Shulou(Shulou.com)06/02 Report--

Http://www.sohu.com/a/239112487_733114

The wave of artificial intelligence is surging, and face recognition companies are developing rapidly. Relevant institutions predict that by 2020, the market size of face recognition is expected to reach 200 billion, of which customs clearance security products will reach 70 billion and online payment will reach 50 billion. This will be an industry that is likely to produce new Alibaba, Tencent or Baidu heavyweight companies.

What is face recognition?

Face recognition is a biometric recognition technology based on human facial feature information. Use cameras or cameras to collect images or videos containing human faces, automatically detect image information and track human faces, and carry out a series of face-related analysis techniques for the detected faces.

The composition of face recognition system:

Face recognition algorithm mainly consists of three modules:

Face detection (Face Detection)

Face alignment (Face Alignment)

Facial feature representation (Feature Representation)

As shown in the following figure:

▌ face detection

Face detection is used to determine the size and position of the face in the image, that is, to solve the problem of "where is the face" and cut out the real face region from the image, which is convenient for subsequent face feature analysis and recognition. The following picture shows the results of face detection in an image:

Taken in 2014, it was the largest selfie in the world at that time.

(the detection method is Baidu's PyramidBox)

▌ face alignment

The same person may show different postures and expressions in different image sequences, which is not conducive to face recognition. Therefore, it is necessary to transform the face images to a unified angle and posture, which is face alignment. Its principle is to find several key points of the face (such as the corner of the eye, the tip of the nose, the corner of the mouth, etc.), and then use these key points to transform the face to the standard face as much as possible through similar transformation (Similarity Transform, rotation, scaling and translation). The following image is a typical face image alignment process:

The basic flow of ▌ face recognition:

Classification factions of ▌ face recognition algorithm

▌ face recognition corresponds to the solution direction:

Why is face recognition so popular in China?

On the one hand, in the artificial intelligence competition, face recognition is a more mature application field.

On the other hand, face recognition is in line with the trend of national policy and benefits people's livelihood. The National 863 Program, the National Science and Technology support Program and the Natural Science Foundation have all allocated special funds to support face recognition related research. With the support and improvement of national policy, face recognition technology will be pushed to a broader daily field.

Attached: list of top 20 companies for face recognition

▌ Chinese face recognition Unicorn

When a typhoon blows in an emerging segment, it will always attract a large number of entrepreneurs and hot money to invest in it, but in the end, after several rounds of knockout competitions and competitions, there will be only a few oligarchs or even a super giant in the market. from the hundred regiments, taxi-hailing software wars, to shared bikes, to artificial intelligence and more detailed face recognition, will be such a process of survival of the fittest.

At present, among the rivers and lakes of face recognition, Shangtang, Xiangshi, Yuncong and Yitu are called the four unicorns of face recognition by Li Kaifu. Hot money and financing will burn for another year or two, and face recognition companies other than these four may be at the end of their life cycle.

Source: pulse data Research Institute, SigAI, Zhihu MaxSam

Comprehensive arrangement and editing by robot civilization

-end-

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.