How does facial recognition technology determine facial attributes?

In recent years, smart phone equipped with facial recognition has become the new product development direction of various major manufacturers. Apple has always been the smart phone product specifications and technical indicators. Since 2009, it has been actively researching and developing. In 2010, it began to acquire facial recognition related companies to 2017. Years ago, a total of seven companies related to the development of facial recognition technology were acquired, demonstrating their enthusiasm for the development of facial recognition technology.

The process of Facial Recognition includes: capturing and inputting an image through the image capturing device; performing face detection, identifying an image area depicting the face to determine whether the face exists in the image; confirming the presence of the face Then, the features of the face are further extracted and compared with the known faces in the database, thereby confirming or recognizing the personal identity in the input image. In general, good image processing is performed on the face detection segment, which reduces the amount of computation of subsequent recognition actions and improves facial feature recognition. The quality of facial feature extraction greatly determines the effect of subsequent recognition.

From the application point of view, the face recognition application that confirms or recognizes the identity of another person is called "identity authentication"; in addition, facial recognition technology can also be used to determine various attributes of the face, such as status, expression, and even Emotional discrimination, this is called "attribute analysis."

Face recognition technology needs to process a large number of images, and the amount of calculation is huge, which has become an important topic for technology promotion. In recent years, computer computing performance has improved, including the progress of cloud computing, and improving computing problems. Through cloud computing, it is also possible to perform security monitoring and other applications that need to process a large amount of image information; on smart handheld devices such as smart phones, it is more feasible to perform face recognition operations such as photo classification and personal identity authentication.

Apple used to apply facial recognition technology to character recognition in images, including:

1. In 2009, iPhoto introduced the photo classification function, iPhoto will recognize the face in the photo, and then classify the face that is recognized as the same face.

Second, the face recognition camera function launched in the iPhone 4S in 2011 mainly supports facial recognition focusing, and the face detection can reach up to 10 faces.

Third, in 2013, iOS 7 supports face (head) motion control, which can be set to select the selected operation options when the iPhone detects the right or left head shaking, such as clicking and returning to the main screen.

Fourth, iOS 10, launched in 2016, began to support photo sorting on the local end of smart handheld devices. 5. In 2017, the new face recognition application "Clips App" was launched. The face recognition technology was used to compare the photos in the Photos App to confirm the characters in the film, and the user was suggested to transfer the files to the people appearing in the film.

After 2009, Apple began investing more R&D energy in facial recognition technology development. Most of the earlier patents were acquired through acquisition. The technology developed independently is mainly in the face detection (including image pre-processing) segment, and the technology for identification is not deep. In order to accelerate the commercialization of facial recognition technology, in 2010, the M&A strategy was adopted to improve the layout. The M&A technology covers all the facial recognition technology gaps. By 2017, the layout of facial recognition technology has been quite complete.

First, Polar Rose and RealFace's technology complements Apple's gap in identification technology and enhances the most important precision and immediacy of facial recognition technology. Polar Rose's 3D facial recognition technology, PrimeSense's 3D image capture technology, and LinX's high-precision miniature image sensor, Apple is fully capable of developing 3D facial recognition technology. Perceptio and Turi artificial intelligence basic technology will help Apple develop higher-precision facial recognition technology, while Perceptio enables smart handheld devices to run artificial intelligence systems. It is expected to help RealFace develop facial recognition technology based on artificial intelligence. Apple smart handheld device

Apple's attempt is to develop a 3D facial recognition application based on artificial intelligence that can be mounted on smart handheld devices. In the future, in product development, facial recognition technology will be used to enhance the multimedia application function of smart handheld devices, thereby serializing the applications in the handheld device, simplifying the operation process and making the smart handheld device more intelligent. At the same time, applications such as controlling the use of devices or content through face recognition technology are expected to appear soon in Apple smartphones. The use of expressions or sentiment analysis to enable devices and users to generate more "emotional communication" applications will be one of the next development priorities.