With the maturity of hardware and software technology, face recognition technology is expected to enter a real outbreak period.
Author: huifan Time: 2019-11-20
In recent years, many app software have provided consumers with a unique brush face experience. In the field of biometrics development, face recognition has great potential. Many research teams are also committed to organically combining face recognition and life scene applications to truly serve the people.
Face recognition, also known as face recognition, or face recognition, is a method of capturing an image or video stream containing a face by using a camera or a camera, and automatically detecting and tracking the face in the image, thereby performing the detected face. A series of related techniques for face recognition.
In recent years, the broad market prospects of the face recognition industry have attracted various capitals. In 2014, Oubit acquired 100% equity of Platinum Information for 525 million yuan, thus entering the face recognition business and intelligent security field; in April 2015, Jiadu Technology 50 million yuan strategic investment focused on face recognition intelligent analysis algorithm and Product development of cloud from information technology; in September 2016, Ant Financial Services 70 million US dollars to acquire the US biometrics company EyeVerify, further deepen its layout in the field of biometrics.
It can be said that with the increasing popularity of face recognition, this biometric technology, which has existed for more than 10 years, is suddenly exposed to the spotlight. However, while bringing convenience, the potential threats are also prohibitive. In the United States, the hidden dangers brought by the face recognition technology have caused many technology giants such as Facebook and Google to get into the fire.
According to the "2016-2020 China Face Recognition Industry Investment Analysis and Forecast Report", from the non-military application field of face recognition, the current domestic face recognition technology is basically in the financial and security fields. In the field of entertainment, only Baidu's face is really using face recognition technology in the background. In other fields, face recognition is not possible in the short term due to technical difficulty and investment.
It is reported that the current face recognition algorithm is still less than 100% accurate. Moreover, compared with fingerprints and irises, the face shape is not constant for life, and it is likely to change with time. This is also a difficult point that cannot be avoided in face recognition technology. The current face recognition technology can only be identified in the case of hair changes, conventional makeup, a certain degree of fat and thin changes, natural aging and the like. But when the face changes dramatically, the computer's face recognition technology can't do anything.
face recognition technology
china company,face attendance machine china manufacturer
In addition, face recognition is sensitive to the scene lighting environment. The current face recognition technology mainly recognizes the relative position of the face and the facial features by plane 2D, and it has the possibility of deceiving the machine in the form of a photo. However, in the case of no cooperation, identification deviation may occur, such as the situation where your cousin is more like you than you. But overall, the accuracy of face recognition is worthy of recognition. It is believed that with the maturity of hardware and software technology, face recognition technology is expected to enter a real outbreak period.
Jules Verne in his most neglected book, "A Diary of American Journalists in 2890 AD," once described the magical and magnificent description of the world after the millennium:
At that time, New York was called Universal City. There were thousands of feet of skyscrapers on both sides of the road. The interior was heated and air-conditioned. The climate was completely controlled by humans. The crops could grow in the Arctic. People traveled on 1,500 km/h of submarine hydraulic pipelines to and from all continents. Advertisements can be mapped in the cloud. Journalists relay news from Jupiter, Mars, and Venus. People sitting at home can understand the world.
But today, a hundred years later, many imaginations have become reality.
The speed of technological progress can't even keep up with the sub-hands. "It is obvious that you can eat your face, but you have to rely on talent." Biometric technology is rapidly evolving with the support of artificial intelligence and big data. Brushing your face as a common life scene More and more mature, brushing the face like a card is almost a trend.
Face recognition opens more imagination space outside of authentication
At the airport and the bank, you will be asked to put the right Pose in front of the camera. This is not surprising. Alipay's face landing and payment are not unfamiliar to most people. Even the driver of the network has to brush the face line. .
At the Wuzhen Drama Festival in October this year, Baidu also guested a “special host” to identify hundreds of international guests of different races and ages in real time among thousands of viewers. Deep-learning and imaginary breakthroughs in artificial intelligence, Wu Enda said at the Shenzhen IT Leaders Summit that the error rate of Baidu face recognition is 0.16%, which is higher than Google's 0.37%, which certainly supports the extension and popularity of technology.
The difficulty of face recognition lies not in the acquisition and recognition of images, but in the judgment of artificial intelligence under the support of technology, logic and big data. The addition of convolutional neural network makes face recognition technology get rid of serious single identity verification and live detection. Come out and create more fun scenes.
The most interesting performance is that face recognition achieves a series of features such as age, gender, expression, etc., and learns to analyze the specific state of the person. For example, people's daily micro-expressions are mainly seven kinds, joy, anger, sadness, Surprise, disgust, fear, neutrality, coupled with the highly accurate five-position positioning, the system can accurately capture the emotions of the interacters.
How to make the offline consumption scene also "IQ online"?
Traditional enterprise game marketing often has breakpoints. This is a mismatch between marketing methods and technology. At the same time, high-cost online access, offline interactivity and lack of entertainment, it is difficult to effectively integrate the entire process of consumption.
In a series of scenarios, artificial intelligence actually has the ability to comprehensively digitize the attributes of people, behavioral intentions, hobbies, etc., combined with image, face and speech recognition, based on the targeted communication, Help traditional enterprises to rebuild and upgrade their brand power.
When Jules Verne had passionately portrayed the way of future advertising, “it averages $3 million a day. Due to a clever system, some of the ads are spread in new forms, which is done with clouds. Huge advertisements can be seen by people all over the region. Thousands of projectors in this corridor continue to launch large-scale advertisements to the clouds, and the clouds are displayed in color."
Today's technology can make the marketing effect better than Verne's imagination. It is not only more accurate, but also the means of communication and the medium of communication. On December 23, KFC and Baidu cooperated in the artificial intelligence experience store. The opening of Beijing Financial Street has applied the cutting-edge technologies built around artificial intelligence to offline scenes for the first time.
The first is to use Baidu face recognition technology to play KEMOJI expressions, to recognize the mood after obtaining facial information, easy to use and fun, the most surprising is the mood to order, the camera scans facial features according to the user's age, gender, image, For elements such as mood, give appropriate labels and recommend matching KFC packages. In the second consumption, the user can quickly identify by taking a photo, and the order is fast, and the integrated payment link becomes a complete consumption closed loop.
Behind the fun scene is the powerful transformation of technology driven by traffic portals.
Taking face recognition as an example, complex techniques must be implemented in three steps:
The first is to acquire the facial features of the human body through photoelectric sensors, which becomes very easy after the popularity of smartphones;
The second is to digitize the acquired information. Finally, the most important thing is the big data comparison after the algorithm is processed.
In general, the recognition rate of human naked eyes is 97.52%, and Li Yanhong said at Baidu World Congress that the success rate of Baidu face recognition technology is over 99.7%, which is the technical guarantee for entering the era of brushing.
But the further evolution of face recognition technology depends on two points.
The foundation is the synergy of peripheral technologies.
For example, the Baidu artificial intelligence system with Baidu brain as the core relies on the simulation neural network, the computing power built on hundreds of thousands of servers, and the billions of network big data every day. The front-end application scenario is based on this. It is the most vital, such as face recognition, such as voice ability.
This is followed by the aggregation of online traffic portals and offline application scenarios.
face attendance machine
The wide application of face recognition technology depends on the support of online traffic portal, because whether it is the magnificent scene built by AR or the various fun applications after brushing, it relies on the powerful drainage of the entry-level APP to be able to mature with offline consumption. The scene forms a benign interaction.
Technology + the future of the scene
The boundaries between the online and offline of the second half of the Internet will be broken. The two most important forces of the Chinese economy will be integrated by the promotion of consumption. The first to change is the traditional enterprises.
For a long time, traditional companies have done a lot of self-proclaimed market-free marketing, although it seems to be lively, it has long-term damage to the brand, and the rise of artificial intelligence brings another idea, that is, through Technical means to make consumers more precise, more accessible and interactive.
For the Internet platform, the cooperation with the mature offline scene is also a process from quantitative change to qualitative change. The success rate of face recognition today exceeds the human eye, relying on the self-learning ability after a large amount of data accumulation. Beyond the scene, Tencent is desperately trying to use the small program and short video to explore the potential of WeChat. Ant Financial is desperate to put social elements into the financial scene. Baidu tries to use AR technology and face recognition as a guide to play with the end consumers. Interactions, such as the use of secrets as assistants, are now half-interactive through voice and images, so creating a large number of offline applications is critical to further technology.
It is conceivable that on the basis of adding cutting-edge technologies such as AR, intelligent voice, face and image recognition, the consumption scene of traditional enterprises will undergo subversive changes, and the integration with the Internet will be closer.
