Facial recognition system is a technology capable of identifying or verifying someone from a digital image or a video frame from a video source. There are several methods in which facial recognition systems work, but in general, they work by comparing the selected facial features of a given image with a face in the database.
Although originally a form of computer application, it has seen wider usage in recent times on mobile platforms and in other forms of technology, such as robotics.
These are typically used in security systems and can be compared with other biometrics such as fingerprint or eye recognition systems. More recently, it has also become popular as a commercial identification and marketing tool.
Video Facial recognition system
History of facial recognition technology
Automatic face recognition pioneers include Woody Bledsoe, Helen Chan Wolf, and Charles Bisson.
During 1964 and 1965, Bledsoe, along with Helen Chan and Charles Bisson, worked with computers to recognize human faces (Bledsoe 1966a, 1966b; Bledsoe and Chan 1965). He was proud of this work, but because funding was provided by an unnamed intelligence agent that did not allow much publicity, little of it was published. Given the large image database (basically, a mug image book) and a photo, the problem is to choose from a small database of records so that one of the drawings matches the photo. The success of a method can be measured in terms of the ratio of the answer list to the number of records in the database. Bledsoe (1966a) describes the following difficulties: 2
The project is labeled man-machine because humans extract coordinates a set of features from photographs, which are then used by computers to be recognized. Using a graphics tablet (GRAFACON or RAND TABLET), the operator will extract coordinate features such as student center, inner corner, outer corner of eye, peak widow point, and so on. From this coordinate, a list of 20 distances, such as the width of the mouth and the width of the eye, the pupil to the pupil, is calculated. These operators can process about 40 images per hour. When building the database, the person's name in the photo is associated with a list of calculated and stored spaces on the computer. At the recognition stage, the specified distance is compared to the appropriate distance for each photo, resulting in a distance between the photo and the database record. The nearest record is returned.
Since it is unlikely that the two images will fit in the rotation of the head, lean, tilt, and scale (distance from the camera), each set of distances is normalized to represent the face in the frontal orientation. To achieve this normalization, the first program tries to determine tilt, lean, and rotation. Then, using this angle, the computer cancels the effect of this transformation on the calculated distance. To calculate these angles, the computer must know the three-dimensional geometry of the head. Since the actual head is not available, Bledsoe (1964) uses a standard head derived from measurements on seven heads.
After Bledsoe left the PRI in 1966, this work continued at the Stanford Research Institute, primarily by Peter Hart. In an experiment conducted on a database of over 2000 photos, computers consistently outperformed humans when presented with the same recognition task (Bledsoe 1968). Peter Hart (1996) enthusiastically recalled the project with a cry, "It really works!"
Around 1997, the system developed by Christoph von der Malsburg and postgraduates of Bochum University in Germany and the University of Southern California in the United States outperformed most of the systems with the Massachusetts Institute of Technology and the University of Maryland assessed subsequently. The Bochum system was developed through funding by the United States Army Research Laboratory. The software is sold as ZN-Face and is used by customers such as Deutsche Bank and other airport operators and busy locations. The software is "powerful enough to make identification of the imperfect facial features, it can also often see through barriers to identification such as whiskers, beards, changing hairstyles and eyeglasses - even sunglasses".
In 2006, the performance of the latest face recognition algorithm was evaluated in the Face Recognition Grand Challenge (FRGC). High resolution face images, 3-D face scans, and iris images are used in the test. The results show that the new algorithm is 10 times more accurate than the face recognition algorithm of 2002 and 100 times more accurate than the algorithm of 1995. Some algorithms are able to outperform human participants in recognizing faces and can identify uniquely identical twins.
US government-sponsored evaluation and challenge issues have helped spur more than two orders of magnitude in the performance of facial recognition systems. Since 1993, the error rate of automatic face recognition system has decreased by a factor of 272. The reduction applies to systems that match people with facial pictures taken in a studio environment or mugshot. In Moore's legal terms, the error rate decreases by one-half every two years.
Low-resolution face images can be enhanced using facial hallucinations.
Maps Facial recognition system
Techniques for facial acquisition
Traditional
Some face recognition algorithms identify facial features by extracting landmarks, or features, from subject faces images. For example, an algorithm can analyze the relative position, size, and/or shape of eyes, nose, cheekbones, and jaws. These features are then used to search for other images with matching features.
Other algorithms normalize the facial image gallery and then compress the facial data, storing only data in the image useful for face recognition. The probe image is then compared to face data. One of the most successful early systems was based on a template matching technique applied to a set of prominent facial features, providing some sort of compressed face representation.
The recognition algorithm can be divided into two main approaches, geometric, which see distinguishing feature, or photometric, which is a statistical approach that filters images into values ​​and compares values ​​with templates to eliminate variance.
Popular recognition algorithms include major component analysis using eigenfaces, linear discriminant analysis, elastic bunching graphic matching using Fisherface algorithm, hidden Markov model, multilinear subspace learning using tensor representation, and neuronal motivated dynamic link matching.
3-Dimensional Introduction
3D face recognition techniques use 3D sensors to capture information about face shape. This information is then used to identify the characteristics of the facial surface, such as the contours of the eye sockets, nose, and chin.
One advantage of 3D face recognition is that it is unaffected by changes in lighting like any other technique. It can also identify faces from different angles, including profile views. The three-dimensional data point of the face greatly improves the accuracy of facial recognition. 3D research is enhanced by the development of advanced sensors that do a better job of capturing 3D face images. The sensor works by projecting a structured light to the face. Up to a dozen or so of these image sensors can be placed on the same CMOS chip - each sensor captures different parts of the spectrum.
Even the perfect 3D matching technique can be sensitive to expression. For that purpose the group in Technion appliance is applied from metric geometry to treat expression as isometry
The new method is to introduce a way to capture 3D images using three camera trackers pointing at different angles; one camera will point in front of the subject, the second to the side, and the third at the corner. All these cameras will work together so they can track the subject's face in real time and can detect and recognize faces.
Analysis of skin texture
Other trends emerge using visual skin details, such as those captured in a standard digital or scanned image. This technique, called Skin Texture Analysis, changes the lines, patterns, and unique spots that appear on a person's skin into a mathematical space.
Surface Texture Analysis, works in the same way as face recognition. An image is taken from a patch of skin, called a skinprint. Patches are then broken down into smaller blocks. Using algorithms to convert patches into mathematical and scalable spaces, the system will distinguish real lines, pores and skin textures. It can identify the differences between identical twins, which are not yet possible using facial recognition software only.
Tests have shown that with the addition of skin texture analysis, the performance in recognizing faces can increase by 20 to 25 percent.
Face recognition incorporates various techniques
Because each method has its advantages and disadvantages, the technology company has combined the introduction of 3D, Traditional and Leather Textual Analysis, to create a recognition system that has a higher success rate.
The combined technique has advantages over other systems. It is relatively insensitive to changes in expression, including blinking, frowning or smiling and having the ability to compensate for the mustache or the growth of the beard and the appearance of the glasses. The system is also uniform with respect to race and gender.
Thermal camera
Different forms of input data input for face recognition is by using a thermal camera, with this procedure the camera will only detect the shape of the head and will ignore subject accessories such as glasses, hats, or makeup. The problem with using thermal images for face recognition is that the database for face recognition is limited. Diego Socolinsky, and Andrea Selinger (2004) examine the use of thermal face recognition in real life, and the scenery of operations, and at the same time build new databases of thermal facial images. This study uses low-sensitive and low-resolution electro-electric sensors capable of obtaining long-wave thermal infrared (LWIR). The results show that LWIR smelting and regular visual cameras have greater results in outdoor inspection. Indoor results show that the visual has an accuracy of 97.05%, while LWIR has 93.93%, and Fusion has 98.40%, but outdoors proves visual has 67.06%, LWIR 83.03%, and fusion has 89.02%. The study used 240 subjects over a 10-week period to create a new database. Data collected on sunny days, rain, and cloudy.
Apps
Mobile platform
Social media
Social media platforms have adopted facial recognition capabilities to diversify their functionality to appeal to a wider user base amidst fierce competition from different applications.
Established in 2013, Looksery went on to raise money for his face modification app at Kickstarter. After successful crowdfunding, Looksery was launched in October 2014. This app allows video chat with others through a custom filter for faces that change the look of the user. Although there are additional image apps like FaceTune and Perfect365, they are limited to static images, while Looksery allows augmented reality for live video. By the end of 2015, SnapChat buys Looksery, which will then become the function of its monumental lenses.
SnapChat animated lenses, which use face recognition technology, revolutionize and redefine selfie, allowing users to add filters to change their look. Filter selection changes daily, some examples including one that makes users look like old and wrinkled versions of themselves, who water their skin, and that puts a virtual flower crown over their heads. The dog filter is the most popular filter that helps drive the success of SnapChat, with popular celebrities like Gigi Hadid, Kim Kardashian and people who like to post videos of themselves with dog filters.
DeepFace is an in-depth facial recognition system created by research groups on Facebook. It identifies the human face in a digital image. It employs nine-layer nerve webs with over 120 million connection weights, and is trained on four million images uploaded by Facebook users. The system is said to be 97% accurate, compared to 85% for the FBI Next Generation Identification system. One of the creators of the software, Yaniv Taigman, came to Facebook through the acquisition of Face.com in 2007.
Mobile technology
Face ID
Apple introduced the Face ID on the flagship iPhone X as the successor of biometric authentication to Touch ID, a fingerprint-based system. Face ID has a face recognition sensor that consists of two parts: the "Romeo" module that projects over 30,000 infrared points into the user's face, and the "Juliet" module that reads the pattern. This pattern is sent to the local "Secure Enclave" in the central device processing unit (CPU) to confirm the match to the phone owner's face. Facial patterns are not accessible by Apple. The system will not work with the eyes closed, in an attempt to prevent unauthorized access.
The technology learns from changing the appearance of the user, and therefore works with hats, scarves, glasses and a lot of sunglasses, beard and makeup.
It also works in the dark. This is done by using "Flood Illuminator", a special infrared flash that emits invisible infra-red light onto the user's face to read 30,000 focal points correctly.
Deployment in security services
Police
The Australian Border Force and New Zealand Customs Service have set up an automated border processing system called SmartGate that uses face recognition, which compares the nomad's face with data in the e-passport microchip. Major Canadian airports will be using a new face recognition program as part of the Main Inspection Kiosk program that will compare people's faces with their passports. The program will first come to Ottawa International Airport in early 2017 and to other airports by 2018. Panama's Tocumen International Airport operates an airport surveillance system using hundreds of face-recognizing cameras to identify individuals who want to pass through the airport.
Police forces in Britain have been testing live face recognition technology at public events since 2015. However, recent reports and investigations by Big Brother Watch found that the system is up to 98% inaccurate.
National security
The US State Department operates one of the largest face recognition systems in the world with a database of 117 million American adults, with photos usually taken from SIM photographs. Although still far from finished, it is used in certain cities to give a clue who is in the photo. The FBI uses photos as an investigative tool not for positive identification. In 2016, face recognition was used to identify people in photos taken by police in San Diego and Los Angeles (not on real-time video, and only on booking photos) and planned use in West Virginia and Dallas.
In recent years, Maryland has used face recognition by comparing people's faces with their SIM photos. The system attracted controversy when it was used in Baltimore to arrest the unruly protesters after the death of Freddie Gray in police custody. Many other countries use or develop similar systems but some states have laws that prohibit their use.
The FBI has also instituted the Next Generation Identification program to include face recognition, as well as more traditional biometrics such as fingerprints and iris scans, which can draw from criminal and civil databases.
In 2017, Time & amp; ClockedIn attendance company released facial recognition as a form of attendance tracking for businesses and organizations that want to have a more automated system of tracking working hours as well as for safety and health and safety control.
In May 2017, a man was arrested using an automatic face recognition system (AFR) mounted on a van operated by the South Wales Police. Ars Technica reported that "this seems to be the first time [AFR] has led to the capture".
By the end of 2017, China has applied face recognition technology in Xinjiang. Journalists visiting the area found surveillance cameras mounted every hundred meters in some cities, as well as facial recognition checkpoints in various places such as gas stations, shopping centers and mosque entrances.
Additional use
In addition to being used for security systems, authorities have found a number of other applications for facial recognition systems. While post-9/11 deployments were previously well-publicized experiments, more recent dissemination is rarely written about because of their secret nature.
At Super Bowl XXXV in January 2001, police in Tampa Bay, Florida used Viisage facial recognition software to search for potential criminals and terrorists present at the event. 19 persons with minor criminal records are potentially identified.
In the 2000 Mexican presidential election, the Mexican government used facial recognition software to prevent voter fraud. Some people have signed up to vote under several different names, in an attempt to put some votes. By comparing new face images with those already in the voter database, authorities may reduce duplicate enrollment. Similar technology is being used in the United States to prevent people from obtaining fake ID and SIM cards.
Facial recognition has been utilized as a form of biometric authentication for various computing platforms and devices; Android 4.0 "Ice Cream Sandwich" adds face recognition using the smartphone's front camera as a means of unlocking the device while Microsoft introduces face recognition login to the Xbox 360 game console via Kinect accessories, and Windows 10 via the "Windows Hello" Platform (which requires infrared- illuminated). Apple iPhone X smartphones introduce facial recognition to the product line with its "Face ID" platform, which uses an infrared illumination system.
The facial recognition system has also been used by photo management software to identify photo subjects, enable features such as looking for images per person, and suggest photos to share with a particular contact if their presence is detected in the photo.
Advantages and disadvantages
Compared to other biometric systems
One of the key advantages of facial recognition systems is being able to identify the mass of people because it does not require the cooperation of the test subjects to work. Properly designed systems installed at airports, multiplexes and other public places can identify individuals among the crowd, without passers even being aware of the system.
However, compared to other biometric techniques, facial recognition may not be reliable and efficient. Quality measurement is essential in facial recognition systems because large degree variations are possible in facial images. Factors such as lighting, expression, pose and sound during face shooting can affect the performance of facial recognition systems. Among all biometric systems, face recognition has the highest false acceptance and rejection rates, so questions have been raised on the effectiveness of facial recognition software in the case of rail and airport security.
Weakness
Ralph Gross, a researcher at the Carnegie Mellon Robotics Institute in 2008, explains one obstacle related to the facial angle: "Facial recognition has been pretty good on full frontal face and 20 degrees off, but as soon as you go to profile, there's already a problem "In addition to the pose variations, low resolution face images are also very difficult to recognize. This is one of the main facial recognition constraints in surveillance systems.
Facial recognition is less effective if facial expressions vary. A big smile can make the system less effective. For example: Canada, in 2009, only allowed neutral facial expressions in passport photos.
There is also an inaccuracy in the dataset used by the researcher. Researchers can use anywhere from several subjects to a number of subjects, and several hundred images to thousands of images. It is important for researchers to provide the datasets they use with each other, or at least have standard data sets.
Data privacy is a major concern when it comes to storing biometric data in a company. The storage of data about faces or biometrics can be accessed by third parties if not stored properly or hacked. At Techworld, Parris adds (2017), "Hackers are already trying to replicate people's faces to trick facial recognition systems, but this technology proves to be more difficult to hack than fingerprints or voice recognition technology in the past." Face recognition is used as additional security on various websites, phone apps, and payment methods, but the question in the minds of researchers is, Is the face recognition method safe for itself?
Ineffectiveness
Technological critics complain that the London Borough of Newham scheme, in 2004, has never admitted a single villain, although some criminals in the database system live in Borough and systems that have been running for several years. "Not once, as far as police know, has an automatic face recognition system that sees the target of life." This information seems to contradict the claim that the system was credited with a 34% reduction in crime (hence why it was launched to Birmingham as well). But it can be explained by the idea that when people are regularly told that they are under constant video surveillance with advanced facial recognition technology, this fear alone can reduce crime rates, whether facial recognition systems are technically functioning or not. This has been the basis for some other face recognition-based security systems, where the technology itself is not working well, but the user's perception of the technology.
An experiment in 2002 by the local police department in Tampa, Florida, also had disappointing results.
A system at Boston's Logan Airport was closed in 2003 after failing to make any match during the two-year testing period.
In 2014, Facebook stated that in a standard face recognition test of two standard options, its online system gained 97.25% accuracy, compared with a human benchmark of 97.5%.
In 2018, a report by civil liberties and rights campaign organization Big Brother Watch revealed that two British police, the Police of South Wales and the Metropolitan Police, used face-to-face recognition at public and public events, but with such accuracy as low as 2%. Their report also warned of the potential for significant human rights violations. It received extensive press coverage in the UK.
Frequently advertised systems have almost 100% accuracy, this is misleading because research often uses much smaller sample sizes than is necessary for large-scale applications. Since face recognition is not entirely accurate, it creates a list of potential matches. A human operator should look through this potential game and research shows the operator choosing the right match from the list for only about half the time. This is causing the problem to target the wrong suspect.
Controversy
Privacy violations
Civil rights organizations and privacy activists such as the Frontier Electronics Foundation, Big Brother Watch and the ACLU express concern that privacy is compromised by the use of surveillance technology. Some fear that it can lead to "total oversight society," with governments and other authorities who have the ability to know the whereabouts and activities of all citizens around the clock. This knowledge has been, is, and can continue to be deployed to prevent the exercise of the right of citizens by law to criticize those in power, certain government policies or corporate practices. Many centralized power structures with such supervisory abilities have abused their privileged access to maintaining control of political and economic apparatus, and to restrict populist reform.
Facial recognition can be used not only to identify individuals, but also to explore other personal data related to individuals - such as other photos featuring individuals, blog posts, social networking profiles, Internet behavior, travel patterns etc. - all through facial features only. Concern has arisen over who will have access to knowledge of the whereabouts of a person and those with them at any given time. In addition, individuals have limited ability to avoid or block facial recognition tracking unless they hide their faces. This fundamentally changes the dynamics of everyday privacy by allowing any marketer, government agent, or random stranger to secretly collect identities and related personal information from any individual caught by the facial recognition system. Consumers may not understand or realize what their data is used, which denies them the ability to approve how their personal information is shared.
Facial recognition is used in Russia to harass women suspected of involvement in online pornography. In Russia there is a 'FindFace' application that can identify faces with an accuracy of about 70% using a social media app called VK. This app will not be possible in other countries that do not use VK because their social media platform photos are not stored in the same way as VK.
In July 2012, the hearing was held before the Subcommittee on Privacy, Technology and Law of the Judiciary Committee, the United States Senate, to address issues surrounding what facial recognition technology means for privacy and civil liberties.
In 2014, the National Telecommunications and Information Association (NTIA) initiated a multi-stakeholder process to engage privacy advocates and industry representatives to set guidelines on the use of facial recognition technology by private companies. In June 2015, privacy advocates left the bargaining table for what they felt was an impasse based on industry representatives who did not agree to the terms of approval for face recognition data collection. NTIA and industry representatives continue without a privacy representative, and draft rules are expected to be presented in the spring of 2016.
In July 2015, the United States Government Accountability Office reported to Rank Members, the Privacy, Technological and Legal Subcommittee, the Judiciary Committee, the US Senate. This report discusses the commercial use of face recognition technology, privacy issues, and applicable federal law. It states that previously, issues regarding facial recognition technology are discussed and represent the need for updated federal privacy legislation that is constantly in line with the degree and impact of advanced technology. Also, that some industry, government, and privacy organizations are in the process of developing, or have developed, "voluntary privacy guidelines". These guidelines vary among groups, but overall aim to gain consent and inform citizens about the use of desired facial recognition technology. This helps overcome the privacy concerns that arise when citizens do not know where their personal data is, the privacy can be used as reporting shows as a common problem.
The greatest concern with the development of biometric technology, and more specifically facial recognition, has to do with privacy. The rise in face recognition technology has made people worried that big companies, like Google or Apple, or even government agencies will use it for public mass monitoring. Regardless of whether they have committed a crime or not, in general people do not want their every action to be monitored or tracked. People tend to believe that, since we live in a free society, we must be able to go out in public without fear of being identified and observed. People worry that with the increased prevalence of facial recognition, they will begin to lose their identity.
Facebook DeepFace
Social media websites like Facebook have a large number of people's photos, annotated with names. This is a database that may be misused by the government for facial recognition purposes. DeepFace Facebook has been the subject of some class action lawsuits under the Biometric Information Privacy Act, with claims that Facebook collects and stores face recognition data without obtaining informed consent, a direct violation of the Biometric Information Privacy Act. The latest case was dismissed in January 2016 because the court has no jurisdiction. Therefore, it remains unclear whether the Biometric Information Privacy Act will be effective in protecting the privacy rights of biometric data.
In December 2017, Facebook launched a new feature that notifies users when someone uploads a photo that includes what Facebook says is their face, even if they are not tagged. Facebook has been trying to frame a new function in a positive light, in the midst of a previous counterattack. Facebook's head of privacy, Rob Sherman, discussed this new feature as a feature that gives people more control over their photos online. "We have thought of this as a very empowering feature," he said. "There may be photographs that you do not know."
Technology is not perfect in law enforcement
Around the world, law enforcement agencies are beginning to use facial recognition software to help identify criminals. For example, the Chinese police force was able to identify the twenty-five people suspected of using facial recognition equipment at the Qingdao International Beer Festival, one of which has been running for 10 years. The equipment works by recording a 15 second video clip and taking some subject photos. The data were compared and analyzed with images from the police database and within 20 minutes, the subject could be identified with 98.1% accuracy. In the UK, the use of face recognition technology by police has been found to be up to 98% inaccurate.
Facial recognition technology has been shown to work less accurately in colored people. One study by Joy Buolamwini (MIT Media Lab) and Timnit Gebru (Microsoft Research) found that the rate of error for gender recognition for female color in three commercial face recognition systems ranged from 23.8% to 36%, while for lighter-skinned men was between 0.0 and 1.6%. The overall accuracy rate for identifying males (91.9%) was higher than for women (79.4%), and no system accommodated non-binary gender understanding.
Experts fear that the new technology may actually hurt the community that police say they are trying to protect. This is considered an imperfect biometric, and in a study conducted by Georgetown University researcher Clare Garvie, he concluded that "there is no consensus in the scientific community that provides a positive identification of a person." It is believed that by a very large margin of error. in this technology, both legal advocates and facsimile software companies say that technology should only provide part of the case - not evidence that can lead to individual capture.
The lack of regulations holding facial recognition technology companies for rigorous bias testing requirements can be a significant flaw in the application of use in law enforcement. CyberExtruder, the company that markets itself to law enforcement says they have not done any testing or research on bias in their software. CyberExtruder does note that some skin colors are more difficult for software to recognize with the limitations of current technology. "Just as individuals with very dark skin are difficult to identify with high significance through facial recognition, individuals with very pale skin are the same," said Blake Senftner, a senior software engineer at CyberExtruder.
Emotional detection
The facial recognition system has been used for emotional recognition. In 2016, Facebook acquired FacioMetrics emotional detection startup.
Face recognition system
In January 2013, Japanese researchers from the National Institute of Informatics made the 'privacy visor' glasses that use infrared light almost made the face underneath it not recognized by facial recognition software. The latest version uses a titanium frame, light reflecting material and a mask that uses angles and patterns to disrupt facial recognition technology through light sources that absorb and bounce back. In December 2016, an anti-CCTV form and face recognition glasses called 'reflectacles' were invented by Chicago-based custom-spectacle craftsman Scott Urban. They reflect infrared and, optionally, visible light that makes the user face blurry white for the camera.
Other methods of protecting from facial recognition systems are special haircuts and makeup patterns that prevent the algorithm used to detect faces.
See also
- List
- List of computer vision topics
- List of emerging technologies
- Outline of Artificial Intelligence
References
Tucker, J (2014). How faicial cam recognition technology becomes, taken from https://www.bostonglobe.com/ideas/2014/11/23/facial-recognition-technology-goes-way- return/CkWaxzozvFcveQ7kvdLHGI/story.html
Further reading
- What is Biometrics? White Book Published by Aware, Inc., January 2014
-
Farokhi, Sajad; Shamsuddin, Siti Mariyam; Flusser, Jan; Sheikh, U.U; Khansari, Mohammad; Jafari-Khouzani, Kourosh (2014). "The infra-red face recognition is close to combining Zernike's moments and discrete wavelet transformations that can not be measured". Digital Signal Processing . 31 (1). doi: 10.1016/j.dsp.2014.04.008. - "The Face Detection Algorithm Set to Revolutionize Image Search" (February 2015), MIT Technology Review
- Garvie, Clare; Bedoya, Alvaro; Frankle, Jonathan (October 18, 2016). Line Up Perpetual: Police Face Recognition Not Regulated in America . Privacy Center & amp; Technology at Georgetown Law . Retrieved October 22 2016 . Ã,
-
Like It Happened . Canadian Broadcasting Corporation. October 20, 2016 . Retrieved October 22 2016 . Ã, Interview with Alvaro Bedoya, executive director of the Privacy Center & amp; Technology at Georgetown Law and coauthor of Perpetual Line Up: Uncovered Police Uncovered in America .
-
External links
- Media related to the facial recognition system on Wikimedia Commons
- A Stereo Photometric Face Recognition Approach ", University of Western England. Http://www1.uwe.ac.uk/et/mvl/projects/facerecognition.aspx
Source of the article : Wikipedia
0 Comments