Whole-body biometric recognition: coming to a CCTV camera near you?

What is innovation if not a game of one-upsmanship? Having established widespread use cases for fingerprint biometrics and facial matching, and burgeoning markets in iris and palm vein recognition, the horizon is now the total package: whole-body biometric recognition.

The Economist has picked up on the story of the Michigan State University researchers who, since 2021, have been developing technology for long-range biometric identification under its Biometric Recognition and Identification at Altitude and Range (BRIAR) program.

Called FarSight, the tech combines biometric algorithms – called Large Vision Models, or LVMs – that, in the course of about three seconds, measure gait, generate a 3D body model and run the subject’s face through “a turbulence model that seeks to undo the refractive effects of the choppy air on the light that comes into the camera.”

The gait feature is based on Meta’s DINOv2 LVM. Meanwhile, the third feature points to the tool’s intended use: on airborne drones. The Economist says this opens up the potential to “extend the spying powers of artificial intelligence (AI) into the skies.”

“The project is funded by the Intelligence Advanced Research Projects Activity (IARPA), part of the American government responsible for marshalling fanciful spy-gadget ideas into real-world use.” Other current IARPA projects, it notes, include “an effort to build a device that can modulate a voice in real-time in order to avoid detection by speech recognition tools, as well as an initiative to make snooping devices small and pliable enough to be woven directly into clothing.”

Cameras work at high angles, but not too high

By fusing three biometric markers into a single profile, the tech enables those with the data to run matches. And they show promising results; the National Institute of Standards and Technology (NIST) tested FarSight’s full-body recognition on a set of low-resolution images and videos collected at hundreds of metres, often from a high angle; it outperformed all other systems tested on the same set.

That said, a second NIST test on a wider data set revealed limitations, and Dr. Xiaoming Liu, who leads the FarSight project, says its performance “drops considerably if the camera’s angle is very high, if the weather is warm (which causes fiercer turbulence) or if the distance to the target exceeds a kilometer.”

Regardless, a kilometer, or about 0.6 miles, is plenty of room to work with – for drones or other hosts. With BRIAR, IARPA intends to create a technology that can be adaptable to “any high or distant camera, such as those mounted on tall buildings or border-surveillance towers.”

Jay Stanley at the American Civil Liberties Union says “that implies its routine use on civilian populations.”

Many major cities are already wired with thousands of CCTV cameras, some of which are gradually integrating facial recognition capabilities. Recent history has shown that, if law enforcement is presented with a technology it deems helpful to its mission, it will want to use it. The graduation from facial recognition to full-body recognition is not hard to imagine, especially if the latter is pitched as a less invasive option. There are already companies offering the service, and suggesting full body biometric recognition could be deployed in cities where facial recognition is banned.

Article Topics

biometric identification  |  Biometric Recognition and Identification at Altitude and Range (BRIAR)  |  biometrics  |  IARPA  |  long-distance  |  NIST  |  whole-body identification