Re: RES: [nvda] Microsoft's object recognition technology and Cloud questionsfor NVDA



For everyone else who’re wondering what’s up: yes, it is a cloud-based image recognition technology at work. Deep within is artificial intelligence (AI) network that processes images and sends back textual descriptions. As this is a cloud-based service, internet connection is a must, and NV Access should raise funds to use it if it’ll cost money to do so. Ideally, it should be an add-on first before being considered for inclusion in NVDA.




From: [] On Behalf Of Robert Mendoza
Sent: Tuesday, January 23, 2018 1:58 AM
Subject: Re: RES: [nvda] Microsoft's object recognition technology and Cloud questionsfor NVDA


Hi Alexandre ,

It was not that as clear to recognize some of the pictures but, somehow they are like similar in terms of recognition. I think in the orientation or the platform on mobile is not similar as for desktop or laptop. Though, you may still try the OCR still by activating the narrator, then once you heard the voice of the narrator you simply used the key command to capture the image by pressing Shift plus Caps lock plus D. Again, you have to make sure that the cursor is pointed to the graphics and the narrator gives you some description but, again not that full to describe images with compare of the Seeing AI app for mobile. I still believe that this will be more improves on the future releases of Windows 10. For now that is the only thing I've got for you or perhaps others could shed some details for this one.

Robert Mendoza

On 1/23/2018 5:16 PM, Alexandre Alves Toco wrote:


Are you saying that narrator can now describe a photo’s contente as seeingAi?

If so, it is a very interesting feature!!!

How do you acomplish this?



Enviado do Email para Windows 10


De: Shasa
Enviado:domingo, 21 de janeiro de 2018 14:47
Assunto: [nvda] Microsoft's object recognition technology and Cloud questionsfor NVDA


Developer friends and all NVDA users, greetings.


Microsoft has introduced a description feature that allows Windows 10 to

recognize objects, images, and other non-resolvable content on its

screen reader.

Is this open source by Microsoft? Can not we do this with the same DLL

as we do in the same OCR?


Finally, Python has great capabilities in the visual part. Legends

emerged especially with the development of artificial intelligence. Can

we deal with a cloud system?





Join to automatically receive all group messages.