1
edit
Changes
→Recorded Details
= Recorded Details =
'''1. Introduction & Definition of Face Recognition Technology'''
http://en.wikipedia.org/wiki/Facial_recognition_system#Early_development
'''2. How does the technology work (overview)?'''
Effectiveness of Point Distribution Model and Active Shape Models:
“The point distribution model can be used for Face detection and face recognition.”
ROKITA, JOANNA, ADAM KRZYŻAK, and CHING Y. SUEN. "MULTIMODAL BIOMETRICS BY FACE
AND HAND IMAGES TAKEN BY A CELL PHONE CAMERA." International Journal of Pattern
Recognition & Artificial Intelligence 22.3 (2008): 411-429. Academic Search Premier. EBSCO. Web. 1
Nov. 2010.
'''Recorded Details:'''
To find local facial features, such as eyes, mouth and nose, we can apply a point distribution model
and active shape models. Point distribution model is used to generate facial points. Each shape in the
face is represented by a set of n labeled landmark points, which must be consistent from one shape to
the next. As an example, the first point may correspond to the right corner of the right eye. Given a set
of such labeled training examples, we align them into a common coordinate frame. This is
accomplished by translating, rotating and scaling each training shape so as to minimize the sum of
squared distances to the mean of the set. Each shape can then be represented by a 2n-dimensional
vector. This points are followed by some healthy mathematical calculations (will be explained in the
draft).
'''3. Uses of the technology:
3.1 Facial recognition for Authentication and Authorization:'''
“Face detection methods can be categorized into the following two large groups:
1) heuristic-based methods and
2) classification-based methods.”
“The following two popular ways of doing the face registration can be found in the literature:
1) holistic methods and
2) local methods.”
“Illumination normalization, therefore, is a very important pre-processing step before verification.”
Information Fusion:
“From each frame, we obtain a value of its likelihood ratio and compare it to a threshold to make the
decision”
'''Recorded Details:'''
Personal authentication requires a different approach to the face detection and face recognition. For
face detection, we can use either heuristic-based methods or classification-based methods. For face
registration we can use either holistic methods or local methods. All the methods are followed by
heavy math (may be included in the draft). However, all these method can used for authorization or
authentication. Instead of remembering the username and password, one can use face detection and
recognition to authenticate the person to computer login, application login , and websites login. The
idea of using face recognition for authentication also supports global movement of one username and
password for all the online websites (global username and password).
'''3.2 Facial recognition for fighting terrorism:'''
“A technique such as face recognition could, at least in principle, be used to recognize people
“passively,” without their knowledge or cooperation.”
“Instead of a person continuously monitoring the video, the goal is to have a computer monitor the
video and alert a human operator if an “interesting person” is in view.”
System Dependencies:
“the system can only recognize the persons whose images have been enrolled in the gallery”
“If a terrorist is known by name and reputation, but no picture of the terrorist is available, then the face
recognition system is useless.”
“system must be able to acquire face images of reasonable quality to use as probes.”
“system has a sensitivity threshold that must be set appropriately.”
Bowyer, Kevin W. "Face Recognition Technology: Security versus Privacy." IEEE Technology &
Society Magazine 23.1 (2004): 9-20. Academic Search Premier. EBSCO. Web. 1 Nov. 2010.
'''Recorded Detail:'''
Recognizing a terrorist using a FRT might be useful having important assets monitored by cameras
creates a system that is always looking for persons of interest. However, there are ways for the
terrorists to avoid such a system due to it‟s dependencies. FRT can only recognize persons whose
images have been included in a gallery, terrorist could possibly have no picture available and when
images are acquired they need to be of a reasonable quality. FRT threshold must be set appropriately.
- the passive nature of this technology increases the number of faces that are verified
- a possible terror suspect can be apprehended in a discrete manner without making a scene.
- passing off the monitoring to a machine creates impartiality? since a machine is doing the verifying.
(not sure is this is true or not)
'''3.3 Facial recognition for search realistically.'''
“Imagine never forgetting a face, and always being able to recognize a friend-of-a-friend on the street.
And because the systems arelinked to the Internet, once AR recognized someone, you could easily
pull up his or her online footprint, such as a Facebook page.”
Cascio, Jamais. "Seeing Too Much." Atlantic Monthly (10727825) 304.4 (2009): 34. Academic Search
Premier. EBSCO. Web. 1 Nov. 2010.
Recorded Details:
Face recognition can be used to search faces online. This technology has made possible to search a
person using his image online. Just upload the image online; Hit the search; Surprisingly enough, all
the publicly available information is listed. Although it is very impressive to search in such a way, there
are some security concerns. Next part will be discussing about security concerns.
'''4. Problems due to technology.
4.1 Facial recognition vs. privacy in social media driven world:'''
1. “The United States Constitution declares a level of protection for the rights of individual citizens
against oppression by their government that has made the U.S. unique. The right to privacy is
one right that has become a firmly entrenched American value, even though it is not explicitly
enumerated in the Constitution. Since face recognition systems represent leading-edge and
rapidly changing technology, this technology is currently a major issue in the area of social im-
pact of technology.”
Bowyer, Kevin W. "Face Recognition Technology: Security versus Privacy." IEEE Technology
& Society Magazine 23.1 (2004): 9-20. Academic Search Premier. EBSCO. Web. 1 Nov. 2010.
2. “We don‟t regard ourselves as „Big Brother‟. We‟re more like a friendly uncle or aunt watching
over you.” Hale, Benjamin.
"Identity crisis: Face recognition technology and freedom of the will."
Ethics, Place & Environment 8.2 (2005): 141-158. Academic Search Premier. EBSCO. Web. 1
Nov. 2010.
'''Recorded Detail:'''
Social networks have become more ubiquitous in day to day business. A persons online
persona is created, sometimes without the persons consent. Having the ability to query using
Facial Recognition Technology creates a situation where the party doing the surveillance can
do this in a passive manner. The time taken to investigate an individual and do an assessment can
be done in a quickly. Having a FRT system suggests to a society that one has to conform to norms
and obey the law.
''Face recognition technology Notes:''
(We may use one of the quotes from the listed below)
''The Privacy Argument''
"The difference between what a computer might do to make the connection between
two publicly available data sets (like a publicly captured faceprint and a mugshot) and
what a police officer, an insurance representative, or a private agent might do is only a
matter of speed."
"detractors of FRT might do better to argue that it is a restriction of freedom, and not
privacy alone"
''The Objective Freedom Argument''
“the capacity to delve into the moment- to-moment lives of citizens, the potential for
abuse is tremendous”
“arbitrarily passing and enforcing laws if citizens who might otherwise object to those
laws can be monitored to the degree that any dissent might be quashed before it be-
comes dissent worth worrying about?”
“it holds the potential to make all laws fully enforceable, regardless of whether the laws
themselves are just”
''The Subjective Freedom Argument''
“One need no longer will maxims to be universal laws, because the question of univer-
sal law has been supplanted by the positive authority of institutional law. The categori-
cal imperative, in a nearly just society of perfect enforcement, will read instead: „One
ought to act according to that maxim that is the positively enforced state law‟.”
“The problem here is that the operationalization of ethical deliberation transforms the
meaning of freedom altogether, shifting it from the free establishment of self-imposed
rules about how to act (from autonomy) to a calculable choice between a list of exter-
nally provided ends (to heteronomy).”
''Recognition and Freedom''
“ In the world of FRT, the institutional identity provides a surrogate practical identity,
supplanting (1) the reflectively reasoned and willed adoption of normative stances
about how to act with (2) a characterization of an individual‟s institutionally recognized
past actions.”
“One is either a lawbreaker or a law-abider, and when one is a lawbreaker, one is a
lawbreaker according to what one has already done.”
(past actions determine how people society treats you)
“At the objective level, freedom is eclipsed because people are no longer able to break
the law and to succeed at breaking the law.”
“With highly and heavily administered FRT, the suggestion that one „ought to obey the
law‟ is transformed into the more binding mandate that „one cannot avoid obeying the
law‟.”
“the idea of freedom necessarily implies that one maintain the possibility of deliberating
over courses of action, of acting according to practical reason.”
“The threat to individual autonomy is therefore twofold: first, regarding possibilities for
setting individual laws; and, second, regarding possibilities for the social recognition of
self-determination and public establishment of autonomous agency. FRT provides the
means for men to undo the progress of modernity. It makes real the previously unimag-
inable prospect that free will, that darling of the Enlightenment, will henceforth be dog-
ged by cameras.”
'''4.2 Facial recognition false positives.'''
“Distinctiveness of a face predicts both miss errors (MEs) and false positives (FPs) but correlations
between these errors are low”
Lewis, Michael B. "Familiarity, Target Set and False Positives in Face Recognition." European Journal
of Cognitive Psychology 9.4 (1997): 437-459. Academic Search Premier. EBSCO. Web. 1 Nov. 2010.
''Recorded Detail:''
People recognize faces that have personal familiarity this determines if false positives will
occur. Face Recognition Technology in some instances produce false positives when subjects
are wearing eyeglasses, lighting glare. For FRT to work a frontal view of the face is needed.