Wang Jinqiao, the father of face recognition robot: the era of brushing face is coming
In the second season of "Refueling! To the Future" broadcast on CCTV's comprehensive channel last Sunday, a face recognition robot called "Little Plus" attracted the attention of the national audience. It is reported that this face recognition robot was built by the research team of Wang Jinqiao, a researcher at the State Key Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences. Wang Jinqiao, who participated in the variety show for the first time, was a bit shy and nervous, but when he talked about his "children", his face was full of smiles and pride.
When Wang Jinqiao came to power, he was joked by the host Sabinin: "You are so young." I don’t know that this year is the 13th year of Wang Jinqiao’s face recognition technology. Now Wang Jinqiao is a leader in face recognition technology! The face recognition technology developed by his team has repeatedly assisted the public security department to find out. Lost population. In addition to the important contribution in the field of “recovering lost populationâ€, he is even more aware of the future application of face recognition technology in real life: “The era of brushing is coming soonâ€.
Face recognition technology is more difficult to upgrade the face recognition challenge
In 2004, it was the first year of Wang Jinqiao's technology related to face recognition. At that time, his main research population was limited to actors in the video, and his research purpose was only to analyze and understand the content of TV dramas. Who would have thought that in the next 13 years, Wang Jinqiao sneaked into the field of face recognition technology and developed a highly accurate face recognition robot. This time, Xiao Jia staged a unique first show in "Refueling! To the Future", facing 18 male dogs with the same size and uniform hair, which is almost indistinguishable by human eyes. Xiao Jia did not hesitate to say the correct answer. According to Xiao Jia’s “father†Wang Jinqiao, in the comparison of ID cards and on-site photos, the accuracy of Xiaojia’s face verification is over 99% under one millionth of the false alarm rate. Xiaojia's recognition algorithm is very prominent in the dynamic monitoring of face recognition in the middle and long distance. Under one hundred thousandth of the false alarm rate, the accuracy of face recognition is 85%. And this impressive dog face recognition display is actually more difficult than face recognition.
In fact, the characteristics of the face are more abundant than that of the dog's face. For example, according to the distinctive features such as scar, skin color, sputum, hair style, etc., we can remember and distinguish different faces, and these features are basically not reflected on the dog's face. Come out, let alone dogs of the same breed and body size. Faced with such a challenge, the Wang Jinqiao team used a self-designed deep-coupled neural network algorithm to overcome this difficulty. This algorithm can tap into the rich local and global features of the dog face, highlighting the fineness that can effectively distinguish different dogs. Differentiated differences, while ignoring the common features between dogs of the same breed, thus achieving efficient dog face recognition. Compared with the dog face recognition technology, the face recognition technology of the Wang Jinqiao team also uses this deep learning-based low-power deep-coupled neural network algorithm, which can capture the global structural features and local nuances of the face at the same time. Thereby learning a person's personalized face features. The difference is that the current market algorithms are often difficult to overcome the impact of attitude and illumination changes. In order to solve this historical problem, the Wang Jinqiao team proposed a three-dimensional deformation neural network by introducing a three-dimensional face model. It can realize face dynamic recognition with high precision in medium and long distance.
Face recognition technology helps the lost relatives to reunite the social welfare attributes
In Wang Jinqiao's view, face recognition technology is difficult to challenge, but it also has great application prospects. Face recognition is a non-contact biometric technology that does not require users to cooperate with it. It has real life. Very wide range of applications. One of the most eye-catching is the contribution of face recognition technology in the field of social welfare.
On the stage of "Come on! To the Future" on Sunday, Wang Jinqiao told such a story. In 2016, a man in Jixian County, Henan Province, met a person who was very similar to himself. He immediately remembered the long-lost twin brother, so he specifically sought help from the police. After receiving the news, the police immediately took the face data image of the party and then helped the party successfully retrieve the brother who had been separated for forty years through Xiaojia’s face recognition technology. Not only that, Xiaojia's face recognition technology also helped a lost old lady in Chenzhou, Henan Province to go home smoothly. Under the condition that the old lady can't remember the personal information, the police collected the old lady's face area image and relied on Xiaojia's face recognition technology to successfully retrieve the old lady's identity information within 1 second. Family.
Loss of loved ones often brings unbearable pain and suffering to a family, and Xiaojia’s face recognition technology gives them hope. Wang Jinqiao also mentioned in the program that in the future, Xiaojia’s face recognition technology will continue to play a major role in finding lost children and the elderly, helping more lost relatives to reunite.
Accurate recognition of face brushing face in any scene is coming soon
In addition to outstanding achievements in the social welfare field such as “recovering lost relativesâ€, Wang Jinqiao also revealed that face recognition technology has been widely used in the fields of ticket verification, banking processing, online payment, and hotel occupancy.
When mentioning the broad prospects of face recognition technology in the future, Wang Jinqiao feels that although the face recognition technology has achieved good results, there are still some difficulties that need to be broken. Among them, the face recognition technology mainly needs to be solved from three aspects: how to identify under harsh conditions, how to apply the fast recognition speed of the mobile application, and how to identify the training data analogy under extremely unbalanced conditions. When the above difficulties are overcome, the future face recognition technology will achieve high-precision recognition of faces in arbitrary poses of arbitrary scenes, and realize medium-distance dynamic face recognition.
As face recognition technology becomes more mature, technologies such as face unlocking and face payment will also appear in our lives. Alipay has previously proposed using "face payment" to solve problems such as lack of power and no network. . Wang Jinqiao is even more imaginative. In the near future, there will be a supermarket for unmanned sales due to the breakthrough of face recognition technology. Customers can enter the face and freely purchase goods. When they check out, they don’t have to face the front of the checkout counter. The team can automatically deduct the face when they walk out of the supermarket. For the future of face recognition technology, Wang Jinqiao feels: "The era of brushing the face is coming."
Car Super Tweeter,Audio Car Tweeter,Car Tweeter Speakers,Audio Multimedia Driver
NINGBO BOILINGSOUND ELECTRONICS CO.,LTD , https://www.tweeterspeaker.com