Has anyone tested this addon?


I was wondering i anyone has tested the image description addon and how good is it?

From: Manshul Belani <manshulvbelani@...>
To: nvda-addons@nvda-addons.groups.io

Hi All,

Dear All,

I have created an addon for NVDA which provides description for graphics/images present on the web pages. The add on uses Microsoft's Computer Vision cognitive API to get a description of the image provided to it. Please find below the steps to use the add on and the dropbox link to download it.

Steps to use add on:

Step 1: Generate a subscription key for using Microsoft Computer Vision cognitive API from the link below:


Please note that the key has to be created for the Computer Vision API. Currently Microsoft provides the option to create a trial key for 30 days.

Step 2: Download and Install the NVDA Addon named Assistech_IIT_Delhi from the link “” or from download button above.

For Addon installation instructions visithttps://www.nvaccess.org/files/nvda/documentation/userGuide.html#toc94


Step 3: From the add-on menu, select the options add API key and provide your subscription key for the cognition API.


Step 4: If you are behind a proxy, from the add on menu (the add-on is named as Assistech_IIT_Delhi) , select the options add proxy details and provide your proxy IP and port. 

Step 3 and step 4 only need to be performed the first time or whenever either of your proxy details or subscription key changes.


Step 5: While navigating a webpage, whenever a graphic is encountered and is in focus, use keyboard shortcut NVDA+g. The image description and the confidence level returned b the API are provided as audio output. Use NVDA+gg to get the results (including the dominant colors present in the image) in a browseable message box. (Fetching results takes a few seconds.). If no result is obtained from API, appropriate speech message is returned.

I request you all to kindly test the add on and provide your feedback.

Manshul V Belani

This message is transmitted on 100% recycled electrons.