AI Case Study

The US Department of Homeland Security plans to pilot identifying travellers leaving the United States via automobile using a facial recognition system

The US Department of Homeland Security will be implementing a new facial recognition system at land border crossings which will use image and video to identify travellers as they exit the US.

Industry

Public And Social Sector

Security

Project Overview

"Customs and Border Protection will deploy a new system for scanning drivers’ faces as they leave the US. The pilot, called the Vehicle Face System (or VFS), is planned for installation at the Anzalduas border crossing at the southern tip of Texas and scheduled to remain in operation for a full year. According to a Customs spokesperson, the purpose of the project will be 'to evaluate capturing facial biometrics of travelers entering and departing the United States and compare those images to photos on file in government holdings'."

Reported Results

Planned; not yet available

Technology

"The VFS project is based on plenoptic technology, which allows a single sensor to capture images at multiple focal lengths simultaneously." The AI behind the facial recognition system is undisclosed.

Function

Risk

Security

Background

"Traditional facial-recognition systems are often confused by the moving reflections in a car windshield, which make it difficult to isolate facial features. Researchers working on in-car recognition hope that the right focus would separate the driver from the reflections. The new pilot is the result of years of effort by Customs and Border Protection to develop a camera capable of recognizing faces through the windshield of a car, a long-standing challenge for facial recognition systems. Tests related to the new camera have been underway since 2016, initially in collaboration with Oak Ridge National Labs in Tennessee."

Benefits

Data

Facial images captured by camera