[Njtechdiv] iDentifi app

[Mario Brusco]

  A Toronto teen is hoping to change the lives of visually impaired 
people around the world with a new app that can identify virtually

any object with the quick tap of an iPhone or iPad.

  Anmol Tukrel, a 17-year-old grade 12 student at Holy Trinity School in 
Richmond Hill, has always been fascinated with technology,

particularly artificial intelligence. By the time he was in grade 7, he 
was already teaching  himself how to code.

  Growing up, Tukrel often travelled to Pune, India to visit his aunt 
who worked at the K. K. Eye Institute, a hospital dedicated to

providing eye care for people who can't afford it. That experience, 
combined with an internship at a startup that uses computer

vision to make products for advertising firms, led him to the perfect 
idea for a Canada-Wide Science Fair project.

  "I thought I could use computer vision for a more humanitarian use, 
and  help visually impaired people," he said.

  Tukrel's iPhone app, iDentifi, allows users to take a photo of 
virtually  any object, and then describes that item in great detail back

to the user.

People can also take photos of text and have it read back to them, in 
one of 27 languages. Tukrel hopes it makes every day tasks 

like picking out the can of pop you want  easier for people who are 
visually impaired.

  Jason Fayre, the head of accessibility and assistive technology at the 
Canadian National Institute for the Blind, tested out the app

and, although there are similar apps on the market, gave it a rave review.

  "I'm extremely impressed, especially that it was written by a grade 12 
person," he said. As a blind person himself, Fayre said

iDentifi would make his life in easier when trying to identify things in 
the kitchen.

  "If I don't know what a particular can of something is, being able to 
take  a picture and have that information read back to me in

great detail is very useful," he said.

  It took Tukrel more than a year to develop the app, a process that 
involved months of painstaking research and enough code to fill

a two-inch binder.  He had initially planned on making his own 
convolutional neural network computer speak for the data structure

used to make the program that recognizes objects. Eventually, he opted 
to integrate existing programs.

  Tukrel casually speaks about computer vision, convolutional neural 
networks, and application program interfaces as though he

were a university graduate of computer science - not an 
about-to-graduate high schooler.

  "I've always liked technology, but as much as I like playing video 
games  and using different apps, I wanted to be able to make

them myself," Tukrel  said.

  For Tukrel, the work doesn't stop now that the science fair is over.

  He has already met with various organizations to get feedback on the 
app, and plans on making tweaks to improve the user

experience. So far, the app has been downloaded by several thousand 
people and is being used in 60
  countries. And, it's free, something Tukrel doesn't plan on changing.

  "I want people who are visually impaired to use it without thinking of 
the financial consequences of doing so," he said. "We have

such great  technology and I think it's important that everyone has 
access to it."


https://www.thestar.com/news/gta/2016/11/28/toronto-teen-uses-app-to-give-visually-impaired-a-new-look-at-the-world.html

and here's a short youtube on how it works.

https://www.youtube.com/watch?v=zpAZHyZIzyg