+- Deep Politics Forum (https://deeppoliticsforum.com/fora)
+-- Forum: Deep Politics Forum (https://deeppoliticsforum.com/fora/forum-1.html)
+--- Forum: JFK Assassination (https://deeppoliticsforum.com/fora/forum-3.html)
+--- Thread: Pixel Counting Biometric Comparison of Oswald photos (/thread-13934.html)
Pages:
1
2
Pixel Counting Biometric Comparison of Oswald photos - Drew Phipps - 04-06-2015
I don't do Facebook or other social media, so I really can't tell you about their TOS. However, I might suggest that, after all the claims they've made about it being "better than the human brain" and all, a little informal testing of their claim would be fair use. I would imagine that if you merely posted your results on a Facebook page, you know, to demonstrate Facespace's ability to distinguish between faces... that would be like a little free advertising for them.
However, I wouldn't suggest trying to use their system to, say, sell a book, or otherwise turn a profit.
Pixel Counting Biometric Comparison of Oswald photos - R.K. Locke - 04-06-2015
They all look like the same person to me.
Pixel Counting Biometric Comparison of Oswald photos - Drew Phipps - 04-06-2015
R.K. Locke Wrote:They all look like the same person to me.
Yeah, me too...
Pixel Counting Biometric Comparison of Oswald photos - Jim Hargrove - 06-06-2015
Drew,
Do you have a copy of Jack White's poster called "The Evolution of Lee Harvey Oswald"? It has 80 or so mugshots purportedly of LHO, though a few barely resemble him/them.
The shots are printed halftones, of course, maybe an inch and half by two and a half each, but they are tight on the face and identified. The resolution might be a little better than you'd expect. I made a high resolution photo for David Josephs (not great, but you can read the smallest print on the poster). I'd be happy to send you a copy if you thought it might be helpful.
This is groundbreaking work you've done here, and should be the starting point of even more serious analysis. I'm hoping John A will take a look and give us his impressions, but he's doing a long write-up right now on Landesberg and, as always, is pretty compulsive about it. Maybe soon.
Many of the two individuals you've postulated do match the H&L analysis; some clearly do not. Whether this indicates errors in J.A.'s paradigm, shortcomings in the biometric analysis, or misrepresentations in the public record (deliberate or accidental) is a subject that deserves a whole lot of additional scrutiny.
Pixel Counting Biometric Comparison of Oswald photos - Drew Phipps - 07-06-2015
I welcome any additional photos, and I'll keep my excel spreadsheet handy, so new faces can be added. Please post or send me the images.
Pixel Counting Biometric Comparison of Oswald photos - Jim Hargrove - 07-06-2015
The image below of Jack White's "The Evolution of Lee Harvey Oswald" poster was compressed by Linux GIMP from 7MB to less than 1MB to save bandwidth for Magda. Comparing the original and compressed images side-by-side shows remarkably little degradation. (The biggest loss surely comes from my execrable photography skills.) Anyone who ever met or communicated with Jack White in a substantial way would know that he would approve in a heartbeat my decision to upload this still-copyrighted image. Hope the nearly 1MB upload works....
Pixel Counting Biometric Comparison of Oswald photos - Drew Phipps - 14-06-2015
I'm working with the above pictures, and the extra pixels have allowed a reduced error (down to around 8% as opposed to upwards of 15% in some cases). I don't suppose you could link me, or send me, to the 7 MB picture, which would instill even greater confidence...
I haven't selected all of them to work with, as some of them are too blurry and some of them aren't full frontal... I have selected 14 of the "best" poses.
Pixel Counting Biometric Comparison of Oswald photos - Jim Hargrove - 15-06-2015
Check your private messages. I sent a link.