Face recognition technology and application cases

- Jul 09, 2018-

Strong demand for face recognition downstream applications, the market is expected to usher in explosive growth.

In 2018, artificial intelligence, Internet of Things, and blockchain technologies have become hot, and at the same time, biometric recognition technology represented by face recognition has attracted more and more attention. Face recognition is based on human facial feature information for identity recognition. In recent years, this technology has continuously made new breakthroughs. During the Spring Festival, the concept of “Olympic Pass” at the PyeongChang Winter Olympics allowed the Olympic Games to enter the “brushing era”; in addition, electronic ID cards using face recognition technology in some provinces and cities nationwide have been activated; in addition, the technology is still being governed. Scenes such as “Chinese-style crossing the road” and assisting the public to capture fugitives have once again brushed out the sense of existence and are getting closer to people’s lives.

The domestic face recognition market is expanding. What are the advantages of this technology compared to other biometric technologies? The current six popular biometric technologies, face recognition, fingerprint recognition, palmprint recognition, iris (or retina) recognition, and voice recognition With vein recognition, these techniques have their applicability. Among them, face recognition, as the most widely used and most convenient biometric technology, has obvious advantages in terms of security, cost, ease of use, and public acceptance. It has excellent solutions in both civil and commercial fields. Program.

Face recognition technology has developed over the past decades, and the country has attached great importance to it for many years. Nowadays, the domestic industrial chain has become more and more perfect, and the basic conditions of software and hardware for large-scale popularization have been available. With the increasing market demand for downstream applications of face recognition technology (especially finance, security, Internet, etc.), the industry is about to explode.

The relevant analysis report shows that the face recognition market is expected to maintain a growth rate of around 20%. By 2022, the global face recognition market will reach $7.595 billion. Among them, the size of the Chinese face recognition market accounts for about 10% of the global market, and it is expected to continue to expand in the future.

Two major application directions and advantages and disadvantages of face recognition

In the era when people are connected with each other, the first thing to solve around people's service is to identify people's identity. For this reason, people have invented many cards and certificates as the basis for identifying people's identity. This identity is essentially "seeing things like See people, recognize things and don't recognize people." In order to comply with the trend of the times, making full use of the advantages of face recognition technology, emphasizing the sharing and opening of face big data, it is very necessary to build a portrait library and a face bayonet system.

The main application directions of face recognition are face control and face verification (ie, person-in-one).

Face control refers to obtaining the face in the video through the device and comparing it with the control library (blacklist). When the similarity is higher than a certain value, the system prompts and records. The main application areas are the national security level, and the scene of the safe city of the safe city.

Face verification is a 1:1 comparison of a face or ID photo with a live video; or a 1:N screening comparison from a self-built face database. The main application areas are public security, information security, and financial security. For example, conference delegate identification, confidentiality control, and user authentication.

Regardless of the direction, face recognition technology has the same advantages and disadvantages. The advantage is that the identified person does not need to cooperate with the identification device, and the image can be acquired and compared during the traveling process; the user is easy to accept without direct contact with the device; the concealment is good, the result is intuitive; the cost is moderate, and the security is high. The disadvantage is that it is also a focus of attention due to light factors, shelters, post-production, cosmetic surgery, and the judgment of twins and similar faces.

In the context of new technologies, how to use biometrics as the basis for verification and certification, and to solve problems such as fraud, fraudulent use, replacement, etc., Skod has made substantial progress.

Typical case analysis: face verification system and human identity verification system

Regarding the face verification system, it solves the problem of light and motion blur through strict algorithm control technology. In turn, it is possible to instantaneously capture high-quality faces in various environments and achieve high-precision comparison. At the same time, the document is still the main part of the verification as the irreplaceable main certificate, and face recognition is an important supplement to the verification. The irreplaceability of documents is mainly reflected in the authorization-authorization to enable non-useless judgment, pre-trial-management department review before issuing certificates, security-solving similar faces, make-up facelifting and other issues.

The documents used in the face verification system are not limited to “identity cards”. Users can issue their own documents, such as work permits, pass cards, student cards, annual cards, representative cards, passes, and visitor receipts.

And the questions that users are very concerned about, such as the face comparison charging mode, the face comparison can not be used off the network, the face comparison library capacity is limited; there is the inability to integrate with other business systems, etc., users do not have to worry . Since the face verification system algorithm is deployed locally, the WEB comparison service fee is not generated, and the network can be used without any network; in addition, the system is applicable to applications of various sizes and can interact with any business system.

The face verification system can perform a 1:1 comparison of the face or ID photo with the live video, or a 1:N screening comparison from the self-built face database. Application areas include: conferences, personnel management, safe area access, and scenic spots. The system not only realizes the precise management of the verification identity, but also refines the consumption behavior from the data, providing decision-making basis for more humanized management.

For the human-integrated verification system, how does the system determine the facial similarity of the target person and release it after the face capture machine starts the capture? The whole process of face recognition is from face judgment, face tracking, face capture, face The comparison and comparison results feedback consists of a series of links. The focus and difficulty of face recognition is how to capture a clear and well-characterized face, especially from the face of a moving person. Thus, the on-site environment (such as light) and tracking face to capture timing control is the key to solving the above problems.

The face verification system perfectly solves the problems of light, low light, back light and motion blur through strict algorithm control technology, which can realize high-precision comparison in a variety of environments. Similarly, the whole process of comparison operation is completed in the local server, without passing through the external network, without generating face matching costs, realizing the closed-loop management of the self-built face database.

New requirements in the new form of security prevention New journey of face recognition

The face recognition system can be widely used in public security, finance, airports, subways, border ports and other important areas for the natural identification of personnel identification. For example, the face and the ID card are matched and checked to ensure that the person's identity is in one; the face evidence is retained, and the ability to collect evidence afterwards is enhanced. In large-scale event meetings, the introduction of face recognition technology is a new requirement in the new form of security prevention. From the perspective of public security management, it can provide effective monitoring data for the public security department. From the perspective of the event organizer, it is possible to make a one-to-one correspondence with the ticket and prevent the use of the ticket.

In addition, when it comes to the application cases that have been successfully landed, the application case of face recognition in the 3rd Maritime Silk Road International Art Festival is very representative. Through RFID + face recognition technology to achieve intelligent conference management, provide online invitation registration, background multiple audit, on-site face-to-face, data real-time monitoring of a full set of business services; to achieve staff, participants to face the face, improve sign-on management efficiency With the user experience. After testing, the face recognition accuracy rate is 99.9%, and the response rate reaches the millisecond level.

In the end, it is also said that the face plane image is only the projection of the three-dimensional face on the plane, in the process, part of the information will be lost. How to collect, analyze and compare living parts through special face camera and computer image processing technology to make accurate and rapid identification is the embodiment of face recognition maturity. Face verification uses dynamic face recognition and static face recognition in different scenarios, and gives full play to their respective strengths to achieve the best recognition.

Things always move forward in constant movement. Skod's face verification technology is constantly pursuing perfection in the process of practice, understanding, practice and re-recognition.