Combating Fraud with Face Recognition Some FBI Initiatives

Combating Fraud with Face Recognition: Some FBI Initiatives Richard W. Vorder Bruegge Senior Photographic Technologist Federal Bureau of Investigation July 2011

Outline • Context: NGI, BCOE & FBI FR Use Cases • Examples of R&D and recent FBI FR use • Facial Identification Scientific Working Group (FISWG) • FBI Outreach to DMVs 2

Next Generation Identification (NGI) 4 An upgrade to IAFIS – The FBI is developing an automated, interoperable multimodal biometric system Enhanced Palmprints tenprint and latents services Photos/Facial, 2012 2011 scars, marks, and tattoos 2013 Iris 2014 4 The NGI system: – Incremental replacement of the current IAFIS – Improving current functionality and providing new functionality – Developing multimodal collection and search identification services 3

Beyond System Development The BCOE is the FBI’s program for exploring and advancing the use of new and enhanced biometric technologies and capabilities for integration into operations

Proposed FR Use Cases 4 Identifying , subjects, fugitives, missing persons, and persons of interest by using FR systems – From authorized LE & Govt. Sources (DMVs? ) – From open sources – From images in seized systems 4 Tracking subjects movements to/from critical events (e. g. , 9/11) 4 Conducting automated surveillance at lookout locations 4 Verifying mug shots against National Criminal Information Center (NCIC) records 4 Controlling access 5

Facial & Image-Based Recognition 4 Facial Identification (FI) - the manual examination of the differences and similarities between two facial images to determine if they represent different or the same person 4 Facial Recognition (FR) - the automated searching of a facial image in a computer database, resulting in a group of facial images ranked by computer-evaluated similarity 4 Image-Based Recognition – the automated searching of features, such as ears, hands, blemishes, and scars, marks and tattoos, in a computer database 4 Current FBI face and image biometric services include: – Operational Technology Division (OTD) facial identifications and court testimony at the Forensic Audio, Video, and Image Analysis Unit (FAVIAU) – Criminal Justice Information Services (CJIS) Division FACE Search Team 6

FBI Identification from Images Primarily Supports Internal FBI Casework 4 The FBI’s Facial Recognition/Identification work is performed within the Forensic Audio, Video, and Image Analysis Unit (FAVIAU) 4 The FAVIAU is one of only a few accredited laboratories in the world that conducts examinations in the disciplines of video, image, and audio analysis 7

FBI-DMV Face Pilot 4 Departments of Motor Vehicles (DMVs) have deployed FR systems to reduce fraud since 1997 4 The BCOE-sponsored pilot program with the North Carolina DMV has led to more than 700 leads including: – 1 federal fugitive apprehension – 6 state fugitive apprehensions – 1 missing person resolution 4 For the pilot, CJIS provides probe photos and names to run in the NC FR system against DMV photos 4 The BCOE is working to expand the FR pilot to additional DMVs Probe Photo 8

How is FACEMASK-S&T being used operationally today? Face Search, then name search 4 Fugitives include subjects with wants and warrants in NCIC records in North Carolina and surrounding states (Virginia, Tennessee, Georgia, South Carolina) Subject demographics not brought up unless there is a probable match based on face or name search. FR search yields image, DL # and score. “Matches” used for lead development, only – not conclusive. FBI does not see, review, or retain any PII data on “non-match” (“innocent citizens”)

FACEMASK – S&T Statistics to date Total numbers of searches: ~3000 Hits on face search: 380 Hits on name search: 320* *Cautionary Statistic: Significant number of FR MISSES – Provides insight into how many times one might miss a subject.

Automatic Face Detection and Recognition (AFDAR) in Videos and Still Images 4 USG-owned software application 4 Expedite analyst review of videos and still images 4 Potential for use in terrorism and child pornography cases – Automatically locate faces and compare against others (either internal to video/image set or external). 4 Review/analysis of unmanned/unmonitored surveillance videos. 11

Cluster Base - Origins 4 Originally called “AFDAR-C” 4 Analyst tool designed to expedite review of still images. 4 Applies AFDAR face detection and clustering to still images. – Includes both “clusters” and “close clusters” 4 Allows analyst to identify & verify links between images and image clusters. 13

Cluster Base – Applied to Recent Investigation 4 Criminal Enterprise obtained multiple names and SSNs from temporary workers. 4 Multiple IDs would be sold to a single person. 4 These people would use these IDs to get mortgages, credit cards, etc. 4 Investigators identified several states in which fraudulent IDs were being used. 14

Cluster Base – Applied to Recent Investigation 4 One state identified as known location of activity – FBI requested DMV photos of subjects with specific SSNs and names (known fraud) – 14, 000+ provided 4 ~100 Known Suspects from 2 nd state included with 14 K+, then Cluster Base run… – 36 “Strong Hits”, including leader of criminal enterprise – Total 70 images matching to 36 subjects found in this set 4 Cluster Base still being used to find multiple offenders 15

Standards - FISWG 4 Facial Identification Scientific Working Group (FISWG) 4 Sponsored by the BCOE and established in 2009 4 Mission - “develop consensus standards, guidelines, and best practices for the discipline of image-based comparisons of human features” 4 Participants include federal, state, local, and international government agencies, as well as invited non-governmental organizations 4 Next meeting November 14 -18, 2011 in Orlando, FL 4 Visit www. fiswg. org 17

FISWG Organization – 5 Subcommittees 4 FR Systems 4 Capture and Equipment 4 One-to-One Comparisons 4 Training 4 Outreach 4 Documents intended to establish standards and consolidate information. 18

Developing the FBI Training Curriculum 19

FBI Outreach To DMVs 4 FBI has established a Facial Searching Service Team to aid FBI investigations. 4 Hope is to have this team facilitate authorized searches of DMV FR databases to locate wanted, warranted, and missing persons. 4 Letters requesting information on DMV FR systems have been mailed to 46 states to date. 4 13 responses so far (but some only mailed in last two weeks). – Thanks to those who have answered… 20

FBI Outreach To DMVs – Letter Sent 4 Does your state DMV have a FR system or plan to have a FR system in the future? 4 Does your state permit the FBI or Law Enforcement agencies to search your FR database remotely or in person? 4 Will your agency search photos for the FBI? 4 Has a Memorandum of Understanding or any other agreement been signed between your state or agency and the FBI regarding searching your DMV FR system? 4 Who is your primary point of contact for the use of your state’s FR system? Please provide their contact information. 4 From a legal perspective, are there any privacy protection provisions in the controlling legal authority for your state, including legislative code, administrative regulation or executive order, if so, what are they? 21

Questions / Comments Contact Information: Richard Vorder Bruegge Email: Richard. Vorder. Bruegge@ic. fbi. gov Additional Resources: www. Biometric. Co. E. gov Email: Biometric. Co. E@leo. gov 22

23