The Technical Characteristics And Future Development Trend Of Face Recognition In China Are Explained In Detail.
- May 23, 2018 -

In recent years, face recognition technology has been widely used in the field of security. With the continuous development of technology, it is becoming more and more close to our daily life. It can be seen by mobile phones, shopping malls and parks. According to the related analysis report, it is expected that the face recognition market will grow at about 20% growth this year, and the global face recognition market will reach US $7 billion 595 million by 2022. It can be said that the next few years will be the key year for the maturity and popularity of face recognition technology.


Face recognition, also known as face recognition, is a biometric technology based on human face feature information for personal identification. It uses video cameras or cameras to collect images or video streams that contain faces, and automatically detects and tracks faces in the image, and then makes a series of related techniques for the face of the detected face.



Domestic mainstream face recognition enterprises and features


At present, there are many enterprises engaged in face recognition technology worldwide, such as Google, Facebook, Standford, Carnegie and so on. In these years, many power groups have emerged in China, such as Hai Kang Wei, science and technology, science and technology, and so on.


These mainstream face recognition enterprises in China have their own technical solutions. For example, Hikvision's face recognition technology, its human face recognition system is very good in recognition accuracy, the front-end intelligence - deep eyes, distributed intelligence - Hyper brain NVR, Central Intelligence - Hai Kang Facebook, both fusion depth learning algorithm, front and back depth intelligence to provide powerful support for large data applications.


Face recognition system is widely used in all walks of life, such as Ping An City, transportation hub, intelligent medical treatment, commercial chain, bank, school, park, frontier inspection and so on.


There are many breakthroughs in the accuracy rate of face recognition technology. The off-line face recognition and video stream detection technology are also very bright. The off-line face recognition is located in the detected face frame to further locate the five features and contour positions of the face. At present, the location of the 21 point model of key points and the key point of the face are used. The average accuracy is up to 96%. Video stream face detection and key point detection and face detection in static pictures are used in off-line state, which can improve the response time. This technique can be used in image editing and other fields.


Its face recognition technology is widely applied in many fields, such as attendance system, remote authentication, access control system, entertainment application and so on.


Face recognition technology based on Shang Tang technology shows good accuracy and delay in face calibration of video streaming, and face recognition algorithm is also one of its advantages. According to the OFweek artificial intelligence network, Shang Tang technology has been used in the international authoritative face database LFW. The accuracy of human face recognition has surpassed the human eye for the first time. In the field of security, it introduces the static face alignment system SenseTotem (totem) and dynamic face alignment SenseFace. Totem is a set of map searching system, which can be used to collect face screenshots in monitoring video, compare the standard face photos in the search target library, and help the police to quickly confirm the identity of suspects involved in the case, and SenseFace is used for the real time large database face recognition in the large-scale visual frequency monitoring system in airports, railway stations and other public places. It can provide real-time capture of face, attribute recognition and key personnel trajectory reduction in surveillance video.


In addition to the application of human face recognition technology in finance, entrance guard, unmanned driving and so on, such as small coffee show, panda live broadcast and so on all kinds of facial AR special effects provided, behind the technical support provided by the merchant soup.


Cloud technology face recognition technology is very distinctive. Its face recognition technology uses structural light body anti attack detection. It has a good effect in improving the accuracy rate. The 3D face recognition system based on structured light technology has made a revolutionary breakthrough in accuracy, response speed and in vivo detection. 3D structure optical face recognition technology can be widely used in the Internet of things, mobile Internet, bank, security, traffic and other human face recognition scenes, which can better improve the effect of attack prevention, improve the accuracy of face recognition, and combine the latest algorithm with the 1/10000000 error rate. More than 99% of the accuracy rate.


Its human face recognition technology is widely used in banks, public security systems, shopping malls, etc., and the establishment of a joint laboratory with the Ministry of public security, the state-owned large commercial banks, stock exchange shares and the General Administration of civil aviation to promote the establishment of artificial intelligence product standards.


The current domestic face recognition technology is facing the pain point in the popularization of the market.


As far as the most popular area of face recognition is concerned, security, banking and public security systems are the most widely used. In the future, face can shine in many fields, which depends on breakthroughs and innovations in technology. As far as the current domestic situation is concerned, face recognition is faced with two major pain points in the market popularization.


First, the business model needs to be improved


The popularity of a new technology is related to many factors. As an enterprise, cost control and profitability are the purposes. The face recognition technology has a high threshold and a large investment, and the application scene in the short term is very limited, so many enterprises appear hesitant in the layout, and want to make face recognition technology enter thousands of families like online shopping. A good business model is very important.


In the popularization of face recognition technology, many domestic companies have also done a lot of efforts, such as Baidu has introduced a product called face best, it is an attempt application of Baidu in artificial intelligence technology. Some people think that the introduction of this application makes APP of mobile phone break from "mode age" to "the age of operation". Because the application is not only the user and the future business model, more is connected to the Baidu artificial intelligence and face recognition technology, with a high enough technical threshold to realize the difficulty. Internet giants such as Alibaba and Tencent have tried to achieve business realisation based on face recognition technology, but the results are not as good as expected.


Two, the current face recognition technology still has room for improvement.


The current face recognition system mainly consists of four components: face image acquisition and detection, face image preprocessing, face image feature extraction, and matching and recognition. With the growth of future data, data matching will face greater challenges.


Face recognition is very useful, but at present, face recognition technology has many shortcomings, such as sensitive to ambient light environment, it may affect the accuracy of recognition. In the face of dark, similar and other complex scenes, the recognition rate is poor.


In recent years, although domestic enterprises have made certain achievements in face recognition technology, there are still many aspects to be done, such as window glass recognition, night recognition, infrared recognition and other technologies to be developed by enterprises.


Where can face recognition technology be improved at present?


To make progress in face recognition technology, we must seek breakthroughs in several key technical points.


Feature based face detection technology is based on color, contour, texture, structure or histogram features for face detection.


Based on template matching face detection technology, the face template is extracted from the database, then a template matching strategy is adopted to match the face image with the template library, and the size of the face and the position information are determined by the height of the correlation and the size of the template matching.


The face detection technology based on Statistics - the face and negative sample library of face and non face are collected in a large amount, and the system is trained by statistical method to detect and classify face and non face patterns.


From the technical level, face recognition can be improved from the above three aspects. Any breakthroughs in the details of the technology may bring about an improvement in the recognition rate.


In addition to the above three technical aspects of face recognition, there are many upstream enterprises to improve the face recognition technology, for example, improving the accuracy of face recognition by improving the components.


The infrared light source used for the user's face or eye is not only bright enough but also uniform, which is especially important for face recognition and eye tracking system. OSRAM's latest Synios P2720 uses infrared wavelengths of 940 nm to reduce red explosion. The infrared camera has the best sensitivity to the infrared light source of 850nm.


According to the OFweek artificial intelligence network, the IRED is designed for two-dimensional face recognition and is the latest member of the existing biometric product line for OSRAM photoelectric semiconductors.


The main research direction and development prospect of face recognition


With the advent of the era of big data and sharing, the issue of data security has become more and more important. The new generation of technology revolution represented by face recognition has been launched. The requirements of these technologies are getting higher and higher, which not only require the accuracy of the data, but also ensure the security of the data. The face recognition is very promising. As the main force of the industry, the technological strength and innovation ability of the enterprise determines the trend of the whole industry, and any little technological innovation may bring about the reform of the industry.


The main research direction of face recognition in the future will be around some problems, such as the similarity of face structure, face attitude, age change, light change of complex environment, face decorating and so on.


The development of face recognition is not only restricted by its own technology, but also closely related to the whole industry. In order to enable the early popularization of artificial intelligence, many enterprises are also active in the layout of face recognition, which includes both the BAT Internet giant and the enterprise for improving the optical devices for face recognition. Ma Yun investment face recognition technology company is open to technology and Shang Tang technology, Tencent set up the map laboratory, focus on image processing, deep learning and other fields of technology development; Baidu has also set up a face recognition team, research and development of core technology. OSRAM, the optical giant, recently acquired Vixar Inc. to strengthen 3D facial recognition skills and actively improve face recognition technology. By introducing Vixar technology in vertical cavity surface emitting lasers (VCSEL), OSRAM will master more security technologies including ultra precision 3D facial recognition, which can be used not only for solutions. Locking smart phones and consumer electronic devices can also be applied to industries requiring higher security access control.


Relying on the rapid advance of the Internet of things and artificial intelligence, the application scene of face recognition will be more and more extensive. With the R & D investment of the national scientific research institutions, the technological research of enterprises and the promotion of the market, these will be a sign of the bright future of face recognition. According to industry analysts, face recognition will become the mainstream of effective identification in the future, and face recognition is not a new word.