sajka <janina rednote net>]

Didier, Rastislav:

Quick comment from me also in line below ...

Orca screen reader developers writes:
> Hello,
>
> answer inline.
>
> Le 01/08/2020 à 03:11, Rastislav Kiss via orca-list a écrit :
> > Hello,
> > as for the project, I have few very interesting ideas of combining ocr
> > with my own image analysis algorithms, which could result in more or
> > less new ways of using characters recognition for us.
> > The problem is, that I don't know yet, whether they'll really work, as
> > theory is one thing, and practice another. :)
> > I don't even know yet, whether the whole stuff will be researched and
> > developed as one project or as more smaller parts, the former would be
> > more practical, the latter more realistical, as individual parts have
> > varying development times and it would be unpractical to wait with one
> > taking few days to develop for another, which needs a month to find the
> > best possible way and create supporting framework. Especially taking in
> > my inability to keep working on things for longer time. :)
>
> Do you really intend to reinvent tesseract 4? I hope not...
> https://github.com/tesseract-ocr/tesseract

Yes, if Rastislav's ideas are already incorporated in Tesseract, than
reinventing it might be simply a learning exercise, not a practical
benefit.

On the other hand, Tesseract is relatively old technology. Nowdays, with
AI available from multiple cloud services for relatively affordable
costs, it might be very useful to investigate enhancements, or even
entirely different approaches.

Also, if we move away from the 'C' in 'OCR,' there's lots of room for
cool new applications of technology and AI could help develop those.

Some of these have already shown up in applications, e.g. image
recognition, song recognition, etc.

Let me give just one possible application that came up in recent W3C
conversations preparatory to an upcoming W3C Workshop on Machine
Learning ...

Google Earth is a repository that could be mined for analysis to
identify the accessible route from point A to point B. "Accessible
route" might be different for a blind person's requirements, vs the
requirements of someone using a wheel chair.

Just one thought.

The Workshop info is here:

https://www.w3.org/2020/06/machine-learning-workshop/


Best,

Janina

-- 

Janina Sajka
https://linkedin.com/in/jsajka

Linux Foundation Fellow
Executive Chair, Accessibility Workgroup:       http://a11y.org

The World Wide Web Consortium (W3C), Web Accessibility Initiative (WAI)
Co-Chair, Accessible Platform Architectures     http://www.w3.org/wai/apa