Companies apply the same technology that Facebook uses on photo and video files to find and identify our ancestors
The ability to enjoy the Facebook To find our faces and recognize friends in embarrassing photos taken ten years ago is familiar to most of us. Typically, the technology works by mapping the geometry of a face; the relative positions and distances between the eyes, nose, forehead, mouth and chin. Up to 70 'facial landmarks' can be used to put a face to your 'facial signature' and distinguish you from others.
One of the great things is that this signature can be used to find other faces in a database with very similar signatures and to identify your face in long-forgotten images or footage. In recent years, thanks in part to facial recognition capabilities more readily offered by major cloud providers, the same technique is being applied to identify people in photos not only from our college days but from as far back as the 1860s. XNUMX.
Photo adjustment
Last year, developer Vignesh Sankaran built a tool that recognized faces from the State Library of New South Wales' collection of digitized images. The online application used Amazon Web Services' facial detection and recognition capabilities to select faces in photographs from the library's Sam Hood collection.
Hood worked as a photographer and photojournalist predominantly in the Sydney, Australia area from 1880 to 1950. A collection of over 30.000 of Hood's negatives was acquired by the library in the 1970s.
“Clicking an image shows the facial detection results with bound boxes around the detected faces. Bounding boxes colored in dark blue are faces that had similar faces detected in the image collection, with a confidence level of 95%”, described Sankaran.
Clicking a blue box brings up other photos in which that face appears.
Potentially, the app could be developed to attach names to any recognized faces, making searching for individuals in the collection much easier for library staff.
Similar work is underway on a larger scale in the US to match faces found in archived and submitted American Civil War photos.
In 2017, collaboration between researchers at Virginia Tech, the Virginia Center for Civil War Studies, and Military Images magazine resulted in the development of CivilWarPhotoSleuth (CWPS).
The tool uses facial recognition software to identify 27 “facial landmarks” in period photographs uploaded by the public. CWPS then compares the unique facial landmarks to the tens of thousands of photos in its archive.
“Facial recognition allows us to find matches even when a soldier's facial hair changes, or if a different view of him is in our archive,” the tool's makers said.
“One of the site's greatest strengths is that the more people use it, the more valuable it becomes. When you add an ID photo from your collection, it can instantly match a mysterious photo that another user has been trying to identify for years. Likewise, if you search for an unidentified photo and don't find a match at first, you will automatically be notified if a possible matching photo appears on the site at any time in the future,” they added.
A public version of website was released in August.
File value
The capability also has enterprise applications – particularly for media organizations that want to find relevant footage or photos in their video archives.
“They have millions of hours of video content and are typically stored across multiple legacy systems, there is no meta tagging and the search processes for finding content are extremely old and manual and cut across multiple systems,” explains Angus Dorney, co-CEO. from Sydney and Melbourne cloud technology company Kablamo.
“If you are a newsmaker at a media organization or work for a government archive and someone asks you for a specific piece of footage it is very difficult and time consuming and expensive to try to find,” he adds.
Kablamo creates solutions that have a “YouTube-like user experience” for finding relevant archival material. Using AWS face and object recognition tools, users simply type in a person or thing “and can get a back list of prioritized rankings, where it is, and be able to click and access 'that example' immediately,” says Dorney, former general manager Rackspace.
The machine learning models behind the capability, over time, can refine and adjust its behavior, making the results more accurate and more useful to users.
“You actually have a computer starting to function like a human brain around these things, which is incredibly exciting,” adds Dorney.
Similar work is being carried out by Danish company Vintage Cloud. It uses a visual recognition API offered by Clarifai to apply meta tagging to old film material in a product called Smart Indexing. The company recently announced a database of 100.000 faces that customers can access and compare with those found in archival images.
“Imagine if a producer came to you, demanding footage of Marlon Brando, or a burning skyscraper, or a 1976 Ford,” said Peter Englesson, CEO of Vintage Cloud. “Smart indexing of your archival assets would allow you to not only establish if you have the clip you want, but also access it immediately – providing the opportunity to realize the value of that asset.”
By George Nott / Computerworld AU
FOLLOW-US
