Face recognition technology has been extremely hot in recent years.
How is the development of its industry? Let's unveil its mystery.
1. There is a huge gap between the practical application effect and the experimental effect.
Nowadays, the application of face recognition technology in finance, security and other fields is much worse than in the laboratory. A University introduced face recognition to read and punch cards in the morning. Because of the slow reaction speed, there is a long queue at noon. Visible face recognition technology in practical application, due to various physical factors (illumination, angle, focusing, distance of mermaid camera, etc.), the quality of captured pictures is poor, pictures are transmitted through the network to the LAN for comparison, matching recognition (this process is too slow), making the actual effect. A big discount. In most cases, the actual image quality is much lower than the training image quality.
2. Improving the Standard of Experimental Testing
In the experimental test, the image quality is usually high and is a front face image. In practical application, the collected image may be a local face and the image quality is not high. So in most cases, the practical application standard will be much higher than the training standard.
3. Training Effect and Realistic Effect
In most cases, the actual effect will be much lower than the training effect. Now CV companies on the market say that their training effect is more than 99% (infinitely close to 100%), but this does not mean that the actual application effect is 99%. Face recognition applications with complex industrial scenarios (such as blacklist recognition, which is a 1:N face comparison) have a correct rate of more than 90%, which is already a good algorithm model.
Thoughts on the Future Development Trend of Face Recognition Technology
With the development of artificial intelligence industry, under the background of information, cloud computing and big data, the application of biometrics technology will become more and more popular, which is represented by face recognition. The following trends of development are presented:
1. Networking Trend
Face recognition solves a basic problem of identity recognition in daily life. In the future, the result of this kind of identity authentication will be more and more combined with applications in all walks of life, and information can be shared through the Internet and the Internet of Things. Simply speaking, the development trend of "Identity Recognition + Internet of Things" will be very common in the future.
2. Trend of multi-biometric pattern fusion
Face recognition technology can not reach the expected experience of human beings. For some special applications with high security requirements, such as financial industry, face recognition is easy to be exploited by illegal elements for identity fraud. Therefore, it needs a combination of various biometric recognition technologies (such as biometric detection, iris recognition). To further improve the overall security of identity recognition.
3. Cloud Technology
In the future, cloud technology will also provide data and computational support for the application of face recognition. Access control based on cloud technology can manage hundreds of channels at the same time. With the popularization of the Internet of Things, users can control and manage access control remotely anywhere and identify themselves accurately, which will be widely used in enterprises and schools. Access control solutions for schools, training institutions, large-scale business occasions and office buildings.
Security Risk of Face Recognition Technology
Facial recognition and iris recognition have different degrees of reproducibility. Faces are exposed every day. A person's facial features can be obtained and reproduced by taking photos. Another risk is instability. Heavy makeup, allergies, injuries and cosmetic surgery on face paintings will lead to great changes in facial features, affecting the accuracy of face recognition and even unable to recognize.
As a new technology field, face recognition will face many malicious attacks from illegal elements, just like the technology used to serve in the security field. It needs to improve its technical level and protection ability in the constant offensive and defensive warfare. Like killing computer viruses, attack and defense will be a long-term game process, which will promote the development of technology in a better direction.
You may also be interested in the following information