Re: Windows OCR update in nvda


Quentin Christensen
 

At this stage, we are using the OCR supplied by Windows.  There are entire companies dedicated to OCR solutions, so it's unlikely to be an area we'll directly invest in ourselves.  For reading longer text we would still suggest it is worth considering a commercial OCR solution.  Of course, if Microsoft make better features available, we will be sure to take advantage of them.

On Mon, Sep 10, 2018 at 7:44 PM, Robert Mendoza <lowvisiontek@...> wrote:
Hi, Quentin.


I could concur to his message. Sometimes works, but most of the time saying just blank and not sure specifically underlying on it to recognize fully. In addition,   let say the focus is indicated on the graphics NVDA tells something text content then pressing the Insert or Caps plus R it reads sometimes but not too close when announcing the text. in accurate manner. That is why I am wondering if there is a possibility in the future release to develop this function on the OCR thing. But this is just merely just a thought if you could consider.


Robert Mendoza



On 9/10/2018 12:36 PM, John Isige wrote:
If it's an image on Facebook, it sometimes helps to follow the link and
then pick the "view full size" link. It doesn't always make a
difference, but I've found that sometimes viewing it full size will get
me some text where simply highlighting the graphic, say with the arrows,
and hitting NVDA-r doesn't give me anything.


On 9/9/2018 23:18, Quentin Christensen wrote:
Hi Robert,

In Windows 10, navigate to the graphic with the text, then press
NVDA+r (the NVDA key will be either INSERT or CAPS LOCK, depending on
your settings).  NVDA should report that it is recognizing, and then
"done".  Once done is announced, you can use the regular navigation
and reading keys to read the text, as you would other text on the
page.  When done, press escape to return to the web page.

For anyone reading this using Windows 7 or 8, it's very similar using
the add-on, press NVDA+r but then you use the text review commands
rather than regular navigation.

Kind regards

Quentin.

On Mon, Sep 10, 2018 at 2:10 PM, Robert Mendoza
<lowvisiontek@... <mailto:lowvisiontek@...>> wrote:

    Hi, Quentin.

    From time to time I alternate using of both beta and 2018.2.1 in
    my Windows 10.
    Unlikely could not give you the picture because it was posted in
    the facebook of my friend.
    Although facebook has this technology to describe as Alternate
    text or Alt Text but seemingly somehow help me a bit. But for the
    Windows 10 what is the key command for OCR for describing images?
    I appreciate if you could give those it here.

    Robert Mendoza


    On 9/10/2018 9:31 AM, Quentin Christensen wrote:
    Hi Robert,

    As Brian noted, NVDA uses Microsoft's built-in OCR in Windows 10
    (which is why it is only built-in on Windows 10 and you need the
    add-on for earlier Windows versions).

    Microsoft are evolving Windows 10 with their regular updates, and
    presumably this also involves improving the OCR.

    In the meantime, have you got an example image or link which
    isn't working for you?

    Kind regards

    Quentin.

    On Sat, Sep 8, 2018 at 6:38 PM, Brian's Mail list account via
    Groups.Io <bglists=blueyonder.co.uk@groups.io
    <mailto:bglists=blueyonder.co.uk@groups.io>> wrote:

        Inwindows 10 though this is all down to Microsofts cloud OCR.
        All nvda does is tap into it as far as I'm aware.
        Brian

        bglists@... <mailto:bglists@....uk>
        Sent via blueyonder.
        Please address personal E-mail to:-
        briang1@... <mailto:briang1@....uk>,
        putting 'Brian Gaff'
        in the display name field.
        ----- Original Message ----- From: "Robert Mendoza"
        <lowvisiontek@... <mailto:lowvisiontek@...>>
        To: <nvda@nvda.groups.io <mailto:nvda@nvda.groups.io>>
        Sent: Saturday, September 08, 2018 5:15 AM
        Subject: [nvda] Windows OCR update in nvda


            Hi, wondering if there is development or advancement on
            the Windows OCR in nvda or with the future release for
            recognition of graphics with text content?
            It's been a while I have not seen update from this and
            tried to test this as well to beta version of nvda and
            unlikely the OCR function still not working perfectly
            because some text could read and some text could not
            recognize fully like giving such blank when pressing the
            Insert/NVDA plus R


            Also, could the OCR recognizes fully the graphics for the
            image itself? Thanks.
            Robert Mendoza












    --
    Quentin Christensen
    Training and Support Manager

    Official NVDA Training modules and expert certification now
    available: http://www.nvaccess.org/shop/
    <http://www.nvaccess.org/shop/>

    www.nvaccess.org <http://www.nvaccess.org/>
    Facebook: http://www.facebook.com/NVAccess
    <http://www.facebook.com/NVAccess>
    Twitter: @NVAccess




--
Quentin Christensen
Training and Support Manager

Official NVDA Training modules and expert certification now available:
http://www.nvaccess.org/shop/

www.nvaccess.org <http://www.nvaccess.org/>
Facebook: http://www.facebook.com/NVAccess
Twitter: @NVAccess












--
Quentin Christensen
Training and Support Manager

Official NVDA Training modules and expert certification now available: http://www.nvaccess.org/shop/

Facebook: http://www.facebook.com/NVAccess 
Twitter: @NVAccess 

Join nvda@nvda.groups.io to automatically receive all group messages.