Key considerations for adopting facial recognition technology
By Don Wick
With facial recognition technology now being used for everything from unlocking your phone to checking in at the airport, it’s inevitable that government would jump onto the facial recognition bandwagon.
Law enforcement, homeland security, security in government buildings and other public agencies are all lobbying to purchase their own facial recognition software. Spurred on by lower prices and rising competition, these agencies are also recognizing that they are now able to deploy facial recognition solutions in new ways, including body cameras, dash cams, and on a mobile phone.
Before making the leap to purchase facial recognition software, though, government agencies would be well-advised to consider a number of key factors.
First, it is essential for agencies to understand how facial recognition works, as well as its limitations. Using image processing and machine learning algorithms, facial recognition software matches the photo of an unidentified person against a database of photos of identified persons. The algorithms then produce a list of possible matches, with each match containing a score which indicates the likelihood of a match. Low resolution, poor lighting, motion blur, glare, off-angle faces, facial hair, glasses, and other scenarios, however, can challenge algorithms to produce a good match.
Next, does the solution provide face detection? In face detection, the algorithm detects faces in an image, without necessarily identifying to whom the face belongs. It may also locate features (eyes, nose, mouth, and ears); detect the presence of facial hair, glasses, and hats; and identify gender, race, approximate age, and emotional state. This type of information is particularly useful in gathering statistics on a large group of people (such as visitors to a public building) or determining the possible intent of individuals in a crowd.
Public perception is yet another key consideration. While facial recognition technology may be perfectly legal when used in appropriate circumstances, lack of information or misinformation can result in public backlash. As a result, it is important for government users to fully understand both the possible uses for the technology, and how the public may perceive those uses.
Searching a database of known offenders for matches of an unidentified suspect in a criminal incident is perhaps the most common use of facial recognition software. As the technology matures, though, other uses may become more widespread. These could include:
Matching an unidentified suspect photo obtained in association with a criminal incident with a state database of driver license photos;
Matching people entering a courthouse with a database of wanted person photos;
Matching airport travelers to a database of known terrorists in real time;
Using a vehicle-mounted camera to match passers-by with a database of wanted person photos;
Using a database of driver licenses and state identification photos to do a real-time search from a vehicle-mounted camera, matching and recording the likely identity, time, and location of passers-by.
Prior experience shows that the public is very understanding when both the photo and database are obtained in direct association with criminal activities. If the photo is a security camera image and the database a set of jail booking photos, for example, even strident civil rights advocates tend to be accepting of the situation. That acceptance evaporates quickly – and the potential for public outcry rises rapidly – if the photo is of a person with no known or suspected criminal activity and the database is a non-criminal database.
Finally, with more government agencies moving to the cloud, it is important to know if facial recognition solutions are available in a cloud deployment. The cloud means no software to install and no servers to manage. But while that significantly cuts costs, cloud deployment raises significant questions about how secure the cloud provider is and whether it is compliant with the FBI’s Criminal Justice Information Services Policy.
Clearly, facial recognition software will play a critically important role for government agencies in the future. It is essential, though for agencies to take proper precautions, both in purchasing and using this solution. To create an atmosphere of trust with the community, it will be essential to inform the public about the intended use of facial recognition software, how it works, what it can (and can’t) do, and how it can benefit public safety.
Don Wick recently retired as chief of the Arvada (Colo.) Police Department. He currently serves as director of operations at Numerica Corporation, where he focuses on Lumen, Numerica’s law enforcement search, analysis, and data sharing platform.