[tabi] Smartphone cameras

  • From: "Lynn Evans" <evans-lynn@xxxxxxxxxxx>
  • To: <tabi@xxxxxxxxxxxxx>
  • Date: Tue, 23 Aug 2011 10:38:48 -0400

Link to article, with text to article below link:

Smartphone cameras bring independence to blind people

19 August 2011 Last updated at 03:19 ET By Damon Rose Editor, BBC Ouch!
VizWiz puts out the user's query to a panel of volunteer helpers
Snapping an image with your smartphone camera brings more than just a
pretty picture if you are blind. With the right app, it can increase
your independence.

Knowing what food is inside a packet or details about the post which has
just arrived on your doormat are everyday things that most people take
for granted.

Blind people have traditionally sought this kind of visual information
from family and friends, or from an employed personal assistant. But
this has meant having to fit in with other people's time or spend
significant money on help. Now there are an increasing number of

As smart phones become more accessible, some with built in speech and
Braille output, it is possible for people with sight loss to get slivers
of visual assistance when there's no one else around to ask.

Want to know what colour your shirt is? Use a colour detector app. Want
to know if it is still daylight outside? Use a light detector app. Want
to read a notice on your work's noticeboard? Use a text recognition app,
of course.

What's in this jar?
The most recent visual assistance product to hit the app store is
VizWiz. As well as giving you automated image recognition from
intelligent software, it throws your questions open to a small band of
volunteers standing-by on the internet - a human cloud, willing to
donate ten seconds of their time here and there to describe photos which
come in.

On its website, the VizWiz is described as: "Take a Picture, Speak a
Question, and Get an Answer".

The free app and service, developed by the University of Rochester in
New York, has received between ten and 12 thousand questions in its
first two months. The volunteers are made up of staff and students who
receive a sound alert when a question arrives, either via Twitter, text
message or the web. They tap in a response which is received by the
original sender.

"The most popular type of question is a product that they have which has
text written on it, a label with instructions. People want to know what
it says, how to cook it or when it expires," said Professor Jeff Bigham,
the man behind the service.

"We can very clearly track the time of day," explained Prof. Bigham.

"In the morning people are asking about clothing, the colour or pattern.
A few people ask if their shirt matches their pants."

"Around one or two eastern time we start getting questions about wine
from what we assume is the UK, asking what label, what year, that kind
of thing."

It is this kind of subjective answer that a piece of software can't give
and that a human service can. But humans need sleep. Prof. Bigham admits
that, though computer scientists are famed for staying up very late, the
6am to 7am timeslot can be a bit difficult to fill with volunteers from
the university.

Human cloud
"It's a really exciting time to work in access technology. A great new
resource is that there are people out there on the web. Everyone is
connected and we can do a lot of interesting things with it," he said.

"People have been throwing around terms like Human Cloud for a while,
and Crowd in the Cloud.

"A lot of work which happened in crowd sourcing before it, took time.
Like Wikipedia, it 'took time' for articles to emerge. What's
interesting with our service is the realtime aspect of it. Someone out
there needs help from the cloud and, in almost real time, they get it."

Users know that it is humans at the other end and this has generated
some "crazy" questions that could never have been answered by automated
recognition software.

"We had one person who kept taking a picture of the sky and asking 'what
is this"' every 5 minutes for a couple of hours," said Prof. Bigham.

"I had no idea what was going on. It also happens we loosely monitor
Twitter. Someone later tweeted 'VizWiz just helped me watch the

Blind photography
In a perhaps unexpected 21st century development, blind people are now
finding they need to learn the basics of photography in order to take
advantage of the growing number of text and image recognition services
on smart phones.

How do you hold the camera up? And how close do you put it to the object
you want to know more about? Angles, perspective, distance and light,
are concepts that don't come naturally to people who have never been
able to see.

The oMoby app is capable of recognising products from a photograph Steve
Nutt is an IT consultant in Hertfordshire who has been blind since
birth. It took him two weeks to master how to frame a shot which he does
in a very functional way, quite different to how sighted people would do

He explains: "If you're taking a picture of, say, a tin, you need to
make sure you get the whole tin in there. I would stand it up so you get
all the sides with the label and snap from about 8 inches above it.

"If you are taking a picture of some text on a piece of paper,
centralise the camera and lift it up about ten inches. Keep your hand
dead straight and dead still when taking the image.

"You have to also bear in mind the size of the thing you're taking the
picture of. the smaller the thing, the closer you need to be to it ...
I'd be lying if I said it was easy."

Jeff Bigham's team sees the results of the camerawork coming from users
like Steve. Not everyone gets it right with their first shot.

"We definitely get a few attempts sometimes. It's not always easy to
frame the photos. Sometimes the centre is out of the photo. if they're
asking what is on a can of soup label, we generally say 'we can't tell
what this is, the label is likely on the other side of the can'."

Other related posts: