WO2019100608A1

WO2019100608A1 - Video capturing device, face recognition method, system, and computer-readable storage medium

Info

Publication number: WO2019100608A1
Application number: PCT/CN2018/076140
Authority: WO
Inventors: 陈林
Original assignee: 平安科技（深圳）有限公司
Priority date: 2017-11-21
Filing date: 2018-02-10
Publication date: 2019-05-31
Also published as: CN108038422B; CN108038422A

Abstract

The present application relates to a video capturing device, a face recognition method, a system, and a computer-readable storage medium, the face recognition method comprising: carrying out face detection on video data to obtain a face image; tracking the face image, and acquiring a sequence of face images; performing image quality evaluation on the sequence of face images, and acquiring a preset number of top-ranking face images; carrying out feature point positioning on the preset number of top-ranking face images, and carrying out calibration on the basis of the positioned face images; inputting the calibrated face images into a deep neural network model which is generated by training in advance, and acquiring an outputted face feature vector; and sending the face feature vector to a server so as to carry out a comparison operation with face images in samples in a face image sample library. According to the present application, the calculation pressure of the server during face recognition may be reduced, and the network transmission pressure may be decreased.

Description

Camera device, method and system for face recognition, and computer readable storage medium

Priority claim

The present application is based on the priority of the Chinese Patent Application entitled "Camera device, face recognition method and computer readable storage medium", filed on November 21, 2017, with the application number CN 201711166813.8, which is filed on November 21, 2017. The overall content is incorporated herein by reference.

Technical field

The present application relates to the field of image processing technologies, and in particular, to a camera device, a method and system for recognizing a face, and a computer readable storage medium.

Background technique

At present, the existing 1:N dynamic face recognition system generally uses one server to connect one or more network cameras, and the server collects video data from the camera through the network, and performs face recognition on the video data, such a centralized analysis scheme The computing pressure of the server is very large, especially when the number of cameras is large, usually one server can not meet the demand, the server array needs to be built, and there are high requirements in terms of power consumption and heat dissipation; in addition, since the video data needs to be transmitted from the camera To the server, the pressure on the network is also relatively large, and this pressure will rise as the resolution and quality of the camera increase.

Summary of the invention

The purpose of the present application is to provide an image capturing apparatus, a method and system for recognizing a face, and a computer readable storage medium, which aim to alleviate the calculation pressure of the server in face recognition and reduce the network transmission pressure.

To achieve the above object, the present application provides an image pickup apparatus including a memory and a processor connected to the memory, wherein the memory stores a processing system operable on the processor, the processing The system implements the following steps when executed by the processor:

Detection step: performing face detection on the video data to obtain a face image;

Tracking step: tracking the face image to obtain a sequence of face images;

Image quality scoring step: performing image quality scoring on the sequence of face images, and obtaining a preset number of face images with the highest score;

Feature point positioning step: performing feature point positioning on a preset number of face images with a higher score, and correcting based on the positioned face image;

Feature vector output step: inputting the corrected face image into a depth neural network model generated by pre-training, and acquiring the output face feature vector;

Transmission step: transmitting the face feature vector to the server to perform a step of performing a comparison operation with the face image in the sample in the face image sample library.

To achieve the above object, the present application further provides a method for face recognition, and the method for face recognition includes:

S1, performing face detection on the video data to obtain a face image;

S2, tracking the face image to obtain a sequence of face images;

S3, performing image quality scoring on the sequence of face images, and obtaining a preset number of face images with a higher score;

S4, performing feature point positioning on the preset number of face images with the highest score, and correcting the image based on the positioned face image;

S5, the corrected face image is input into a depth neural network model generated by pre-training, and the output face feature vector is obtained;

S6: Send the face feature vector to the server to perform a step of performing a comparison operation with the face image in the sample in the face image sample library.

To achieve the above object, the present application further provides a system for face recognition, the system for face recognition comprising:

a detecting module, configured to perform face detection on the video data to obtain a face image;

a tracking module for tracking a face image to obtain a sequence of face images;

a scoring module, configured to perform image quality scoring on the sequence of face images, and obtain a preset number of face images with a higher score;

a correction module, configured to perform feature point positioning on a preset number of face images that are scored first, and perform correction based on the positioned face image;

An input module, configured to input the corrected face image into a depth neural network model generated by pre-training, and obtain an output face feature vector;

The sending module is configured to send the face feature vector to the server to trigger an operation of comparing the face image in the sample in the face image sample library.

The application further provides a computer readable storage medium having a processing system stored thereon, the processing system being implemented by a processor to implement the steps:

Tracking step: tracking the face image to obtain a sequence of face images;

The beneficial effects of the present application are: the processing of one video data per camera device of the present application, in addition to collecting video, the camera device can perform face detection, tracking, image quality scoring, feature point localization and input depth neural network model on the video. In the middle, the face feature vector is obtained, and finally only the face feature vector is transmitted to the server. When the number of camera devices is large, the calculation pressure of the server can be greatly reduced, and the server array does not need to be built, and the network transmission can be reduced to a large extent. Pressure, and network transmission pressure does not rise with the resolution and image quality of the camera.

DRAWINGS

1 is a schematic diagram of an optional application environment of each embodiment of the present application;

2 is a schematic diagram of a hardware architecture of an embodiment of the camera device of FIG. 1;

FIG. 3 is a schematic flowchart diagram of an embodiment of a method for recognizing a face of an applicant.

Detailed ways

In order to make the objects, technical solutions, and advantages of the present application more comprehensible, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the application and are not intended to be limiting. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without departing from the inventive scope are the scope of the present application.

It should be noted that the descriptions of "first", "second" and the like in the present application are for the purpose of description only, and are not to be construed as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. . Thus, features defining "first" or "second" may include at least one of the features, either explicitly or implicitly. In addition, the technical solutions between the various embodiments may be combined with each other, but must be based on the realization of those skilled in the art, and when the combination of the technical solutions is contradictory or impossible to implement, it should be considered that the combination of the technical solutions does not exist. Nor is it within the scope of protection required by this application.

Referring to FIG. 1 , it is a schematic diagram of an application environment of a preferred embodiment of the method for face recognition of the present applicant. The application environment diagram includes a camera device 1 and a server 2. The plurality of imaging apparatuses 1 can perform data interaction with the server 2 through suitable technologies such as a network and a near field communication technology, respectively.

The server 2 may be a single network server, a server group composed of a plurality of network servers, or a cloud-based cloud composed of a large number of hosts or network servers, wherein the cloud computing is a kind of distributed computing, and is a group of loosely coupled computers. A set of super virtual computers.

The camera device 1 is a common electronic product that includes a camera and can dynamically acquire images, and can automatically perform numerical calculation and/or information processing according to preset or stored instructions.

Referring to FIG. 2, in the embodiment, the image capturing apparatus 1 may include, but is not limited to, a memory 11, a processor 12, a network interface 13, and a camera 14 that are communicably connected to each other through a system bus. The memory 11 is stored and processed. The processing system running on the device 12. It is to be noted that FIG. 2 only shows the camera device 1 having the components 11-14, but it should be understood that not all illustrated components are required to be implemented, and more or fewer components may be implemented instead.

Each camera device 1 includes a processor (the processor is an nvidia tx2 chip for processing images), and the nvidia tx2 chip can be connected to the camera device 1 via a usb or csi or a network interface to run the processing system. The imaging device 1 and the server 2 are connected by a network, and the server 2 stores a database of human face image samples. The camera device 1 is installed in a specific place (for example, an office place and a monitoring area), and a video is captured in real time for a target entering the specific place, and the processor processes the video to obtain a face feature vector, and then only sends the face feature vector through the network. To the server 2, the server 2 performs comparison based on the face image sample library to implement face recognition.

The memory 11 includes a memory and at least one type of readable storage medium. The memory provides a cache for the operation of the camera device 1; the readable storage medium may be, for example, a flash memory, a hard disk, a multimedia card, a card type memory (eg, SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM). A non-volatile storage medium such as a read only memory (ROM), an electrically erasable programmable read only memory (EEPROM), a programmable read only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, or the like. In some embodiments, the readable storage medium may be an internal storage unit of the camera 1, such as a hard disk of the camera 1; in other embodiments, the non-volatile storage medium may also be external to the camera 1. The storage device is, for example, a plug-in hard disk provided on the camera 1, a smart memory card (SMC), a Secure Digital (SD) card, a flash card, or the like. In this embodiment, the readable storage medium of the memory 11 is generally used to store an operating system installed in the image pickup apparatus 1 and various types of application software, such as program codes of the processing system in an embodiment of the present application. Further, the memory 11 can also be used to temporarily store various types of data that have been output or are to be output.

In this embodiment, the processor 12 is configured to run program code or process data stored in the memory 11, such as running a processing system or the like.

The network interface 13 may comprise a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the camera device 1 and other electronic devices. In this embodiment, the network interface 13 is mainly used to connect the camera device 1 to the server 2, and establish a data transmission channel and a communication connection between the camera device 1 and the server 2.

The processing system is stored in the memory 11 and includes at least one computer readable instruction stored in the memory 11, the at least one computer readable instruction being executable by the processor 12 to implement the methods of various embodiments of the present application; The at least one computer readable instruction can be classified into different logic modules depending on the functions implemented by its various parts.

In an embodiment, when the processing system is executed by the processor 12, the following steps are implemented:

Face detection is performed on each frame image in the video data based on the features of the face, and there may be one or more faces in each frame image, or no face, and after face detection, the image may be extracted from the image. Face image.

The face image is an image including only a face region (no other background), and the face region can be large or small. For a face image taken in a distant scene, the face region is small, and for a close-up shot of a face image, The face area is large. The face area is a minimum area including a human face, and is preferably a rectangular area including a human face. Of course, it may be an area including a human face of other shapes, such as a circular area, and is not limited thereto.

Tracking step: tracking the face image to obtain a sequence of face images;

In this embodiment, when face tracking is performed, the similarity of the adjacent two frames of the face image can be calculated to implement face tracking. In an embodiment, the similarity of the face may be calculated according to the X and Y coordinate values of the center point of the face region in the adjacent two frames of the face image; in other embodiments, the face image of the adjacent two frames may be used. The X, Y coordinate values of the center point of the face region in the face region, and the height H and the width W value of the face region are calculated, and the similarity of the faces in the adjacent two frames of the face image is calculated. The face tracking is performed based on the similarity of the faces in the adjacent two frames of the face image, and a sequence of face images of the same person is obtained, and two or more persons appearing in the face image may also be respectively obtained. A sequence of face images corresponding to each person.

Wherein, when the image quality of each sequence of face images is scored, the quality of each face image in the series is scored according to the gradient values and coordinate values of predetermined points in the face image.

Wherein, the predetermined points include an eye point, a nose point and a mouth point, the gradient value of the predetermined point is a mean gradient, and the average gradient refers to a boundary of a predetermined point of the face image or a gray near the sides of the shadow line. There is a significant difference in the degree of gray scale change, which can be used to indicate the sharpness of the image, reflecting the rate of change of the minute detail contrast of the predetermined point, that is, the rate of density change in the multi-dimensional direction of the predetermined point, and the characterization The relative clarity of the face image. The coordinate values of the predetermined points include at least the x-axis of the eye point and the nose point.

In the process of scoring the quality of each face image in the series, in the captured face image, the greater the distance between the eyes, the closer the x-axis of the center point of the eyes to the tip of the nose, and the average gradient value is Large, the higher the score of the image, the greater the probability that the face image is a positive face image.

In this embodiment, for each sequence of face images, in order to conveniently select a face image of a positive face, the scoring results are arranged in descending order, that is, the face image is arranged in front of the face image, and the sequence is selected from the arranged sequence. The preset number of face images of the top score, for example, 7 face images are selected.

For feature points positioning on a preset number of face images, the feature points include at least an eye feature point, a mouth feature point, and a nose feature point, and are corrected based on the face image after the feature point is positioned.

In this embodiment, the corrected face image is input into a depth neural network model generated by pre-training, and is calculated by a deep neural network model, and then the face feature vector of each face image is output, and then the camera device only The face feature vector is transmitted to the server for 1:N dynamic recognition.

Compared with the prior art, each camera device in the embodiment processes one channel of video data, and the camera device can perform face detection, tracking, image quality scoring, feature point location, and input depth neural network model in addition to video capture. In the middle, the face feature vector is obtained, and finally only the face feature vector is transmitted to the server. When the number of camera devices is large, the calculation pressure of the server can be greatly reduced, and the server array does not need to be built, and the network transmission can be reduced to a large extent. Pressure, and network transmission pressure does not rise with the resolution and image quality of the camera.

In a preferred embodiment, on the basis of the foregoing embodiment of FIG. 2, when the processing system is executed by the processor 12, before the face detection, the method further includes:

Analyzing whether the video data is compressed video data or uncompressed video data;

If it is uncompressed video data, the format of the video data is converted into a format capable of face detection;

In the case of compressed video data, the video data is decoded and the format of the video data is converted into a format capable of face detection.

In this embodiment, the camera device may compress the video data after the video data is collected. The compressed video data may be non-real-time compressed or compressed in real time according to the real-time performance. This embodiment is preferably real-time compression. In addition, the captured video data may be subjected to lossy compression according to actual conditions, and the compression ratio is a predetermined ratio, preferably 5:1. Video compression algorithms include M-JPEG (Motion-Join Photographic Experts Group), Mpeg (Moving Pictures Experts Group), H.264, Wavelet (Wavelet Compression), JPEG 2000, AVS compression, etc., through the above compression algorithm to obtain compressed data. Before the face detection, it can be analyzed whether the video data is compressed. Specifically, it can be analyzed whether the format is a compressed format, and if it is further processed, for example, the camera is compressed by M-JPEG, and the format is YCrCB. , you need to convert the video data in YCrCB format to RGB format so that face detection can be performed.

In a preferred embodiment, based on the foregoing embodiment of FIG. 2, the tracking step specifically includes:

Obtaining the X and Y coordinate values of the center point of the face region in the adjacent two frames of the face image, and the height H and the width W value of the face region, according to the X and Y coordinate values of the adjacent two frames of the face image, and The height H and the width W value of the person are calculated, and the similarity of the face in the adjacent two frames of the face image is calculated;

Face tracking is performed based on the similarity of faces in adjacent two frames of face images.

The similarity calculation step includes:

The S _i,j is a similarity, and the w _x , w _y , w _w , w _h are the x-direction distance, the y-direction distance, the width difference, and the height difference of the adjacent two frames of the face i and the face j, respectively. Weight, w _x , w _y , w _w , w _h ∈[0,1], where:

The distance between the face i and the center point of the face j in the x direction;

The distance between the face i and the center point of the face j in the y direction;

The difference in width between face i and face j;

The difference in height between face i and face j.

When the similarity of the face in the adjacent two frames of the face image is greater than or equal to a preset threshold, the face in the adjacent two frames of the face image is determined to be the same person's face.

In a preferred embodiment, based on the foregoing embodiment of FIG. 2, the image quality scoring step specifically includes: each of the series according to a gradient value and a coordinate value of a predetermined point in the face image. The quality of the face image is scored.

The predetermined point includes an eye point, a nose point, and a mouth point, and the gradient value is an average gradient of an eye point, a nose point, and a mouth point, and the eye point includes a left eye point and The right eye point, the mouth point includes a left mouth corner point and a right mouth corner point, and the image quality score is calculated as:

p=((x_LeftEye-x_RightEye)^2×grad)/|(x_LeftEye+x_RightEye)/2-x_Nose|;

Where p is the image quality score, x_LeftEye, x_RightEye represents the X coordinate of the left and right eyeballs, x_Nose represents the X coordinate of the tip of the nose, and grad is the average gradient of the eye point, nose point and mouth point.

In this embodiment, the coordinates of the eye point, the nose point, and the mouth point in the face are selected to score the quality of the face image, and the face image can be objectively and accurately evaluated to obtain a person with a high score. The face image is convenient for subsequent correction and other processing.

As shown in FIG. 3, FIG. 3 is a schematic flowchart of an embodiment of a method for recognizing a face of an applicant, where the method includes the following steps:

Step S1, performing face detection on the video data to obtain a face image;

Step S2, tracking the face image to obtain a sequence of face images;

Step S3, performing image quality scoring on the sequence of face images, and obtaining a preset number of face images with a higher score;

Step S4: performing feature point positioning on the preset number of face images that are scored first, and correcting based on the positioned face image;

Step S5, the corrected face image is input into the depth neural network model generated by the pre-training, and the output face feature vector is obtained;

In step S6, the face feature vector is sent to the server to perform a step of performing a comparison operation with the face image in the sample in the face image sample library.

In a preferred embodiment, on the basis of the foregoing embodiment of FIG. 3, before the step S1, the method further includes:

In this embodiment, the camera device may compress the video data after the video data is collected. The compressed video data may be non-real-time compressed or compressed in real time according to the real-time performance. This embodiment is preferably real-time compression. In addition, the captured video data may be subjected to lossy compression according to actual conditions, and the compression ratio is a predetermined ratio, preferably 5:1. Video compression algorithms include M-JPEG (Motion-Join Photographic Experts Group), Mpeg (Moving Pictures Experts Group), H.264, Wavelet (Wavelet Compression), JPEG 2000, AVS compression, etc., obtains compressed video data through the above compression algorithm. Before the face detection, it can be analyzed whether the video data is compressed. Specifically, it can be analyzed whether the format is a compressed format, and if it is further processed, for example, the camera is compressed by M-JPEG, and the format is YCrCB. , you need to convert the video data in YCrCB format to RGB format so that face detection can be performed.

In a preferred embodiment, based on the foregoing embodiment of FIG. 3, the step S2 specifically includes:

The similarity calculation step includes:

The difference in width between face i and face j;

The difference in height between face i and face j.

In a preferred embodiment, based on the foregoing embodiment of FIG. 3, the step S3 specifically includes:

The quality of each face image in the series is scored according to the gradient values and coordinate values of the predetermined points in the face image.

p=((x_LeftEye-x_RightEye)^2×grad)/|(x_LeftEye+x_RightEye)/2-x_Nose|;

The present application also provides a computer readable storage medium having stored thereon a processing system, the processing system being executed by a processor to implement the steps of the method of face recognition described above.

The serial numbers of the embodiments of the present application are merely for the description, and do not represent the advantages and disadvantages of the embodiments.

Through the description of the above embodiments, those skilled in the art can clearly understand that the foregoing embodiment method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be through hardware, but in many cases, the former is better. Implementation. Based on such understanding, the technical solution of the present application, which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk, The optical disc includes a number of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the methods described in various embodiments of the present application.

The above is only a preferred embodiment of the present application, and is not intended to limit the scope of the patent application, and the equivalent structure or equivalent process transformations made by the specification and the drawings of the present application, or directly or indirectly applied to other related technical fields. The same is included in the scope of patent protection of this application.

Claims

An image pickup apparatus, comprising: a memory and a processor connected to the memory, wherein the memory stores a processing system operable on the processor, wherein the processing system is The processor implements the following steps when it executes:

Detection step: performing face detection on the video data to obtain a face image;

Tracking step: tracking the face image to obtain a sequence of face images;

Image quality scoring step: performing image quality scoring on the sequence of face images, and obtaining a preset number of face images with the highest score;

Feature point positioning step: performing feature point positioning on a preset number of face images with a higher score, and correcting based on the positioned face image;

Feature vector output step: inputting the corrected face image into a depth neural network model generated by pre-training, and acquiring the output face feature vector;

Transmission step: transmitting the face feature vector to the server to perform a step of performing a comparison operation with the face image in the sample in the face image sample library.
The image pickup apparatus according to claim 1, wherein when said processing system is executed by said processor before said detecting step, the following steps are further implemented:

Analyzing whether the video data is compressed video data or uncompressed video data;

If it is uncompressed video data, the format of the video data is converted into a format capable of face detection;

In the case of compressed video data, the video data is decoded and the format of the video data is converted into a format capable of face detection.
The camera apparatus according to claim 1 or 2, wherein the tracking step specifically includes:

Obtaining the X and Y coordinate values of the center point of the face region in the adjacent two frames of the face image, and the height H and the width W value of the face region, according to the X and Y coordinate values of the adjacent two frames of the face image, and The height H and the width W value of the person are calculated, and the similarity of the face in the adjacent two frames of the face image is calculated;

Face tracking is performed based on the similarity of faces in adjacent two frames of face images.
The image capturing device according to claim 3, wherein the image quality scoring step comprises:

The quality of each face image in the series is scored according to the gradient values and coordinate values of the predetermined points in the face image.
The image pickup apparatus according to claim 4, wherein said predetermined point includes an eye point, a nose point, and a mouth point, and said gradient value is an average of an eye point, a nose point, and a mouth point. a gradient, the eye point includes a left eye point and a right eye point, the mouth point includes a left mouth corner point and a right mouth corner point, and the image quality scoring step further includes:

p=((x_LeftEye-x_RightEye)^2×grad)/|(x_LeftEye+x_RightEye)/2-x_Nose|;

Where p is the image quality score, x_LeftEye, x_RightEye represents the X coordinate of the left and right eyeballs, x_Nose represents the X coordinate of the tip of the nose, and grad is the average gradient of the eye point, nose point and mouth point.
A method for face recognition, characterized in that the method for face recognition comprises:

S1, performing face detection on the video data to obtain a face image;

S2, tracking the face image to obtain a sequence of face images;

S3, performing image quality scoring on the sequence of face images, and obtaining a preset number of face images with a higher score;

S4, performing feature point positioning on the preset number of face images with the highest score, and correcting the image based on the positioned face image;

S5, the corrected face image is input into a depth neural network model generated by pre-training, and the output face feature vector is obtained;

S6: Send the face feature vector to the server to perform a step of performing a comparison operation with the face image in the sample in the face image sample library.
The method for the face recognition according to claim 6, wherein before the step S1, the method further comprises:

Analyzing whether the video data is compressed video data or uncompressed video data;

If it is uncompressed video data, the format of the video data is converted into a format capable of face detection;

In the case of compressed video data, the video data is decoded and the format of the video data is converted into a format capable of face detection.
The method of the face recognition according to claim 6 or 7, wherein the step S2 comprises:

Obtaining the X and Y coordinate values of the center point of the face region in the adjacent two frames of the face image, and the height H and the width W value of the face region, according to the X and Y coordinate values of the adjacent two frames of the face image, and The height H and the width W value of the person are calculated, and the similarity of the face in the adjacent two frames of the face image is calculated;

Face tracking is performed based on the similarity of faces in adjacent two frames of face images.
The method for the face recognition according to claim 8, wherein the step S3 comprises:

The quality of each face image in the series is scored according to the gradient values and coordinate values of the predetermined points in the face image.
The method according to claim 9, wherein the predetermined point comprises an eye point, a nose point, and a mouth point, and the gradient value is an eye point, a nose point, and a mouth part. An average gradient of the points, the eye points include a left eye point and a right eye point, the mouth point including a left corner point and a right corner corner point, and the step S3 further includes:

p=((x_LeftEye-x_RightEye)^2×grad)/|(x_LeftEye+x_RightEye)/2-x_Nose|;

Where p is the image quality score, x_LeftEye, x_RightEye represents the X coordinate of the left and right eyeballs, x_Nose represents the X coordinate of the tip of the nose, and grad is the average gradient of the eye point, nose point and mouth point.
A system for face recognition, characterized in that the system for face recognition comprises:

a detecting module, configured to perform face detection on the video data to obtain a face image;

a tracking module for tracking a face image to obtain a sequence of face images;

a scoring module, configured to perform image quality scoring on the sequence of face images, and obtain a preset number of face images with a higher score;

a correction module, configured to perform feature point positioning on a preset number of face images that are scored first, and perform correction based on the positioned face image;

An input module, configured to input the corrected face image into a depth neural network model generated by pre-training, and obtain an output face feature vector;

The sending module is configured to send the face feature vector to the server to trigger an operation of comparing the face image in the sample in the face image sample library.
The system for face recognition according to claim 11, further comprising:

An analysis module, configured to analyze whether the video data is compressed video data or uncompressed video data;

a first conversion module, configured to convert the format of the video data into a format capable of face detection if the video data is uncompressed;

The second conversion module is configured to decode the video data and convert the format of the video data into a format capable of face detection if the compressed video data is used.
The system for recognizing a face according to claim 11 or 12, wherein the tracking module is specifically configured to acquire X and Y coordinate values of a center point of a face region in adjacent two frames of face images, and The height H and the width W of the face region are calculated according to the X and Y coordinate values of the adjacent two frames of the face image, and the human height H and the width W value, and the similarity of the faces in the adjacent two frames of the face image is calculated. Degree; face tracking based on the similarity of faces in two adjacent frames of face images.
The system for recognizing a face according to claim 13, wherein the scoring module is specifically configured to: according to a gradient value and a coordinate value of a predetermined point in the face image, each face image in the series Quality is scored.
The system for face recognition according to claim 14, wherein said predetermined points include eye points, nose points, and mouth points, and said gradient values are eye points, nose points, and mouth portions. An average gradient of points, the eye points including a left eye point and a right eye point, the mouth point including a left mouth corner point and a right mouth corner point, and the tracking module is further configured to:

p=((x_LeftEye-x_RightEye)^2×grad)/|(x_LeftEye+x_RightEye)/2-x_Nose|;

Where p is the image quality score, x_LeftEye, x_RightEye represents the X coordinate of the left and right eyeballs, x_Nose represents the X coordinate of the tip of the nose, and grad is the average gradient of the eye point, nose point and mouth point.
A computer readable storage medium, wherein the computer readable storage medium stores a processing system, and when the processing system is executed by the processor, the following steps are implemented:

Detection step: performing face detection on the video data to obtain a face image;

Tracking step: tracking the face image to obtain a sequence of face images;

Image quality scoring step: performing image quality scoring on the sequence of face images, and obtaining a preset number of face images with the highest score;

Feature point positioning step: performing feature point positioning on a preset number of face images with a higher score, and correcting based on the positioned face image;

Feature vector output step: inputting the corrected face image into a depth neural network model generated by pre-training, and acquiring the output face feature vector;

Transmission step: transmitting the face feature vector to the server to perform a step of performing a comparison operation with the face image in the sample in the face image sample library.
The computer readable storage medium according to claim 16, wherein, before said detecting step, said processing system is executed by said processor, further implementing the following steps:

Analyzing whether the video data is compressed video data or uncompressed video data;

In the case of uncompressed video data, the format of the video data is converted into a format capable of face detection;

In the case of compressed video data, the video data is decoded and the format of the video data is converted into a format capable of face detection.
The computer readable storage medium according to claim 16 or 17, wherein the step of tracking comprises:

Obtaining the X and Y coordinate values of the center point of the face region in the adjacent two frames of the face image, and the height H and the width W value of the face region, according to the X and Y coordinate values of the adjacent two frames of the face image, and The height H and the width W value of the person are calculated, and the similarity of the face in the adjacent two frames of the face image is calculated;

Face tracking is performed based on the similarity of faces in adjacent two frames of face images.
The computer readable storage medium according to claim 18, wherein the image quality scoring step comprises:

The quality of each face image in the series is scored according to the gradient values and coordinate values of the predetermined points in the face image.
The computer readable storage medium according to claim 19, wherein said predetermined point comprises an eye point, a nose point, and a mouth point, and said gradient value is an eye point, a nose point, and a mouth An average gradient of points, the eye points include a left eye point and a right eye point, the mouth point including a left mouth corner point and a right mouth corner point, and the image quality scoring step further includes:

p=((x_LeftEye-x_RightEye)^2×grad)/|(x_LeftEye+x_RightEye)/2-x_Nose|;

Where p is the image quality score, x_LeftEye, x_RightEye represents the X coordinate of the left and right eyeballs, x_Nose represents the X coordinate of the tip of the nose, and grad is the average gradient of the eye point, nose point and mouth point.