NVDA add-on CloudVision


juergenkohler23@...
 

Hello,
 
Do any of you have experience with the NVDA add-on CloudVision (https://github.com/alekssamos/cloudvision)?
Images located on websites are described relatively reliably; but when I want to analyse images on my desktop/PC, the add-on somehow doesn't seem to work...
Are you making the same observation or am I just doing something wrong in the operation of the add-on?
 
I would be very pleased to receive answers!
 
With best regards


alexey
 

Hello. I am the developer of this add-on.
Make sure that the window is expanded to full screen.
Try to go one level higher in object navigation (NVDA+NumPad8)


juergenkohler23@...
 

Hi alexey,
 
thank you for your answer!
I have tried the full screen, but it doesn't seem to be enough yet. Often, after starting the recognition, you get a message that the recognition is complete, but still nothing is read out. If I place the mouse cursor on the image and drag the NVDA cursor to the mouse cursor, both the font and the image are analysed in full screen mode. Is there also a way to do this without the mouse cursor, e.g. with the NVDA focus (CTRL+Num-). I have tried this procedure, but unfortunately without success...
 
With best regards!


alexey
 

I don't know that anymore, I can't help. Perhaps someone else will tell you.

The problem is not in my add-on, but in the navigator object. The add-on does not capture the entire screen, but only the navigator object.


Pranav Lal
 

Hi,

 

I have opened an image in paint and have got the navigator object where I think the image is. I hit ctrl + nvda key + I and am told that the analysis is completed.

 

What do I do next? How do I read the descriptions?

Pranav


alexey
 

After the phrase the analysis is completed, there should be a result immediately. It should automatically say.


alexey
 

https://visionbot.ru/img/ari.jpg
Try opening this link in the Firefox browser and recognizing the graphics there.
If it works, then the problem is in the navigator. I just really don't know how to put the navigator focus on the image.


harshad solanki
 

hey guys, could anybody provide the download link for cloud vision?

On 15/01/2022, alexey <aleks-samos@...> wrote:
https://visionbot.ru/img/ari.jpg
Try opening this link in the Firefox browser and recognizing the graphics
there.
If it works, then the problem is in the navigator. I just really don't know
how to put the navigator focus on the image.






alexey
 


harshad solanki
 

thanks alexey, i got it.


Joseph Weakland
 

so what does this clud vision thing do

On 1/17/2022 8:46 AM, harshad solanki wrote:
thanks alexey, i got it.

On 17/01/2022, alexey <aleks-samos@...> wrote:
https://github.com/alekssamos/cloudvision/releases/download/v3.0.0.0/CloudVision-3.0.0.0.nvda-addon





--
to check out my podcast on anchor, visit the RSS feed of

https://anchor.fm/s/999e2c/podcast/rss

Joseph Weakland
email
josephweakland@...
sent from my windows 10 pc


alexey
 

Describes images, for example, a 23-year-old girl with long blond hair,
reads texts from photos
and recognizes QR codes.
There is a translation of the recognized text into another language

There is also a bot in Telegram t.me/visiontest1bot


Sarah k Alawami
 

Seems like it sort of does the same thing windows OCR would do. I used this add on, or a similar one years ago before windows OCR existed.

-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of Joseph Weakland
Sent: Monday, January 17, 2022 6:55 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] NVDA add-on CloudVision


so what does this clud vision thing do

On 1/17/2022 8:46 AM, harshad solanki wrote:
thanks alexey, i got it.

On 17/01/2022, alexey <aleks-samos@...> wrote:
https://github.com/alekssamos/cloudvision/releases/download/v3.0.0.0/
CloudVision-3.0.0.0.nvda-addon






--
to check out my podcast on anchor, visit the RSS feed of

https://anchor.fm/s/999e2c/podcast/rss

Joseph Weakland
email
josephweakland@...
sent from my windows 10 pc


Sarah k Alawami
 

Oh, interesting. Has this been submitted for review for the nvda official site in case someone wants to use it? Also how secure is the AI in terms of describing images just in case anyone needs to know that information.

-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of alexey
Sent: Monday, January 17, 2022 7:45 AM
To: nvda@nvda.groups.io
Subject: Re: [nvda] NVDA add-on CloudVision

Describes images, for example, a 23-year-old girl with long blond hair, reads texts from photos and recognizes QR codes.
There is a translation of the recognized text into another language

There is also a bot in Telegram t.me/visiontest1bot


alexey
 

I have made a new version of  the CloudVision  add-on
v3.0.0.1 Latest
Added recognition of files that are selected in Windows Explorer.
The idea is copied from the "Nao (NVDA Advanced OCR)" addon.
Press NVDA+CTRL+I in Explorer on the selected file without opening this file.
Try.
The original image will be sent for recognition.
https://github.com/alekssamos/cloudvision/releases/download/v3.0.0.1/CloudVision-3.0.0.1.nvda-addon


Pranav Lal
 

Hi,

I tried the updated add-on, hitting nvda key + control + I does nothing.

Pranav

-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of alexey
Sent: Saturday, January 15, 2022 1:29 PM
To: Pranav Lal <pranav@...>
Cc: nvda@nvda.groups.io
Subject: Re: [nvda] NVDA add-on CloudVision

After the phrase the analysis is completed, there should be a result
immediately. It should automatically say.


Pranav Lal
 

I solved that problem by assigning a gesture but when I run the add-on on an
image, I hear the beeps because I have enabled them and am then told that
object of non type is not iterable.

-----Original Message-----
From: pranav@... <pranav@...>
Sent: Thursday, January 20, 2022 11:11 AM
To: 'nvda@nvda.groups.io' <nvda@nvda.groups.io>
Subject: RE: [nvda] NVDA add-on CloudVision

Hi,

I tried the updated add-on, hitting nvda key + control + I does nothing.

Pranav
-----Original Message-----
From: nvda@nvda.groups.io <nvda@nvda.groups.io> On Behalf Of alexey
Sent: Saturday, January 15, 2022 1:29 PM
To: Pranav Lal <pranav@...>
Cc: nvda@nvda.groups.io
Subject: Re: [nvda] NVDA add-on CloudVision

After the phrase the analysis is completed, there should be a result
immediately. It should automatically say.


alexey
 

Yes, this problem sometimes happens, just try a few times after a few minutes, or select a file where the size is smaller


juergenkohler23@...
 

Hello alexey,
 
I have also tested your new addon version in the meantime. However, it does not work for me. I don't get any message announced.
With the old version, the image recognition on the internet worked, i.e. it is probably due to the navigator object.
If anyone else has a tip on how the new version works, I would be very grateful!


harshad solanki
 

could anybody explain, how to use it?

On 20/01/2022, juergenkohler23@... <juergenkohler23@...> wrote:
Hello alexey,

I have also tested your new addon version in the meantime. However, it does
not work for me. I don't get any message announced.
With the old version, the image recognition on the internet worked, i.e. it
is probably due to the navigator object.
If anyone else has a tip on how the new version works, I would be very
grateful!