face recognition

Top PDF face recognition:

Autonomous face recognition

Autonomous face recognition

Another interesting application might be an intelligent system inside cars. A self-learning and adaptive face recognition system can contribute to different functions — such as guard of being stolen, setting of drivers’ preferences, and detection of driver’s attention. The system can automatically learn its owner’s face in different views and keep updating the database. The key of the car is not only used to open the door and engine the motor, but also for keeping the intelligent system in a normal mode. When some one else enters the car without the key, the system is switched into an alarm mode. It can help to send the corresponding warning message to the owner who can decide whether to report to the police. The face shots of the unknown driver are stored in the database. Since GPS and mobile communication systems are more and more popular in cars, the mugshot as well as the car’s current position can be sent by using the in-car mobile phone to the car owner if the alarm is verified by the car owner. In this case, the car is not able to drive faster than a certain limit and it can not restart any more when it stops. If the owner thinks it is a false alarm, he/she can use the mobile phone to switch the system back to the normal mode again. In normal mode, the system learns new faces which might be from the spouse, relatives or friends of the owner. It can then automatically memorize the preferences of different drivers. For some high-end cars, different types of keys are normally used to differentiate the preference settings for different drivers (normally a couple). But when the keys are exchanged, it doesn’t work at all. With the help of the face recognition system, it is not so annoying any more. During driving, the system can also be used to learn and detect the normal status of eyes for each specific driver when he drives. The information might include the blink times per minute, the opening degrees of the eyes etc. When a driver is sleepy, the status changes and a warning voice might occur to notify him. Obviously, it is much more precise to detect the usual status of eyes for a specific person than for a generic person model.
Mehr anzeigen

109 Mehr lesen

Applying fusion in thermal face recognition

Applying fusion in thermal face recognition

According to recent studies, there are several algorithmic approaches used for thermal face recognition [ABB06, HRdSVC12]. The appearance based methods [SS06, FY02, Che03] deal with suche aligned images as a matrix of numbers. They are computationally efficient, but have some problems with head pose variance. Another approach is to extract features using Gabor filter bank [BPK04]. In [BPTB07], authors present thermal face recognition by vascular network extraction.

12 Mehr lesen

A Robust Face Recognition Algorithm for Real-World Applications

A Robust Face Recognition Algorithm for Real-World Applications

are lower, when the window size is W = 9, especially on the expeirments that contain occlusion. The main reason is, due to occlusion, face detection quality is very low in these cases, which in turn causes poor eye center position esti- mates that cannot be covered with a smaller window size. Another interesting observation that can be derived from the table is that, very high correct clas- sification rates are obtained against occlusion problem. Especially eye region occlusion is known to be one of the biggest challenges in face recognition. The obtained results imply that mainly the erroneous feature localization, thus im- precise face alignment causes the poor performance in the case of eye region occlusion. It is also intriguing to observe that the correct recognition rate ob- tained in the ARinterscarf experiment is lower than the one obtained in the ARintersun experiments, although lower face occlusion is known to be an easier problem than the upper face occlusion. The reason can be the textured surface of the scarfs which might affect the classification decision more than the black sunglasses. It should be also considered that, as already shown, the main is- sue with the upper face occlusion is the misalignment and once it is handled, very high performance can be reached. Note that the achieved correct identi- fication rates are significantly higher than the ones presented in the literature [GSC01, Mar02, TCZZ05, PLL05, FSL06, WGYM07, JM08].
Mehr anzeigen

167 Mehr lesen

Anzeige von Binarized Eigenphases for Limited Memory Face Recognition Applications

Anzeige von Binarized Eigenphases for Limited Memory Face Recognition Applications

Most of the appearance-based methods for face recognition that deal with the face images as a whole, depend on calculating the eigenvalues and the eigenvectors of a system representing this face space [8-10]. The calculations related to (and consequently the time required for) finding the eigenvectors and eigenvalues of the covariance matrix of the system is relatively huge. In our case, we can not just apply the principal component analysis (PCA) (or even other known face recognition method such as linear component analysis (LDA)) to the face images directly.
Mehr anzeigen

10 Mehr lesen

How (not) to measure bias in face recognition networks

How (not) to measure bias in face recognition networks

In this paper, we perform an in-depth exploration of a certain methodology for measuring the specific amount of bias that exists in any trained FR CNN. The underlying idea is appealing due to its intuitive approach and similar rea- soning has already been used to argue for specific bias removal algorithms in the past [2]. The quantification itself relies on internal cluster validation measures for clusterings of embeddings based on labels for race, gender and age. It is agnostic towards the specific architecture and training procedure of the model and thus applicable to any FR system that exposes its embeddings; it is also non-invasive with respect to model training and does not effect the model’s performance. Counterintuitively, our experiments speak against the validity of the idea and confirm the contrary: higher bias, as expressed in a drop in face recognition ac- curacy for minority cases, goes along with worse clustering, i.e. less “awareness” / more “blindness” of the model with respect to distinguishable features of the respective minority. We thus offer potential reasons for our observations, leading to preliminary results on how to better quantify bias in FR.
Mehr anzeigen

12 Mehr lesen

Face Recognition across Large Viewpoint Changes

Face Recognition across Large Viewpoint Changes

It is well known that human observers are capable of recognizing familiar faces from any of a number of viewpoints. By face recognition , we mean the classi cation of a face as \familiar"/\known" ver- sus \unfamilar"/\unknown". Clearly, this ability could be due to the fact that when a face is famil- iar to us, it is likely that we have encountered it previously in a variety of orientations. For unfa- miliar faces, however, even those encountered only from a single viewpoint, while human performance across large pose changes is not perfect, it is still well above chance [7]. This latter fact indicates that there is information available in a single view of a face to make relatively accurate recognition judgments even across quite large changes in view- point (e.g., full face to pro le). Intuitively, this is not surprising since much of the information that makes individual faces recognizable is visible from largely di erent poses/viewpoints. This kind of information includes both global features such as skin texture and tone, as well as very local features such as moles, blemishes and dimples.
Mehr anzeigen

9 Mehr lesen

Study on interactive face recognition

Study on interactive face recognition

11.2 R EPORT : V ICTOR R UCH The field of deep learning and face recognition were both new for me. I was excited to work on a term project where we could research in these fields. The term project was very interesting and challenging. I am satisfied with the results. The current state of face recognition technologies has surprised me. Michael and I have similar backgrounds and know each other for many years. We have done various projects together, which contributed to an efficient communication. I am glad Michael had previous experience in the field of deep learning. He took the time to explain many concepts to me and pointed out useful resources.
Mehr anzeigen

27 Mehr lesen

3D Face Recognition For Cows

3D Face Recognition For Cows

To improve the speed, implementing a faster and more accurate 3D face recognition method as explained in [Sp15] for cows on a bigger dataset would be an interesting experiment. Collecting data over a longer period of time from different types of cows will show if facial variations (natural or due to sickness) will affect the system’s performance. A big- ger dataset will enable further research on 3D cow face recognition using conventional and Deep Learning methods. If vision based systems out-perform the traditional electrical ones, cows will be free from IR tags around the neck.
Mehr anzeigen

9 Mehr lesen

FrFT Convolutional Face: Towards Robust Face Recognition Using Fractional Fourier Transform and Convolutional Neural Networks

FrFT Convolutional Face: Towards Robust Face Recognition Using Fractional Fourier Transform and Convolutional Neural Networks

to easily handle the issue of low-quality facial im- ages in facial recognition. Further, the selection of the order p ∈ (0, 1) in FrFT plays a vital role while designing the facial recognition systems. Theoret- ically, small p values result in a considerably struc- tured data representation; conversely, the features, such as texture, can be adequately extracted using large p values. The continuous FrFT of an image can be given as

4 Mehr lesen

Longitudinal examination of olfactory, verbal, and visual (face) recognition performance in different age groups

Longitudinal examination of olfactory, verbal, and visual (face) recognition performance in different age groups

Old age-related differences affect not every memory domain similarly. The LTM storage capacity stays unchanged in comparison to younger adult groups or decreases with age, especially in case of episodic memory (Salthouse, 1996). Meanwhile memory performance of STM and WM varies throughout the lifespan, showing a form of inverted U-shape of executive and inhibitory processes: performance increase in childhood and adolescence, having a peak in young adulthood and decline with adult age. General structure of STM recognition is similar in children, teenagers, and older adults but the WM capacity was found to be lower than in younger adults (Cowan, Naveh-Benjamin, Kilb, & Saults, 2006). The basis of this phenomenon could be explained with the inhibition-deficit hypothesis that describes an increasing difficulty in suppressing or down-regulating the activation of distracting thoughts and stimuli in the retrieval process with older age (Hasher & Zacks, 1988). When we try to recall a stimulus, various related options become conscious throughout activation of associative networks and thereby distract from retrieving the correct memory, down-regulation of networks allows staying focused. The consistency of this hypothesis was shown in many other studies on attention, WM and language processing (Aslan & Bauml, 2012).
Mehr anzeigen

60 Mehr lesen

Perceptual Narrowing in Speech and Face Recognition: Evidence for Intra-individual Cross-Domain Relations

Perceptual Narrowing in Speech and Face Recognition: Evidence for Intra-individual Cross-Domain Relations

During the first year of life, infants undergo perceptual narrowing in the domains of speech and face perception. This is typically characterized by improvements in infants’ abilities in discriminating among stimuli of familiar types, such as native speech tones and same-race faces. Simultaneously, infants begin to decline in their ability to discriminate among stimuli of types with which they have little experience, such as non- native tones and other-race faces. The similarity in time-frames during which perceptual narrowing seems to occur in the domains of speech and face perception has led some researchers to hypothesize that the perceptual narrowing in these domains could be driven by shared domain-general processes. To explore this hypothesis, we tested 53 Caucasian 9-month-old infants from monolingual German households on their ability to discriminate among non-native Cantonese speech tones, as well among same-race German faces and other-race Chinese faces. We tested the infants using an infant- controlled habituation-dishabituation paradigm, with infants’ preferences for looking at novel stimuli versus the habituated stimuli (dishabituation scores) acting as indicators of discrimination ability. As expected for their age, infants were able to discriminate between same-race faces, but not between other-race faces or non-native speech tones. Most interestingly, we found that infants’ dishabituation scores for the non-native speech tones and other-race faces showed significant positive correlations, while the dishabituation scores for non-native speech tones and same-race faces did not. These results therefore support the hypothesis that shared domain-general mechanisms may drive perceptual narrowing in the domains of speech and face perception.
Mehr anzeigen

5 Mehr lesen

Evaluation of wavelet based linear subspace techniques for face recognition / Hima Deepthi Vankayalapati

Evaluation of wavelet based linear subspace techniques for face recognition / Hima Deepthi Vankayalapati

The performance of different linear subspace techniques like principal component analysis (PCA), independent component analysis (ICA) and linear discriminant analysis (LDA) is evaluated [r]

114 Mehr lesen

Determinants of Successful Cooperation in a Face-to-Face Social Dilemma

Determinants of Successful Cooperation in a Face-to-Face Social Dilemma

(ii) Stake size Categorizing the jackpot in five divisions (see Table C.1 and Table C.2 in the Supplemen- tary Material) we find that the cooperation rate decreases with a step-wise increase in the jackpot size. But, surprisingly, the rate declines sharply from 73.6% for jackpots below £500 to roughly 50% for jackpots above £500. The difference, taking the cutoff £500, is highly significant (p=0.000). Concerning the mutual cooperation rate, it is significantly higher if the two players face a jackpot below the level of £500 (p=0.000). This result is even more remarkable if one bears in mind that a stake size around £500 is already much higher than the one used in most laboratory experiments. At the same time, however, the cooperation rate rises with an increase in the potential jackpot, i.e., the highest possible jackpot the players could obtain after the first phase of the final round. Hence, the effects of the actual and the potential jackpot operate in opposite directions. One might presume that the players’ perception of the actual jackpot depends on the potential jackpot, i.e., two actual jackpots equal in size will be judged differently depending on their difference to the potential jackpot. In Table III we explore this issue further.
Mehr anzeigen

51 Mehr lesen

Liebe per Mausklick. Unterscheiden sich online und face-to-face initiierte Paarbeziehungen?

Liebe per Mausklick. Unterscheiden sich online und face-to-face initiierte Paarbeziehungen?

Intimität, dafür aber weniger Leidenschaft aufweisen, als Paare, die sich face-to-face kennen- gelernt haben. An Hand dieses Ergebnisses lässt sich daher schlussfolgern, dass trotz der star- ken Intimität, die bei Paaren, die sich online kennengelernt haben, vorherrscht, diese immer noch Veränderungen unterliegen muss. Denn nach Baumeister und Bratslavsky (1999) han- delt es sich bei Leidenschaft um eine Funktion der Veränderung der Intimität: Sie kann nur ansteigen, solange die Intimität in der Paarbeziehung sich noch wandelt. Dies lässt die Ver- mutung zu, dass die Intimität noch nicht ihr Maximum erreicht hat oder, dass Paare, die sich online kennenglernt haben, auch weiterhin neue Erfahrungen miteinander machen, die in einer Veränderung der Intimität resultieren. Eine andere Möglichkeit wäre auch, dass sich diese Paare häufiger streiten und wieder versöhnen, so dass in Folge beide Beziehungsmerkmale, Intimität und Leidenschaft, hoch sind. Allerdings spricht die überwiegend positivere Beurtei- lung der Beziehung von Online-Paaren gegen diese Annahme. Es könnte aber auch sein, dass eine erhöhte Interdependenz zwischen den Partnern mit einer höheren Sensitivität für Verän- derungen der Intimität in der Paarbeziehung einhergeht, weshalb die Leidenschaft von Frauen, die ihren Partner online kennengelernt haben, positiver beurteilt wird als von denen, die ihren Partner face-to-face kennengelernt haben. Denn gemäß den Ergebnissen der Hypothese 4 zei- gen online initiierte Paarbeziehungen, die größeren Partnereffekte und demnach auch eine hö- here Interdependenz als offline initiierte Paarbeziehungen. Letztlich kann die Frage, warum Paare, die sich online kennengelernt haben, auch Leidenschaft positiver beurteilen, hier nicht abschließend geklärt werden. Dazu bedarf es weitere Untersuchungen, um die Ursache für dieses Ergebnis aufzeigen zu können.
Mehr anzeigen

219 Mehr lesen

Mediation in Teletandem : from face-to-face sessions to reflective journals

Mediation in Teletandem : from face-to-face sessions to reflective journals

reciprocal support and instruction between two learners, each of whom is a native speaker of the other’s target language, be it through face-to-face or any other mode of communication. In this broad sense tandem learning has likely existed in informal arrangements since people first sought to learn foreign languages. Tandem learning was thus not originally a theoretically driven method so much as a widespread practice, independently discovered many times over by individuals wishing to learn the language of an acquaintance and prepared to offer their own native-speaker expertise in return. (…) If this kind of more or less ad hoc arrangement was practised long before it was given the name ‘tandem’, though, what is a recent development is the theorisation and systematic exploitation of tandem as an open learning method. (O’ ROURKE 2007: 43)
Mehr anzeigen

25 Mehr lesen

Adjustment for differences between face to face and telephone interviews

Adjustment for differences between face to face and telephone interviews

In this case, it should not come as a surprise that the results are exactly correct because all estimates are based on the same data and determined by these data. The real test can only be done with new data where the response probabilities of this study are used and the nonresponse weights are obtained from a comparison of two companies who are doing the standard Eurobarometer study and the tracking study. However, this theoretical analysis shows that a prediction from the telephone data to the face to face data is possible.

17 Mehr lesen

Computergestützte Face-to-Face Interviews über das Internet mit Hilfe von MS NetMeeting

Computergestützte Face-to-Face Interviews über das Internet mit Hilfe von MS NetMeeting

Für die Durchführung der von uns entwickelten Methode wurde die Desktop- Conference Software NetMeeting von Microsoft verwandt. Desktop-Conference Systeme und Software gibt es zwar viel[r]

16 Mehr lesen

Die Interaktion im Auskunftsdienst : Face-to-Face, Telefon-, Email- und Chat-Auskunft im Test

Die Interaktion im Auskunftsdienst : Face-to-Face, Telefon-, Email- und Chat-Auskunft im Test

Auch wenn durch die geringe Anzahl an Testsituationen nur Tendenzen aufgezeigt werden können (und diese in unterschiedlichen Ausprägungen), schließt sich das Er- gebnis den in Kapitel 2[r]

126 Mehr lesen

„Das is doch keene Schmalspur“ – Eine Fallanalyse zur Bedeutungsaushandlung in der Face-to-Face- Kommunikation

„Das is doch keene Schmalspur“ – Eine Fallanalyse zur Bedeutungsaushandlung in der Face-to-Face- Kommunikation

wird. Eine Zusammenführung beider Forschungsrichtungen birgt ein großes Potential an Synergieeffekten, um die Bedeutungsaushandlung in der Face-to-Face- Kommunikation aus diesen zwei Perspektiven zu untersuchen. Auf der einen Seite kann Bedeutung damit als interaktional, indexikal und prozesshaft aushandelbar verstanden und untersucht werden. Auf der anderen Seite kann durch die Erweiterung um verschiedene Ausdrucksmodalitäten und Einbeziehung dieser in die Analyse ein deutlich dynamischeres Verständnis von Gesprächsführung geschaffen werden. So kann Bedeutung nicht nur in den verschiedenen Modalitäten an der einzelnen Sprecherin erforscht wird, sondern auch die affektive Dimension der Interaktion einbezogen werden (vgl. dazu auch Kappelhoff/Müller 2011:121ff.). Deutlich wird die Schnittmenge dieser beiden Disziplinen dann auch gerade im „inter“: Der Fokus kann von einer sprecherinzentrierten Analyse hin zu einem interaktionalen Miteinander verschoben werden, in welcher die subjektiven Erfahrungen jedoch nicht vollständig ausgeblendet werden. Bedeutungskonstitution kann damit multimodal und als gemeinschaftlich dynamisch im Gespräch untersucht werden.
Mehr anzeigen

14 Mehr lesen

Kosten-Nutzen der Teilnahme an Face-to-Face-, Telefon- und Onlinebefragungen: der Einfluss des Befragungsinstruments auf die Antwortrate

Kosten-Nutzen der Teilnahme an Face-to-Face-, Telefon- und Onlinebefragungen: der Einfluss des Befragungsinstruments auf die Antwortrate

und Antwortraten zu prognostizieren. Für die Entscheidung, an einer Befragung teil­ zunehmen sind zahlreiche Faktoren relevant. Daher wurden in dieser Arbeit solche Faktoren identifiziert, die sich abhängig vom Befragungsmodus unterscheiden. Im Sinne der zentralen Fragestellung haben sich Anonymität und Interaktion, Incen­ tives, Seriosität der Befragung, Länge und Aufwand der Befragung sowie die Stich­ probenziehung als relevante Variablen herausgestellt. Auf Grundlage dieser Varia­ blen lässt sich die höchste Teilnahmebereitschaft bei Face-to-Face-Befragungen erwarten. Dabei spielt insbesondere der Nutzen, der durch Incentivierung und durch die persönliche Interaktion mit dem Interviewer entsteht, eine wichtige Rolle. Der verwendete Ansatz konnte zeigen, dass Unterschiede in den Kosten und Nutzen der verschiedenen Befragungsmodi existieren und somit die Annahme der zentralen Fragestellung bestätigen. Doch auch wenn empirische Studien die grundsätzliche Prognose bestätigen, unterliegt das Modell dennoch wesentlichen Limitierungen. So kann das Modell aufgrund seines allgemeinen Charakters nicht auf die indivi­ duellen Präferenzen der Befragten eingehen, sondern nur allgemeine Kosten und Nutzen aufstellen. Dass die Teilnahme an einer Onlinebefragung für Rentner und Studierende vollkommen unterschiedliche Kosten und Nutzen bedeutet, kann mit dem Modell zum Beispiel nicht abgebildet werden. Die Kalkulation von Kosten und Nutzen basiert zudem nicht auf unmittelbaren empirischen Messungen, die die mit einer Skala abgefragten Werte direkt in das Modell überträgt, sondern auf einer Ana­ lyse theoretischer und empirischer Literatur. Des Weiteren, und dies mag eine sehr wichtige Einschränkung sein, ist die Teilnahmebereitschaft nicht nur von Ursachen, die in den jeweiligen Eigenschaften der Befragungsmodi angelegt sind, beeinflusst. Vielmehr spielen das Thema
Mehr anzeigen

25 Mehr lesen

Show all 889 documents...