2015 Facial recognition software?

Georgia Coast

Sergeant
Joined
Feb 7, 2013
Location
Florida
I know that there have been several previous questions about facial recognition software, but they seem to be several years ago. Is anyone on the forum currently using facial recognition software to try to ID ACW photographs?

I have two photographs of a mixed race "body servant" to a Confederate general that can be compared to numerous photographs of that general. The history of the two plus the photographic evidence strongly suggests that they are half-brothers, but I'd like to get other opinions on this, esp. through software if possible. Both men are from my area and I have a strong interest in their story, as they seem to have remained connected after the war. (The slave/"body servant" became a doctor, possibly with help from the general's family.)
 
This product kinda does what you are looking to do, and the price is right. Few caveats:
  • As any predictive application that works with an algorithm it needs to be a. trained and b. have many options to chose from, in order to increase the fidelity of a match. You need to train it by tagging a whole bunch of photos with a whole bunch of people and need many photos of those individuals. Hard to do sometimes with ACW subjects
  • If you don't do that, the likelihood of a match of 2 likes (but not cigar) is very high, but that match is of low fidelity and practically worthless.
  • Something like this might provide data points that a couple of people looked alike, and that it as far as you can go. Even if they do not look alike, they could be brothers (I don't look that much like my brother for example.) To start talking about relationships, you need DNA analysis & matches. Straight forward if they were still around, because they would have had identical Y chromosomes if brothers from the father side (which I presume was the premise,) but...
Fun project though. Good luck!
 
Thank you so much, E. This was not one I was familiar with.

I realize that with only two photos of one of the subjects (Alexander Darnes), though many, many of Kirby Smith, this isn't going to have any scientific proof to speak of, just more circumstantial evidence in one direction or another. The St. Augustine historian who has been researching and writing about Darnes also thinks it is highly likely that Kirby Smith's father was the father of both men. Darnes is a fascinating guy. There's a Wikipedia article on him, but beware -- quite a few things in it are wrong; for example, it says his father was "Violent Pinkney," when in fact his father is unknown (most likely Edmund's father or possibly a white Darne [sic] who was a soldier in the Seminole wars), and his MOTHER is "Violet" Pinkney.

I don't think the DNA matching is a possibility. Darnes didn't have any children, though he did have a niece or nephew, I can't recall which off the top of my head. It was these family members who provided the photos of Darnes to the St. Augustine Historical Society, I believe. His possessions also included a photo of Kirby Smith, signed on the back, that Kirby Smith had given him some time after the war. While the young Kirby Smith doesn't look all that much like Darnes, once he lost his hair they began to look more and more alike. Of' course, Kirby Smith did have children, and in fact, I had a great conversation with the sculptor Maria Kirby Smith, his great-granddaughter. Her sculpture, "Sons of the City,"(which imagines of the two men re-meeting late in life) is fascinating.

Our local SCV has taken an interest in Darnes and recently gave him a new gravestone. I'm appreciative that they've renewed interest in his place in history, but I'd rather see the original broken, old marker than a brand-new one, personally. I hope the old one is being kept somewhere, broken or not.
1-darnes 1876.JPG
1-darnes full.JPG
e-kirby-smith-1824-1893-granger.jpg
 
Thank you so much, E. This was not one I was familiar with.

I realize that with only two photos of one of the subjects (Alexander Darnes), though many, many of Kirby Smith, this isn't going to have any scientific proof to speak of, just more circumstantial evidence in one direction or another. The St. Augustine historian who has been researching and writing about Darnes also thinks it is highly likely that Kirby Smith's father was the father of both men. Darnes is a fascinating guy. There's a Wikipedia article on him, but beware -- quite a few things in it are wrong; for example, it says his father was "Violent Pinkney," when in fact his father is unknown (most likely Edmund's father or possibly a white Darne [sic] who was a soldier in the Seminole wars), and his MOTHER is "Violet" Pinkney.

I don't think the DNA matching is a possibility. Darnes didn't have any children, though he did have a niece or nephew, I can't recall which off the top of my head. It was these family members who provided the photos of Darnes to the St. Augustine Historical Society, I believe. His possessions also included a photo of Kirby Smith, signed on the back, that Kirby Smith had given him some time after the war. While the young Kirby Smith doesn't look all that much like Darnes, once he lost his hair they began to look more and more alike. Of' course, Kirby Smith did have children, and in fact, I had a great conversation with the sculptor Maria Kirby Smith, his great-granddaughter. Her sculpture, "Sons of the City,"(which imagines of the two men re-meeting late in life) is fascinating.

Our local SCV has taken an interest in Darnes and recently gave him a new gravestone. I'm appreciative that they've renewed interest in his place in history, but I'd rather see the original broken, old marker than a brand-new one, personally. I hope the old one is being kept somewhere, broken or not. View attachment 61495View attachment 61496View attachment 61497
I feel the same way about old markers. Just in case anyone planning to add new stones is reading this, what I personally like to see is a new stone (maybe even a larger stone nearby with several people on it) and the old stone reset nearby in a way that protects it.

Some of the family matching on ancestry will link people up across generations and oceans... I know a large number of families with common names who have used it to figure out which Samuel Johnson in 1650 something was their branch. If family members are willing to be tested that's your best bet. Of course it only works if his sibling had the same father.

Facial recognition software isn't magic. It just says whether two people look alike and you already know that.
 
Edmund

He was widely known as Kirby Smith, presumably to distinguish from every Tom, Dick and Harry Smith. Commanded the Trans-Mississippi Department in the latter part of the war.

KirbySmithCloseup.jpg
 
Last edited:
Back
Top