After my previous post about reading devices with displays, Peter
Meijer of "voice learning edition" asked me to elaborate on the
Smith-Kettlewell project to develop a system which would read
displays on consumer products via an inexpensive web camera and a
computer.

It is not 'my" project, so I must be rather careful what I say,
I will ask the project leader, Dr. Alan  Yuille if he has further
information for public release.

It is the goal of this project to develop software to run on
inexpensive "off the shelf" hardware to read things like
appliances as automatically and interactively as possible.

If the system works as planned, you would point the camera at the
device about where you thought the display was, and the software
would do all the work of finding the display, identifying
important and changing features on it and "read" it to you.  It
is very likely that some training would be necessary whenever you
tried a new device, but we have no idea yet how this training
would be accomplished.

It is our belief that only with this level of machine
intelligence can a user interact with modern consumer
electronics in a truly accessible way.

Changes in the consumer electronics market lead us to believe
that many appliances are going to become more and more
inaccessible unless we have some stand alone solution such as
this.  One example is that you can not easily buy a modern oven
which has a familiar thermostat of the type we have all learned
to mark and adjust.  Most modern stoves feature a dial that
rotates infinitely in either direction and controls a digital
temperature readout.  Like so much of this visually
overstimulated world, you have to see it to use it.

Sorry that's all I can tell you for now, because the "AI" guys are
still trying to make it work.

Dr. Meijer's system depends on the user learning an entirely NEW
CODE which attempts to relate audio sounds to the visual image.
Just as with much previous work which is based on the belief that
blind people can learn to "see" using non visual senses, this
work demands that people learn entirely new ways of perceiving
the world.  Particularly with people who have never had sight,
you must learn to understand images of which you have no built in
conception using a language you don't know. The assertion that
there is an intuitive relationship between audio pitch and space
with vision is unsupported except in very very simple examples.

We here at Smith-Kettlewell developed a successful audible
oscilloscope which uses a simple version of just such audio
representation, however our extensive experience with attempts to
use this system with complex visual images does not lead us to
believe that it will work.

There is wide belief that the brain can learn to be "plastic"
such that it can learn to "see" with portions of its self which were
not intended for that purpose, and that the audio perception
capabilities of blind people are somehow magical and can be
trained to process unfamiliar complex images in an entirely new
way.

Despite much misunderstood rumor on this subject I have seen no
actual scientific evidence to support this claim, and my personal
experience does not support it either.

I have listened to several images on Mr. Meijer's web site and
find them entirely meaningless despite the fact that I use the
Smith-Kettlewell oscilloscope regularly.

It can be said in favor of Peter's work that it is certainly
harmless, as long as you don't try crossing streets with it, and
certainly inexpensive, so more power to him and those who are
interested.

You don't have to believe this, but my opinions are, I hope,
not based in the "not invented here" syndrome, I  make
every attempt to be sure that I judge others inventions based on
whether or not I think they will work, not on whether I wish I had
invented them my self.

After all if someone else builds something cool, then I don't
have to do all that work my self.

Remember this is one opinion, not to be taken as a representation
of anything more than that, however it is based on nearly 20
years of experience in this job dealing with all kinds of access
issues.

Now, let's see a goodly number of blind folks make use of Peter's system
and prove me wrong <G>  Keeping in mind that you can almost
always find one subject who can use almost anything, there isn't
space here for the examples I could give which would make you all
laugh.

Tom Fowle
Rehabilitation engineer,Embedded systems developer
The Rehabilitation engineering research center
Smith-Kettlewell Eye Research Institute
San Francisco
[log in to unmask]


VICUG-L is the Visually Impaired Computer User Group List.
To join or leave the list, send a message to
[log in to unmask]  In the body of the message, simply type
"subscribe vicug-l" or "unsubscribe vicug-l" without the quotations.
 VICUG-L is archived on the World Wide Web at
http://maelstrom.stjohns.edu/archives/vicug-l.html