The Facial Identification Subcommittee focuses on standards and guidelines related to the image-based comparisons of human facial features.
Lora Sims, Subcommittee Chair, Ideal Innovations, Inc.
Britt Toalson, Subcommittee Vice Chair, Seattle (Washington) Police Department
Jane Wankmiller, Subcommittee Executive Secretary, Northern Michigan University
Walter E. Bruehs, U.S. Federal Bureau of Investigation
Terry Cowman, Iowa Department of Public Safety
Mark Dolfi, Los Angeles County Sheriff's Department
Neal Gieselman, Aware, Inc.
Steven B. Lee, San Jose State University
Ping Ma, University of Georgia
Allison Miller, Biometrics Operations Division (BOD)
Emily Mullins, USG
Trish Murphy, U.S. Department of Defense
Todd Putorti, New York State Department of Motor Vehicles, Division of Field Investigation
Kirt Simmons, D.D.S./Ph.D., Arkansas Children's Hospital
Debra Tennant, Federal Bureau of Investigation, Criminal Justice Information Services
Antonio Trindade, U.S. Customs and Border Protection, U.S. Border Patrol
Steven Wilkins, Pierce County (Washington) Sheriff's Department
See the Research & Development Needs identified by the Facial Identification subcommittee.
The Forensic Science Standards Board (FSSB) has provided the opportunity for OSAC Subcommittees to identify baseline documents and reference materials that best reflect the current state of the practice within their respective disciplines.
These documents contain practical information regarding these disciplines that can help forensic scientists, judges, lawyers, researchers, other interested parties and the general public, to better understand the nature, scope, and foundations of the individual disciplines as they are currently practiced.
It is important to note that the identification of these documents in this venue does not represent an endorsement by OSAC or NIST. Only documents that are posted on the OSAC Registry constitute OSAC endorsement. All copyrights for these documents are reserved by their owners.
Frequently Asked Questions
1. What is automated facial recognition?
Automated facial recognition is the process of submitting an image containing an unknown face (probe) into a computer software program employing mathematical equations to compare the submitted image against a biometric database containing images of known faces. The process can be extended to display the return of known facial images (“candidate list”) which are in rank order according to a computer-evaluated measure of similarities.
2. What is facial image comparison?
Facial image comparison describes the manual process to identify similarities between (a.) two (or more) facial images or (b.) facial image(s) and a subject for the purpose of determining if they represent the same person or a different person.
3. Are all facial image comparisons the same?
No. There are several types of facial image comparison: Facial Assessment, Facial Review, and Facial Examination. Facial Assessment generally involves a superficial, often rapid comparison process; Facial Review involves more complex comparison processes; and Facial Examination involves an intensive, systematic comparison process.
- Facial Assessment is the one-to-one comparison of facial images or the comparison of a facial image to a live subject for the purpose of identity screening in a high throughput environment (e.g., border crossing) or in a law enforcement setting (e.g., police traffic stop).
- Facial Review refers to the human review of the candidate list returned by a Facial Recognition system. This process involves a series of one-to-one image comparisons to determine if the subject in the probe may also be depicted in any of the candidate images. The comparison process is similar to the process used for Facial Examination, but is less rigorous.
- Facial Examination is the formal, systematic examination process (e.g., Analysis, Comparison, Evaluation - Verification [ACE-V]) by which two or more faceimages are compared to determine whether or not they depict the same person. This is an intensive process that can reach the level of a forensic analysis.
4. How reliable is facial comparison?
With recent advancements in facial recognition technology, the reliability of facial comparison is very accurate. A recent study, Perceptual Expertise in Forensic Facial Image, by D. White, P.J. Phillips, C.A. Hahn, M. Hill and A.J. O’Toole (2015) states that, “Examiners outperformed untrained participants and computer algorithms, thereby providing the first evidence that these examiners are experts at this task.” Scientific research and testing is ongoing to assess and improve the reliability of facial comparisons.
5. With the advancement in readily available facial recognition technology, why do we need human examiners?
Automated facial recognition depends on the mathematically determined similarity between facial images and is often most effective in controlled situations (frontal-facing pose, even lighting, neutral expression), which is rarely the situation encountered with forensic material. The software might not recognize details that humans can perceive (e.g., the effects of changes in expression, weight, or age), leading to images of the same person scoring weakly, highly similar facial images of different people scoring strongly. A human is therefore necessary to make the comparison of the images and come to a conclusion of the comparison.
6. Why are some image comparisons determined by experts to be inconclusive?
Comparisons conducted by experts are based on the quality and quantity of details in the image(s) provided. In many cases there is not enough detailed visual information for the expert to make a definitive conclusion.
7. How long does a facial image comparison take?
Depending on the quality of the image(s), the purpose of the comparison, and experience of the examiner, the length of time required ranges from minutes to hours.
8. What is the recommended method for forensic facial examinations?
The Organization of Scientific Area Committees Facial Identification Subcommittee (OSAC-FI) and the Facial Identification Scientific Working Group (FISWG) recommend Morphological Analysis as the primary method of image comparison. Morphological Analysis is an intensive, systematic process of image comparison in which the features of the face are described and compared, and conclusions regarding similarity or difference are based on subjective observations and evaluations. For additional information regarding facial comparison methods, see the FISWG website at www.fiswg.org.
9. Who is conducting facial image comparisons?
Facial image comparisons are generally performed by individuals working within law enforcement, government, national defense and academic institutions for research purposes.
10. Is there a certification for facial image comparison?
At this time, no public certification programs for facial image comparison exist. However, the International Association for Identification (IAI) is currently working on guidelines to further develop certification programs which are expected to be available to practitioners within the next few years.
11. What training programs are available?
Many agencies/institutions have in-house training programs which are suited to their specific facial image comparison needs. There are training programs available to the general public, but they are limited.
12. I am an artist who draws/sculpts faces for a living. Am I qualified to conduct facial image comparisons?
Experience as an artist can be helpful in conducting facial image comparisons, however, the process requires training, evaluation, and awareness of the potential consequences of the comparison.
In general, the development of standards and guidelines is transitioning from the Scientific Working Groups (SWGs) to the OSAC. Some SWGs will continue to operate to provide other resources within their discipline. The existing SWG documents will remain in effect until updated documents are disseminated by the OSAC or the SWG. SWGDAM will retain the responsibility for updating the FBI DNA Quality Assurance Standards.
– Forensic Science Standards Board: March 2015