Rothco set out to find the truth behind a legendary escape story from Alcatraz—and claims to have solved it by unlocking the hidden data.
As a bleeding-edge type of news, Accenture Interactive-owned Irish agency, Rothco teamed up with AI specialist company Identv to confirm the identities of two Alcatraz prison escapees depicted in a 1975 photo. Frank Morris, along with brothers John and Clarence Anglin, were made famous by Clint Eastwood in the 1979 film Escape from Alcatraz, but were assumed to have drowned after fleeing the famous California prison. However in 2015, evidence surfaced of a grainy and blurry old photo taken by some family friend of the Anglin Brothers allegedly living in Brazil in 1975—13 years after their supposed drowning.
Technology wasn’t available to verify if the photo really was of the Anglins until now, but AI has been used to unlock the hidden data in the printed photo recently and caused to reveal this detective project. Identv used a facial matching system based on a type of machine learning algorithm called a deep neural network.
Mark Hughes, Chief AI Scientist of Identv commented in a statement,
We start with what is called a training process, where we feed the algorithm many images of a single person’s face along with many images of different people. We repeat this process millions of times over and use mathematical models optimized during the training phase to learn how to differentiate one person’s face from another. Identv has developed techniques to carry out this matching process in tens of milliseconds over millions of faces.
Alan Kelly, CCO at Rothco commented,
The Long Shot is a novel example of successfully applying the creative knowledge and digital tools of today to a hidden piece of the past, unlocking newfound stories and opportunities. Technology is advancing at such a rate it can leave you a little dizzy and, short of obtaining DNA proof, there will always be a little room for mystery. But as far as technology is concerned—the prisoners made it.
Hughes also added that once they have that “trained” model, it allows the team to create a sort of mathematical, “facial fingerprint” that is extremely discriminative—it can be compared to other such “fingerprints” in the database in the way that modern-day police fingerprint matching systems work.