Ancestry Saturday: Using Photo Resemblance Tools For Genealogy Research

It's Saturday.  I'm writing about my genealogical research hobby.

My latest celebrated  research find came from a book--Boys of '98: A History of the Tenth Regiment Ohio Volunteer Infantry.  I now know that the Library of Congress has a copy online, though I found the book itself at the Columbus Public Library courtesy of an extremely helpful librarian and some dumb luck.

My great grandfather Andrew Alfred Baker was very likely in this photo.  He was a musician in the band of the Tenth Regiment of the Ohio Volunteer Infantry for the Spanish-American War.

I was extremely pleased with the find, but, alas, which one is he?  There's no captioned list with the photograph.

I asked my aunt, his nearest living relative, and she'd never seen a photo of him to which to compare. He died 15 years before she was born and no photographs passed down.  Of course, I don't know what he looked like either.

A later page in the book says he was among the trombonists in the band.  That narrowed it down to six.

The likely three youngest trombone players.

I was able to find ages for the other trombone players.  He was one of three of the youngest members and, in my opinion, age distinctions are visible enough in this photo to narrow down to three distinct bandsmen.

I polled some folks.

All six I asked picked the same person--the trombone player with no facial hair is the unanimous pick to have been my great grandfather.



The unanimous pick.

Of course, Presidents make national policy choices on overnight polling, but choosing what my great grandfather looked like for my family tree collection requires more science. Right?

There are ways to get through this somewhat scientifically.  Here's a few I pondered and tried:



The unanimous pick among trombone players side by side with my grandmother (left) and great aunt (right).

I compared his photo to my grandmother and Aunt and did a little amateur biometrics.  I think there's reason to think the pick among trombone players could be the father of these two.  The noses are extremely similar.  The mouths and chins are simliar.  The space between the eyes is close in comparison.

Using facial recognition software to find a resemblance was worth a try.

Windows 7's Windows Live Photo Gallery program has a way to tag faces and then have it search other photos on the PC that it's algorithm believes match that face.  I did that, thinking that if it tried to match my grandmother's photo to his that there was some increased likelihood of a match.  It didn't take.

I tried MyHeritage.com's feature advertised to be the first family genealogy service to deploy face recognition software matching faces. My experience from browsing is that they haven't graduated to this type of relatives comparison, yet.  I'll share some more on this one in the future.

The Resemblance Lens feature on my kids' Nintendo DSI XL's was my next try. I've found that capability pretty impressive in the past matching my kids to their uncle, even. I couldn't get it work in this case though. I think it doesn't like black and white photos.

Technology is getting close though I think.  It can't be long before someone figures out how to deploy facial recognition software to tag relatives.

The best, most scientific option left is trying to find an identified photograph somewhere.  I'm going to keep trying, of course.

In the meantime, I opted to post the photo of the polled favorite trombone player with my great grandfather in my Ancestry.com family tree.

Do you think the match is right?