[Blindmath] Facial recognition -- food for thought

Carolyn MacLeod Carolyn_MacLeod at ca.ibm.com
Fri Mar 30 19:42:50 UTC 2012


I guess this must have been in response to my post about the face-tagging 
software?
If so, I feel compelled to set the record straight by saying that I did 
not in any way mean to imply that folks should sit by and wait for big 
companies to do stuff.
I only wanted to share what I had discovered when I tried the face-tagging 
in Photo Gallery, because I was pretty impressed by its performance.

If you had been able to read my mind while I was posting, you would have 
heard something like, "Cool topic! Here's something I learned on the 
subject."
(Now there's a challenging math problem for the future... analyzing 
people's thought data <grin>).

Please allow me to elaborate on my Photo Gallery experience a bit more, 
just for the fun of sharing an exploration of new technology.  :)
I have thousands and thousands of photos on my computer, and I ran Photo 
Gallery partly to see what it could do, and partly because my husband said 
he needed a photo of himself for a Bio.
Richard, you are correct: it is not real-time, but it is almost fast 
enough - it was processing about 3 or 4 photos per second, maybe more.
The accuracy depended on whether or not the person has distinctive facial 
features.
For example, after showing me several faces which I identified as my 
husband, it suggested over 800 more of him without a single error!
This included some photos that were dark, grainy, small, or with an angled 
or tilted view.
I was surprised at the accuracy, and then I realized that my husband is 
the only person in all of my photos who has a dark mustache and glasses, 
so the software obviously optimizes for those.
It had a more difficult time with my 9-year-old son, however, getting only 
about 80% of his photos correct, as it quite often suggested one of his 
cousins, or occasionally, his friends.
It had some real bloopers, like once it suggested that his Grandma might 
be him.
The algorithms behind all this must be really cool, but I do have a 
suggestion for them... look at the date the photos were taken.
If you have a photo of a person in their seventies and one of a child, and 
both photos were taken on the same day, it is not possible for them to be 
the same person.  :)

I want to share one more thing that I learned, if you will permit me. I 
believe it pertains to the topic.
After Richard posted about the real-time facial recognition, I searched 
the App Store, just for fun, to see what iPhone apps might be out there.
Most of them are simple entertainment apps that find a face in a photo and 
decorate it in some way.
One app, called Cyborg, did claim to have real-time facial recognition, 
looking up and displaying the identified person's name, gender, home town, 
birthday and current location, and whether or not they are smiling.
I don't recommend the app because the reviews are not very good, but what 
I found interesting was that they get the data by requesting that you log 
in to your Facebook account.
The advantages of that are that the user does not need to carry any photos 
and other data around, and that there is an enormous quantity of data that 
your Facebook friends and family have already entered for you.
The disadvantages, of course, are the lack of privacy issues, and the fact 
that the data comes from other people and sources and therefore might not 
be as trusted as if it came directly from the user.

One last thing. Apparently Google is planning to sell something called 
Google Goggles by year end.
They have a camera, and a possibly interesting head-tilting mechanism for 
selecting things, but unfortunately they do not seem to have an earpiece 
(I could be wrong).
However, if they catch on, then it may just be considered cool to be 
wearing high-tech headgear.  <grin>
http://bits.blogs.nytimes.com/2012/02/21/google-to-sell-terminator-style-glasses-by-years-end/

Carolyn



From:
Richard Baldwin <baldwin at dickbaldwin.com>
To:
Blind Math list for those interested in mathematics <blindmath at nfbnet.org>
Date:
03/28/2012 05:56 PM
Subject:
Re: [Blindmath] Facial recognition -- food for thought
Sent by:
blindmath-bounces at nfbnet.org



Steve,

You wrote "My point here is not to scold as we have been sort of scolded,"

My earlier post was not intended to scold you or anyone else in the blind
community.

I pointed out when I made the (apparently offending) statement that it
might not make sense to some readers due to those readers not having seen
all of the previous posts due to distribution issues with earlier posts.

My post was in response to a post made earlier by a SIGHTED individual, I
repeat a SIGHTED individual, who in effect said that we shouldn't expend
efforts doing things that the big companies might do better if we are
patient and willing to wait long enough.

My point is that if I had the need, (and perhaps there is no need), I
wouldn't be patient and wait for something that the big companies might do
for me at some time in the future. Instead, if possible, I would do it for
myself.

Dick Baldwin

On Wed, Mar 28, 2012 at 4:36 PM, Richard Baldwin 
<baldwin at dickbaldwin.com>wrote:

> I don't understand the question.
> Dick Baldwin
>
>
> On Wed, Mar 28, 2012 at 2:00 PM, Pickrell, Rebecca M (TASC) <
> REBECCA.PICKRELL at tasc.com> wrote:
>
>> Dick,
>> You percieve this as a problem. Why?
>> I don't. The technology is interesting, but it doesn't mean not seeing
>> someone's face is a problem.
>>
>>
>> -----Original Message-----
>> From: blindmath-bounces at nfbnet.org [mailto:blindmath-bounces at nfbnet.org
]
>> On Behalf Of Richard Baldwin
>> Sent: Wednesday, March 28, 2012 9:04 AM
>> To: Blind Math list for those interested in mathematics;
>> seeingwithsound at freelists.org
>> Subject: Re: [Blindmath] Facial recognition -- food for thought
>>
>> Apologies to those who haven't seen all of the response postings due to
>> cross listing of the original post and to whom this post probably won't
>> make sense.
>>
>> If I were blind and could purchase all of the necessary hardware,
>> including
>> gigabytes of USB storage for a few hundred dollars, I wouldn't be 
sitting
>> around waiting for Microsoft, Google, Apple, or some other large 
company
>> to
>> solve my problems. I would be working to solve those problems myself.
>>
>> If Bill Gates and Michael Dell had that attitude, there wouldn't be a
>> Microsoft. IBM would "own" the computer market on a worldwide basis.
>>
>> Dick Baldwin
>>
>> On Wed, Mar 28, 2012 at 7:03 AM, Carolyn MacLeod <
>> Carolyn_MacLeod at ca.ibm.com
>> > wrote:
>>
>> > Not sure if this is useful, and you probably already know, but
>> Microsoft's
>> > Windows Live Photo Gallery and Picasa and iPhoto (and some others) 
have
>> a
>> > feature called face-tagging.
>> > They work surprisingly well, although they need someone to identify
>> photos
>> > and confirm guesses, and they do make some ridiculous mistakes
>> sometimes.
>> > If you have Windows 7 or Vista, you may already have Windows Live 
Photo
>> > Gallery installed - to find out, open the Start menu and type 
"Windows
>> > Live Photo Gallery" in the Search field, then Enter.
>> > If not, here's a link with a download button:
>> >
>> 
http://windows.microsoft.com/en-US/windows7/products/features/photo-gallery

>> > Carolyn
>> >
>> >
>> >
>> >
>> > From:
>> > Richard Baldwin <baldwin at dickbaldwin.com>
>> > To:
>> > BlindMath Mailing List <blindmath at nfbnet.org>,
>> > seeingwithsound at freelists.org
>> > Date:
>> > 03/27/2012 11:05 PM
>> > Subject:
>> > [Blindmath] Facial recognition -- food for thought
>> > Sent by:
>> > blindmath-bounces at nfbnet.org
>> >
>> >
>> >
>> > Most of the math courses that I completed in public school and 
college
>> > mainly involved completely deterministic concepts such as finding
>> unknown
>> > in sets of algebraic equations, proving theorems in geometry, proving
>> > identities in trigonometry, solving triangle problems using
>> trigonometry,
>> > differentiating functions, integrating functions, playing around with
>> > solids of revolution, etc. They were all good exercises for the brain
>> but
>> > were not very close to real-world problems.
>> >
>> > When I made it into engineering college, the problems and their
>> solutions
>> > were closer to the real world but only barely so.
>> >
>> > When I completed my first engineering degree and went to work in the
>> real
>> > world, I learned very quickly that problems in the real world are far
>> from
>> > deterministics. In other words, there are very few problems in the 
real
>> > world that have deterministic solutions. Problems in the real world
>> > usually
>> > involve a mix of mathematics, statistics, physics, engineering, 
computer
>> > science, and other technologies, and there is rarely a single correct
>> > solution for any problem.
>> >
>> > Furthermore, the solution to most problems requires the design and
>> > implementation of complex mathematical algorithms, and those 
algorithms
>> > are
>> > most commonly implemented using a computer of some sort. (In my 
opinion,
>> > every student that receives a technical degree should be required to
>> learn
>> > to program well in at least one programming language.)
>> >
>> > By now you must be wondering where this is all heading.
>> >
>> > I saw on TV today that shopping malls and large department stores are
>> > installing electronic billboards that use facial recognition to 
display
>> > advertisements that are likely to be of interest to those persons who
>> can
>> > see the billboard.
>> >
>> > I have no idea what the facial recognition algorithm is for 
categorizing
>> > the viewers in a way that allows for a selection of appropriate
>> > advertisements. However, this tells me that the algorithm doesn't
>> require
>> > a
>> > supercomputer to implement. The algorithms must be implemented using
>> > modestly priced computer hardware. Otherwise, they would be too
>> expensive
>> > to include in such billboards.
>> >
>> > This makes me wonder if it might be possible to use a small portable
>> > computer to develop a system that will do facial recognition on 
people
>> > whose faces appear in the field of view of a miniature video camera
>> > embedded in eyeglass frames and to speak information about those 
people
>> to
>> > the wearer of the glasses.
>> >
>> > Science fiction? Maybe so and maybe not.
>> >
>> > Dr. Peter Meijer has demonstrated that it is possible to couple a 
video
>> > camera built into eyeglass frames with a small portable computer and 
an
>> > appropriate software program (The vOICe) and to create soundscapes 
that
>> > some blind users find very beneficial (see
>> http://www.seeingwithsound.com/
>> > )
>> > as they move through the world.
>> >
>> > Not being blind, I can't imagine what it would be like to interact 
with
>> > other people that you can't see. However, it seems to me that it 
would
>> be
>> > beneficial for a blind person to know something about another persons
>> > before a conversation begins. Depending on capability, this could 
range
>> > all
>> > the way from rudimentary information such as the probable sex and 
likely
>> > age of the person, to detailed information such as the identification 
of
>> > prior acquaintances by name.
>> >
>> > Perhaps it is time for a group of blind mathematicians, physicists,
>> > engineers, statisticians, and computer scientists to band together to
>> > produce such a system and to publish it as an open source
>> > hardware/software
>> > system.
>> >
>> > Food for thought,
>> > Dick Baldwin
>> >
>> > --
>> > Richard G. Baldwin (Dick Baldwin)
>> > Home of Baldwin's on-line Java Tutorials
>> > http://www.DickBaldwin.com
>> >
>> > Professor of Computer Information Technology
>> > Austin Community College
>> > (512) 223-4758
>> > mailto:Baldwin at DickBaldwin.com
>> > http://www.austincc.edu/baldwin/
>> > _______________________________________________
>> > Blindmath mailing list
>> > Blindmath at nfbnet.org
>> > http://nfbnet.org/mailman/listinfo/blindmath_nfbnet.org
>> > To unsubscribe, change your list options or get your account info for
>> > Blindmath:
>> >
>> >
>> 
http://nfbnet.org/mailman/options/blindmath_nfbnet.org/carolyn_macleod%40ca.ibm.com

>> >
>> >
>> >
>> >
>> > _______________________________________________
>> > Blindmath mailing list
>> > Blindmath at nfbnet.org
>> > http://nfbnet.org/mailman/listinfo/blindmath_nfbnet.org
>> > To unsubscribe, change your list options or get your account info for
>> > Blindmath:
>> >
>> >
>> 
http://nfbnet.org/mailman/options/blindmath_nfbnet.org/baldwin%40dickbaldwin.com

>> >
>>
>>
>>
>> --
>> Richard G. Baldwin (Dick Baldwin)
>> Home of Baldwin's on-line Java Tutorials
>> http://www.DickBaldwin.com
>>
>> Professor of Computer Information Technology
>> Austin Community College
>> (512) 223-4758
>> mailto:Baldwin at DickBaldwin.com
>> http://www.austincc.edu/baldwin/
>> _______________________________________________
>> Blindmath mailing list
>> Blindmath at nfbnet.org
>> http://nfbnet.org/mailman/listinfo/blindmath_nfbnet.org
>> To unsubscribe, change your list options or get your account info for
>> Blindmath:
>>
>> 
http://nfbnet.org/mailman/options/blindmath_nfbnet.org/rebecca.pickrell%40tasc.com

>> CONFIDENTIALITY NOTICE: This message and any attachments or files
>> transmitted with it (collectively, the "Message") are intended only for 
the
>> addressee and may contain information that is privileged, proprietary
>> and/or prohibited from disclosure by law or contract. If you are not 
the
>> intended recipient: (a) please do not read, copy or retransmit the 
Message;
>> (b) permanently delete and/or destroy all electronic and hard copies of 
the
>> Message; (c) notify us by return email; and (d) you are hereby notified
>> that any dissemination, distribution or copying of the Message is 
strictly
>> prohibited.
>>
>> _______________________________________________
>> Blindmath mailing list
>> Blindmath at nfbnet.org
>> http://nfbnet.org/mailman/listinfo/blindmath_nfbnet.org
>> To unsubscribe, change your list options or get your account info for
>> Blindmath:
>>
>> 
http://nfbnet.org/mailman/options/blindmath_nfbnet.org/baldwin%40dickbaldwin.com

>>
>
>
>
> --
> Richard G. Baldwin (Dick Baldwin)
> Home of Baldwin's on-line Java Tutorials
> http://www.DickBaldwin.com
>
> Professor of Computer Information Technology
> Austin Community College
> (512) 223-4758
> mailto:Baldwin at DickBaldwin.com
> http://www.austincc.edu/baldwin/
>



-- 
Richard G. Baldwin (Dick Baldwin)
Home of Baldwin's on-line Java Tutorials
http://www.DickBaldwin.com

Professor of Computer Information Technology
Austin Community College
(512) 223-4758
mailto:Baldwin at DickBaldwin.com
http://www.austincc.edu/baldwin/
_______________________________________________
Blindmath mailing list
Blindmath at nfbnet.org
http://nfbnet.org/mailman/listinfo/blindmath_nfbnet.org
To unsubscribe, change your list options or get your account info for 
Blindmath:
http://nfbnet.org/mailman/options/blindmath_nfbnet.org/carolyn_macleod%40ca.ibm.com







More information about the BlindMath mailing list