By Owen McShane, Director, Division of Field Investigation, New York State Department of Motor Vehicles; and Anne Dowling, Deputy Director, Institute for Traffic Safety Management and Research, State University of New York, University at Albany, New York
he key to a state having a secure driver license and identification card issuance system is a comprehensive approach that focuses on prevention and deterrence against fraud. From the initial application process to the final document issuance, it is important that the system focuses on ensuring that the applicants are “who they say they are.” Over the past 10 years, the New York State Department of Motor Vehicles (NYSDMV) has implemented a variety of new programs and initiatives to help ensure this process, such as electronically verifying the social security numbers for all license applications and installing new verification machines that help verify breeder documents that are used to establish a customer’s identity.
During the past three years the NYSDMV process has taken an important leap forward with the implementation of cuttingedge facial recognition technology. The new facial recognition program is intended to advance the NYSDMV’s important goal of “one driver, one license” to deter identity fraud and improve highway safety.
Under a $2.5 million grant from the U.S. Department of Homeland Security, NYSDMV’s Division of Field Investigation (DFI) initiated an effort in 2010 to incorporate facial recognition technology into its business practices for the primary purpose of preventing and deterring the issuance of multiple licenses to a single individual. This important undertaking involves searching New York’s driver license file of approximately 21 million records (all with facial images) for duplicate records and for drivers with multiple licenses or identities. To accomplish this, NYSDMV is using facial recognition technology in a dual manner: (1) to compare existing images on the driver license file for possible matches, and (2) to compare a new image from a prospective license applicant to the existing file of facial images before issuing a license document. This ensures that no more than one document is issued to an applicant. These two uses of facial recognition technology are important to NYSDMV in stopping identity theft and driver license fraud.
Origin of Multiple License Records
The problem of multiple records on the NYSDMV driver license file for a single individual has many origins. Some multiple records are the result of simple data entry mistakes by the reporting agencies involved (for example, law enforcement agencies and courts) and by the NYSDMV data entry staff itself. Some are the result of transactions involving name changes stemming from marriage and divorce. However, investigations have revealed that many multiple records on the DMV database are the result of a deliberate effort to circumvent various laws and possible sanctions or penalties for violation of those laws, including New York’s Vehicle and Traffic Law, tax law, social service laws, and penal laws.
With a goal of having one record for each driver (that is eliminating the creation of multiple licenses per individual), NYSDMV has identified three distinct groups who attempt to establish a second identity or maintain multiple identities:
Group 1: Individuals engaged in identity theft or other criminal acts.
Group 2: Individuals wanted under their true identity who establish a second name as an alias. This group poses a high risk for law enforcement encounters because officers may believe they are dealing with a first time traffic offender when they are actually dealing with a wanted felon.
Group 3: Individuals who have multiple traffic violations (for example, traffic infractions, DWIs, insurance suspensions) who establish an alias in order to (1) continue to appear to drive legally while suspended under their true identities, or (2) avoid suspensions by dividing tickets and infractions among multiple records.
Each of these three groups represents a serious problem, albeit from different perspectives. The first two groups involve individuals who pose a serious risk to the general public’s safety, which requires action by the state’s law enforcement community. In comparison, the third group poses a serious risk to those using New York’s roadways, which requires the attention of the state’s traffic safety community. Since research has consistently shown that driver behavior is a key contributing factor in the large majority of traffic crashes, the issue of a single person having multiple licenses or records has serious traffic safety implications.
Since NYSDMV is concerned with both the public safety and traffic safety issues related to multiple licenses, in January 2010, the Institute for Traffic Safety Management and Research, a not-for-profit
research center affiliated with the University at Albany, was asked to assist the DFI in examining and analyzing the records of persons across the state who have or are trying to obtain more than one driver license or non-driver identification (ID) card.
Using Facial Recognition Technology
A two-pronged approach is being used in the statewide roll out of the facial recognition program. One approach takes all of the photos that are captured daily by NYSDMV’s offices across the state and conducts a one-to-one comparison against older photos of the subject, followed by a one-to-many search against all the other photos in the NYSDMV’s database. Approximately 7,000 customer photos are taken daily and compared to the more than 21 million driver and nondriver ID photos in the database. The second approach, also conducted on a daily basis, takes a percentage of the “legacy” photos and conducts a one-to-many search of the other photos. Due to the size of the legacy file, this second approach took three years to complete.
When the investigation of the identified “matched records” is complete, the records are merged, as warranted. The merged record is then reviewed to determine whether the individual’s license should be suspended or revoked based on the combined number of tickets, points, open suspensions, crashes, and so forth that are on the merged record.
The results of the matching process for the first two years of the program, February 3, 2010, to February 2, 2013, from both a public safety perspective and a traffic safety perspective, are summarized below.
Public Safety Implications
The primary objectives of implementing the use of facial recognition technology are to identify individuals
- engaged in identity theft or other criminal acts, and
- individuals wanted under their true identity who establish a second name as an alias.
These issues are significant, especially identity theft, as it is a serious crime that affects 11 million people nationally in any given year, which suggests that 1 in 20 U.S. citizens are at risk of becoming a victim.
Over the three-year period, more than 21 million photos were compared using the facial recognition technology. Overall, this resulted in over 13,500 cases being generated for subjects that had two or more records with NYSDMV. The system also identified 7,710 potential matches, which were determined to be twins or multiple births.
While investigations of the more than 13,000 possible fraud cases are ongoing, key findings of the first three years of operation include the following:
- More than 2,500 felony arrests were made, including
- A subject holding four New York licenses under separate names, who was also naturalized as a U.S. citizen under the same names and was issued four different valid Social Security numbers. The subject was found to be on the no-fly list under the name he used when he initially settled in the United States.
- Over 250 Commercial Driver License (CDL) drivers were found to have two or more licenses. Many of these drivers had open suspensions, DWI charges, and convictions for other offenses that would have prevented them from obtaining a CDL license under their “true” names.
- A subject who had multiple active warrants under his true name since 1993 for bank robbery who was arrested after the photo from his new identity matched his old license record, which was still suspended for unpaid moving violations.
- In addition, administrative action was initiated on more than 7,000 individuals who were identified as having multiple records where the license transactions were too old for criminal prosecution. Generally, the subject’s license would be revoked for a one-year period and all records for the subject would be merged. In this way, subjects would be held accountable for all tickets or accidents they had accumulated under the different identities they had used.
Traffic Safety Implications
Although the use of facial recognition technology is gaining popularity among the state’s driver licensing agencies to prevent and deter driver license fraud, the use of such technology to identify traffic safety-related problems is limited. The primary issues related to individuals with multiple license records is the extent to which they are obtaining multiple licenses to hide the fact that they have multiple traffic violations or multiple crashes on their driving records. Obtaining and using multiple license records in this manner enables a driver to avoid the appropriate sanctions and penalties associated with such events. As a result, problem drivers remain on the state’s roadways, putting other highway users at risk.
To study the traffic safety implications of multiple records, 6,111 cases of “matched records” with possible fraud problems, from the first two years of the program, were reviewed and analyzed to identify how individuals are using their multiple records and determine whether their use has traffic safety implications. Key findings from the review to date show the following:
- 21 percent (1 in 5 subjects) did not have a valid license.
- 3 percent had been involved in a crash, compared to 42 percent of all New York state (NYS) licensed drivers.
- 9 percent had been convicted of impaired driving, compared to 2 percent of all NYS licensed drivers.
- 29 percent had been convicted of a cellphone violation compared to 9 percent of all NYS licensed drivers.
- 56 percent had been convicted of a seat belt violation compared to 21 percent of all NYS Licensed Drivers.
- 35 percent had accumulated six or more points on their license record within an 18-month period at some point in time after November 18, 2004, compared to 11 percent of all NYS licensed drivers.
The NYSDMV is very excited about its successful use of facial recognition technology to uncover identity fraud and keep its highways safer. The total success of this program relies on the ongoing cooperation among the state’s traffic safety organizations, law enforcement personnel, and prosecutors.
The results of the facial recognition program are being shared with other federal and state agencies in an effort to combat fraud, especially among agencies that provide benefits. It is expected that the results of NYSDMV’s facial recognition program will help these agencies uncover related fraud specific to their missions, such as the double collection of medical benefits, fraudulent tax refund applications, and improper disability claims. In the first three years of the program, the NYSDMV DFI identified subjects collecting benefits under multiple identities, as well as subjects working full time under one identity while collecting full disability under a second identity.
The knowledge gained to date from the program, together with the size of the New York driver’s license file (approximately 21.5 million records with facial images), provides an unprecedented opportunity to explore the feasibility of using facial recognition technology to identify and address both public safety and traffic safety–related issues. ♦
Please cite as:
Owen McShane and Anne Dowling, "A Practical Application: Facial Recognition Technology," The Police Chief 80 (June 2013): 42–44.