CN116189316A - Liveness detection method and system - Google Patents

Abstract

This specification provides a living body detection method and system, the method comprising: acquiring a biometric image containing the biological characteristics of the target user, performing preliminary living body detection on the target user based on the biometric image, and obtaining the preliminary living body detection result of the target user , furthermore, instructing the target user to input the identity verification information of the target length, and based on the actual input information of the target user, perform another biopsy detection on the target user to obtain the target biopsy detection result of the target user, wherein the target length is the same as the preliminary biopsy The test results are related. The above solution can improve the accuracy of the living body detection result.

Description

Liveness detection method and system

technical field

This specification relates to the technical field of artificial intelligence, and in particular to a method and system for detecting a living body.

Background technique

Currently, biometric identification (such as face recognition) is widely used in multiple scenarios, such as payment scenarios based on face recognition, access control systems based on face recognition, and the like. Biometric identification can bring convenience to people's lives, but at the same time it also brings some security risks. Among them, live attacks are one of the security risks faced in the field of biometric identification. For example, attackers use mobile phone screens, photos, printed paper or high-precision masks to carry out live attacks.

For this reason, after the biometrics recognition system collects the biometrics image, it needs to first detect whether the target user is a living body based on the biometrics image, and then execute the subsequent biometrics recognition process when it is determined that the target user is a living body. However, in practical applications, it is found that when liveness detection is performed based on biometric images, the accuracy of detection results is not high, resulting in low security in the field of biometric recognition.

Contents of the invention

This specification provides a living body detection method and system, which can improve the accuracy of living body detection results, thereby improving the security in the field of biometric identification.

In the first aspect, this specification provides a living body detection method, including: acquiring a biological feature image, the biological feature image including the biological features of the target user; performing preliminary living body detection on the target user based on the biological feature image, and obtaining the the target user's preliminary liveness detection result; and instruct the target user to input the identity verification information of the target length, and based on the target user's actual input information, conduct another liveness test on the target user to obtain the target user's target A living body detection result, wherein the target length is related to the preliminary living body detection result.

In some embodiments, the preliminary living body detection result includes: a first probability that the target user is alive; and the target length is inversely correlated with the first probability.

In some embodiments, before instructing the target user to input identity verification information of a target length, further comprising: determining a target identity verification type; and determining the target length based on the target identity verification type and the first probability .

In some embodiments, the determining the target length based on the target identity verification type and the first probability includes: determining a weight coefficient based on the type of the current application scenario, and the weight coefficient is related to the current application scenario The degree of demand for biometric security is inversely correlated; weighting the first probability based on the weight coefficient to obtain the first probability after scene adaptation; and based on the target identity verification type and the The first probability after scene adaptation is to determine the target length.

In some embodiments, said determining said target length based on said target authentication type and said first probability comprises: inputting said target authentication type and said first probability into a pre-trained length mapping model , to obtain the target length, wherein the length mapping model is trained by using multiple sets of training samples, and each set of training samples includes: sample identity verification type, sample probability, and sample length.

In some embodiments, the determining the target identity verification type includes: randomly determining the target identity verification type among multiple candidate identity verification types; or determining the type of the current application scenario, and In the verification type, select the identity verification type that matches the type of the current application scenario as the target identity verification type; or, use the identity verification type specified by the target user among multiple candidate identity verification types as the target authentication type.

In some embodiments, the plurality of authentication types to be selected includes at least two of the following: authentication by user identification; authentication by identification of user terminal; authentication by registration password; Account ID for authentication.

In some embodiments, the performing liveness detection on the target user again based on the actual input information of the target user to obtain the target liveness detection result of the target user includes: storing the identity of the first user stored in the database The information is determined as the target identity information, wherein the first user is the login user who triggers the biometric detection, or the first user is the user obtained by performing biometric identification on the biometric image; and the actual input Information is matched with the target identity information, and based on the matching result, one of the first operation or the second operation is performed, wherein the first operation includes: determining that the matching result is a successful match, then determining that the target The user is a living body, and the second operation includes: determining that the matching result is a matching failure, then determining that the target user is not a living body.

In some embodiments, the preliminary living body detection result includes: a first probability that the target user is alive; and, performing preliminary living body detection on the target user based on the biometric image to obtain the target user The preliminary living body detection result, including: inputting the biometric feature image into a trained living body detection model, performing living body detection processing on the biometric feature image through the living body detection model, and obtaining the second probability that the target user is a living body a confidence level corresponding to the second probability; and determining the first probability based on the second probability and the confidence level corresponding to the second probability.

In some embodiments, determining the first probability based on the second probability and the confidence degree corresponding to the second probability includes: determining the product of the second probability and the confidence degree corresponding to the second probability is the first probability.

In some embodiments, the living body detection model includes: a feature extraction network, a local living body detection network, a global living body detection network, and a confidence detection network; and, the living body detection model is used to perform living body detection processing on the biometric image , obtaining the second probability that the target user is a living body and the confidence degree corresponding to the second probability includes: performing feature extraction processing on the biometric image through the feature extraction network to obtain a feature map; through the local The living body detection network performs living body detection processing on a plurality of local regions of the feature map respectively, and obtains living body detection results of the plurality of local regions; through the global living body detection network, the feature map and the multiple local regions Perform live detection processing on the living body detection results to obtain the second probability that the target user is alive; and determine the confidence level corresponding to the second probability based on the living body detection results of the multiple local areas through the confidence degree detection network .

In some embodiments, the training process of the living body detection model includes a first training phase and a second training phase, wherein the first training phase is configured to perform the training on the feature extraction network, the local living body detection network and The global living body detection network is trained, and the second training stage is configured to train the confidence detection network.

In some embodiments, the first training stage includes: acquiring a sample biometric image and annotation information corresponding to the sample biometric image, and the annotation information includes: global in vivo annotation results and in vivo annotation results in multiple local regions ; Input the sample biometric image into the feature extraction network to obtain a sample feature map, input the sample feature map into the local living body detection network to obtain the first live body detection results in multiple local areas, and use the sample feature Input the atlas and the first living body detection results of the multiple local regions into the global living body detection network to obtain the first global living body detection result; and based on the first global living body detection result, the first living body detection results of the multiple local regions The first target loss is determined from the living body detection result, the global living body labeling result and the multiple local area living body labeling results, and the feature extraction network, the local living body The detection network and the global liveness detection network are trained.

In some embodiments, the first training stage further includes: performing disturbance processing on the sample biometric image to obtain a disturbed biometric image, inputting the disturbed biometric image into the feature extraction network to obtain a disturbed feature map, inputting the disturbance feature map into the local living body detection network to obtain second living body detection results in multiple local regions; and, based on the first global living body detection results, the first living body detection results in the multiple local regions, Determining the first target loss based on the global living body labeling result and the living body labeling results of the multiple local regions includes: based on the first global living body detection result, the first living body detection results of the multiple local regions, the The first target loss is determined by the second living body detection results of multiple local regions, the global living body labeling results, and the living body labeling results of the multiple local regions.

In some embodiments, based on the first global living body detection result, the first living body detection results of the multiple local regions, the second living body detection results of the multiple local regions, the global living body labeling result and the Determining the first target loss based on the living body labeling results of the plurality of local regions includes: determining the first loss based on the difference between the first global living body detection result and the global living body labeling result; determining the second loss based on the difference between the first living body detection result of the local area and the living body labeling results of the multiple local areas; based on the first living body detection result of the multiple local areas and the determining a third loss based on the difference between the second liveness detection results; and determining the first target loss based on the first loss, the second loss, and the third loss.

In some embodiments, the second training stage includes: acquiring a sample biometric image and annotation information corresponding to the sample biometric image, and the annotation information includes: a global in vivo annotation result and a plurality of local region in vivo annotation results ; Input the sample biometric image into the feature extraction network to obtain a sample feature map, input the sample feature map into the local living body detection network to obtain the first living body detection results of multiple local areas, and the multiple local The first living body detection result of the area is input into the confidence detection network to obtain the detection confidence of the multiple local areas, and the corresponding local area of the sample feature map is updated based on the detection confidence of the multiple local areas, Obtaining an updated feature map, inputting the updated feature map into the local living body detection network to obtain third live body detection results in multiple local areas, and combining the updated feature map and the third live body detection results in the multiple local areas Input the global living body detection network to obtain a second global living body detection result; and, based on the second global living body detection result, the third living body detection results of the plurality of local regions, the global living body labeling result and the A second target loss is determined from live body labeling results of multiple local regions, and the confidence detection network is trained with the training target of minimizing the second target loss.

In some embodiments, performing liveness detection on the target user again based on the actual input information of the target user, and obtaining the target liveness detection result of the target user includes: performing liveness detection on the target user based on the actual input information performing another liveness detection to obtain a second liveness detection result; and, based on the preliminary liveness detection result and the second liveness detection result, determining the target liveness detection result.

In some embodiments, the biological feature includes at least one of the following: at least one of human face, iris, fingerprint and palm print.

In the second aspect, this specification also provides a living body detection system, including: at least one storage medium and at least one processor, the at least one storage medium stores at least one instruction set for performing living body detection, and the at least one processing A processor, connected in communication with the at least one storage medium, wherein, when the living body detection system is running, the at least one processor reads the at least one instruction set, and executes the first step according to the instructions of the at least one instruction set The living body detection method described in any one aspect.

It can be known from the above technical solutions that the living body detection method and system provided in this manual obtains a biometric image containing the biological characteristics of the target user, and performs preliminary living body detection on the target user based on the biometric image to obtain the preliminary living body detection result of the target user. Furthermore, the target user is instructed to input the identity verification information of the target length, and based on the actual input information of the target user, the target user is re-tested to obtain the target liveness detection result of the target user, wherein the target length and the preliminary liveness detection The results are relevant. The above scheme further introduces an identity verification process on the basis of preliminary liveness detection of the target user based on the biometric image, that is, instructs the target user to input the identity verification information of the target length, and re-inspects the target user based on the actual input information of the target user. Liveness detection, through the above-mentioned two-stage liveness detection process, can improve the accuracy of the liveness detection result as a whole. Furthermore, since the target length corresponding to the identity verification information is related to the preliminary liveness detection result, the target length has certain variability, which can further improve the accuracy of the liveness detection result compared with the fixed-length identity verification information.

Other functions of the living body detection method and system provided in this manual will be partially listed in the following description. The inventive aspects of the liveness detection methods and systems provided in this specification can be fully explained by practice or use of the methods, devices and combinations described in the detailed examples below.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of this specification, the drawings that need to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of this specification. For those skilled in the art, other drawings can also be obtained based on these drawings without creative effort.

FIG. 1 shows a schematic diagram of an application scenario of a living body detection system provided according to an embodiment of the present specification;

FIG. 2 shows a hardware structural diagram of a computing device provided according to an embodiment of this specification;

Fig. 3 shows a flow chart of a living body detection method provided according to an embodiment of the specification;

Fig. 4 shows a schematic structural diagram of a living body detection model provided according to an embodiment of the specification;

Fig. 5 shows a schematic diagram of the training process of the first training stage of the living body detection model provided according to the embodiment of the specification;

Fig. 6 shows a schematic diagram of the training process of the second training stage of the living body detection model provided according to an embodiment of the present specification; and

Fig. 7 shows an interactive schematic diagram of a living body detection process provided according to an embodiment of the present specification.

Detailed ways

The following description provides specific application scenarios and requirements of this specification, with the purpose of enabling those skilled in the art to manufacture and use the contents of this specification. Various local modifications to the disclosed embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments and embodiments without departing from the spirit and scope of the description. application. Accordingly, the description is not limited to the embodiments shown, but is to be accorded the widest scope consistent with the claims.

The terminology used herein is for the purpose of describing particular example embodiments only and is not intended to be limiting. For example, as used herein, the singular forms "a", "an" and "the" may also include the plural forms unless the context clearly dictates otherwise. When used in this specification, the terms "comprising", "comprising" and/or "containing" mean that the associated integer, step, operation, element and/or component is present, but does not exclude one or more other features , integers, steps, operations, elements, components and/or groups exist or other features, integers, steps, operations, elements, components and/or groups may be added in the system/method.

These and other features of this specification, as well as the operation and function of the relevant elements of the structure, and the economy of assembly and manufacture of the components, may be significantly enhanced in view of the following description. Reference is made to the accompanying drawings, all of which form a part of this specification. It should be clearly understood, however, that the drawings are for purposes of illustration and description only and are not intended to limit the scope of the specification. It should also be understood that the figures are not drawn to scale.

The flowcharts used in this specification illustrate the operation of system implementations according to some embodiments in this specification. It should be clearly understood that the operations of the flowcharts may be performed out of order. Instead, operations may be performed in reverse order or concurrently. Additionally, one or more other operations can be added to the flowchart. One or more operations can be removed from the flowchart.

For the convenience of description, this manual will explain the terms that appear in the following descriptions as follows:

Live attack: means of attacking face recognition systems or other biometric systems, for example, live attacks can be carried out through mobile phone screens, printed photos, high-precision masks, etc.

Liveness detection: refers to the algorithms and technologies used to prevent liveness attacks in face recognition or other biometric systems, and can be used to identify whether the target user is alive.

Before describing the specific embodiments of this specification, the application scenarios of this specification are introduced as follows:

The liveness detection method provided in this specification can be applied to the scene of authentication based on biometrics. The living body detection method provided in this specification can be regarded as a user classification method, which can classify the target user as living body or non-living body. For example, after a face recognition (such as face payment, face access control, face attendance, etc.) system collects a face image, it can use the liveness detection method provided in this manual to perform liveness detection on the target user based on the face image to determine Whether the target user is alive. If it is a living body, the subsequent face recognition process will continue to be performed. If it is not a living body, the subsequent face recognition process will not need to be performed, thereby improving the security of face recognition.

It should be noted that the above-mentioned face recognition scenario is only one of the multiple usage scenarios provided in this manual. The living body detection method provided in this manual can not only be applied to face recognition scenarios, but also can be applied to other biometric features, such as fingerprints. , palmprint, iris and other biometric authentication scenarios. Those skilled in the art should understand that the application of the living body detection method described in this specification to other usage scenarios is also within the protection scope of this specification.

Fig. 1 shows a schematic diagram of an application scenario of a living body detection system provided according to an embodiment of the present specification. Liveness detection system 001 (hereinafter referred to as system 001) can be applied to liveness detection in any scene, for example, support scene based on face recognition, access control scene based on face recognition, attendance scene based on face recognition, etc. As shown in FIG. 1 , the system 001 may include a target user 100 , a client 200 , a server 300 and a network 400 .

The target user 100 may be a user who triggers the liveness detection, and the target user 100 may perform a preset operation on the client 200 to trigger the liveness detection.

The client 200 may be a device that responds to the liveness detection operation of the target user 100 . In some embodiments, the living body detection method described in this specification can be executed on the client 200 . At this time, the client 200 may store data or instructions for executing the living body detection method described in this specification, and may execute or be used to execute the data or instructions. In some embodiments, the client 200 may include a hardware device with a data information processing function and a necessary program to drive the hardware device to work. As shown in FIG. 1 , the client 200 can communicate with the server 300 . In some embodiments, the server 300 can communicate with multiple clients 200 . In some embodiments, the client 200 can interact with the server 300 through the network 400 to receive or send messages, such as receiving or sending biometric images. In some embodiments, client 200 may comprise a mobile device, tablet computer, laptop computer, built-in device of a motor vehicle, or the like, or any combination thereof. In some embodiments, the mobile device may include a smart home device, a smart mobile device, a virtual reality device, an augmented reality device, or the like, or any combination thereof. In some embodiments, the smart home device may include a smart TV, a desktop computer, etc., or any combination thereof. In some embodiments, the smart mobile device may include a smart phone, a personal digital assistant, a gaming device, a navigation device, etc., or any combination thereof. In some embodiments, the virtual reality device or augmented reality device may include a virtual reality helmet, virtual reality glasses, virtual reality patch, augmented reality helmet, augmented reality glasses, augmented reality patch, or the like, or any combination thereof. For example, the virtual reality device or the augmented reality device may include Google Glass, a head-mounted display, VR, and the like. In some embodiments, the built-in equipment in the motor vehicle may include an on-board computer, an on-board TV, and the like. In some embodiments, the client 200 may include an image collection device, configured to collect video or image information of the target user, so as to obtain a biometric image. In some embodiments, the image acquisition device can be a two-dimensional image acquisition device (such as an RGB camera), or a two-dimensional image acquisition device (such as an RGB camera) and a depth image acquisition device (such as a 3D structured light camera, laser detectors, etc.). In some embodiments, the client 200 may be a device with positioning technology for locating the location of the client 200 .

In some embodiments, the client 200 may be installed with one or more application programs (APP). The APP can provide the target user 110 with the ability and interface to interact with the outside world through the network 400 . The APP includes but is not limited to: web browser APP programs, search APP programs, chat APP programs, shopping APP programs, video APP programs, wealth management APP programs, instant messaging tools, email clients, social platforms software and more. In some embodiments, the target APP may be installed on the client 200 . The target APP can collect video or image information of the target user for the client 200, so as to obtain a biometric image. In some embodiments, the target user 100 may also trigger a liveness detection request through the target APP. The target APP may respond to the living body detection request and execute the living body detection method described in this specification. The living body detection method will be described in detail later.

The server 300 may be a server that provides various services, such as a background server that provides support for the biometric image collected on the client 200 . In some embodiments, the living body detection method described in this specification can be executed on the server 300 . At this point, the server 300 may store data or instructions for executing the living body detection method described in this specification, and may execute or be used to execute the data or instructions. In some embodiments, the server 300 may include a hardware device with a data information processing function and a necessary program to drive the hardware device to work. The server 300 can communicate with multiple clients 200 and receive data sent by the clients 200 .

The network 400 is used to provide a communication medium between the client 200 and the server 300 . Network 400 may facilitate the exchange of information or data. As shown in FIG. 1 , the client 200 and the server 300 can be connected to a network 400 and transmit information or data to each other through the network 400 . In some embodiments, network 400 may be any type of wired or wireless network, or a combination thereof. For example, the network 400 may include a cable network, a wired network, a fiber optic network, a telecommunications network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a wide area network (WAN), Public Switched Telephone Network (PSTN), Bluetooth network, ZigBee network, Near Field Communication (NFC) network or similar network. In some embodiments, network 400 may include one or more network access points. For example, network 400 may include wired or wireless network access points, such as base stations or Internet exchange points, through which one or more components of client 200 and server 300 may connect to network 400 to exchange data or information.

It should be understood that the numbers of clients 200, servers 300 and networks 400 in FIG. 1 are merely illustrative. There may be any number of clients 200, servers 300, and networks 400 according to implementation needs.

It should be noted that the living body detection method described in this specification may be completely executed on the client 200 , may also be completely executed on the server 300 , or may be partially executed on the client 200 and partially executed on the server 300 .

FIG. 2 shows a hardware structure diagram of a computing device 600 provided according to an embodiment of this specification. The computing device 600 can execute the living body detection method described in this specification. The liveness detection method is described elsewhere in this specification. When the living body detection method is executed on the client 200 , the computing device 600 may be the client 200 . When the living body detection method is executed on the server 300 , the computing device 600 may be the server 300 . When the living body detection method can be partially executed on the client 200 and partially executed on the server 300 , the computing device 600 can include the client 200 and the server 300 .

As shown in FIG. 2 , computing device 600 may include at least one storage medium 630 and at least one processor 620 . In some embodiments, computing device 600 may also include a communication port 650 and an internal communication bus 610 . Meanwhile, the computing device 600 may further include an I/O component 660 .

Internal communication bus 610 may connect various system components, including storage media 630 , processor 620 and communication ports 650 .

I/O component 660 supports input/output between computing device 600 and other components.

The communication port 650 is used for data communication between the computing device 600 and the outside world, for example, the communication port 650 may be used for data communication between the computing device 600 and the network 400 . The communication port 650 may be a wired communication port or a wireless communication port.

The storage medium 630 may include a data storage device. The data storage device may be a non-transitory storage medium or a temporary storage medium. For example, the data storage device may include one or more of a magnetic disk 632 , a read-only storage medium (ROM) 634 or a random-access storage medium (RAM) 636 . The storage medium 630 also includes at least one instruction set stored in the data storage device. The instructions stored in the instruction set are computer program codes, and the computer program codes may include programs, routines, objects, components, data structures, procedures, modules, etc. for executing the living body detection method provided in this specification.

At least one processor 620 may be communicatively connected to at least one storage medium 630 and a communication port 650 through an internal communication bus 610 . At least one processor 620 is configured to execute the above at least one instruction set. When the computing device 600 is running, at least one processor 620 reads the at least one instruction set, and executes the living body detection method provided in this specification according to the instruction of the at least one instruction set. The processor 620 may execute all or part of the steps included in the living body detection method. Processor 620 may be in the form of one or more processors. In some embodiments, processor 620 may include one or more hardware processors, such as microcontrollers, microprocessors, reduced instruction set computers (RISC), Application Specific Integrated Circuit (ASIC), Application Specific Instruction Set Processor (ASIP), Central Processing Unit (CPU), Graphics Processing Unit (GPU), Physical Processing Unit (PPU), Microcontroller Unit, Digital Signal Processor ( DSP), Field Programmable Gate Array (FPGA), Advanced RISC Machine (ARM), Programmable Logic Device (PLD), any circuit or processor capable of performing one or more functions, etc., or any combination thereof. For purposes of illustration only, only one processor 620 is depicted in computing device 600 in this specification. However, it should be noted that the computing device 600 in this specification may also include multiple processors, therefore, the operations and/or method steps disclosed in this specification may be executed by one processor as described in this specification, or by multiple processors. Processor joint execution. For example, if the processor 620 of the computing device 600 executes step A and step B in this specification, it should be understood that step A and step B can also be jointly or separately executed by two different processors 620 (for example, the first processor Step A is performed and the second processor performs step B, or the first and second processors jointly perform steps A and B).

Fig. 3 shows a flow chart of a living body detection method P100 provided according to an embodiment of the present specification. As before, the computing device 600 can execute the living body detection method P100 described in this specification. Specifically, the processor 620 may read the instruction set stored in its local storage medium, and then execute the living body detection method P100 described in this specification according to the provisions of the instruction set. As shown in Figure 3, the method P100 may include:

S110: Acquire a biometric image, where the biometric image includes the biometrics of the target user.

Wherein, the biometric image may refer to an image containing the biometrics of the target user 100 . Utilizing the uniqueness of the biometric feature, the biometric image can be used to identify or verify the identity information of the target user 100 . The biological features involved in the biometric identification described in the embodiments of the present application may include, for example, eye prints, voice prints, fingerprints, palm prints, heartbeat, pulse, chromosomes, DNA, human tooth bite marks, and the like. In some embodiments, the biometric feature may include at least one of human face, eye print, fingerprint, palm print, or iris. Wherein, eye patterns may include biological characteristics such as iris and sclera. In other words, the biometric image may be at least one of a face image, an eye print image, a fingerprint image, a palm print image, or an iris image.

In some embodiments, when the computing device 600 is the client 200 , the client 200 may use the image collection module to collect the above biometric image of the target user 100 . In some embodiments, when the computing device 600 is the server 300, the server 300 may receive a biometric image from the client 200, wherein the biometric image may be obtained by the client 200 using an image collection module to collect the target user 100 of.

S120: Perform a preliminary living body detection on the target user based on the biometric image, and obtain a preliminary living body detection result of the target user.

Wherein, the above-mentioned preliminary living body detection result is used to preliminarily indicate whether the target user 100 is alive. For example, the preliminary live body detection result may include: a first probability that the target user 100 is alive.

In some embodiments, the processor 620 may input the biometric image into a pre-trained live body detection model, wherein the live body detection model has the ability to identify whether the target user 100 is alive. In this way, the preliminary living body detection result of the target user 100 can be obtained through the preliminary living body detection of the target user 100 through the living body detection model, that is, the first probability that the target user 100 is alive can be obtained. Wherein, it should be noted that the above-mentioned living body detection model may be any machine learning model with the ability of living body detection.

In some embodiments, in addition to the ability to identify whether the target user 100 is alive, the above-mentioned living body detection model may also have the ability to detect the confidence of the model output result. Specifically, the processor 620 inputs the biometric image into the trained live body detection model, and performs live body detection processing on the biometric image through the live body detection model to obtain the second probability that the target user is a living body and the corresponding Confidence. In this case, the processor 620 may determine the preliminary liveness detection result of the target user 100 based on the second probability and the confidence corresponding to the second probability, that is, the first probability that the target user 100 is alive. In some embodiments, it is assumed that the second probability output by the liveness detection model is denoted as P, and the confidence degree corresponding to the second probability P is denoted as C, where the value ranges of P and C are both [0,1]. The product of the second probability P and the confidence C can be determined as the first probability S, namely:

S=C*P formula (1)

It can be seen from the above formula (1) that the first probability S is positively correlated with the second probability P, and is positively correlated with the confidence level C, that is, the higher the second probability P output by the living body detection model, and the higher the confidence level C The higher it is, the higher the first probability S is. It should be understood that the above-mentioned first probability S can be regarded as a probability obtained by correcting the second probability P output by the model by using the confidence degree C, therefore, the accuracy of the first probability S is higher than that of the second probability P.

Fig. 4 shows a schematic structural diagram of a living body detection model provided according to an embodiment of the present specification. As shown in FIG. 4 , the living body detection model 400 may include: a feature extraction network 401 , a local living body detection network 402 , a global living body detection network 403 and a confidence detection network 404 . Referring to FIG. 4 , the liveness detection process of the target user 100 by the liveness detection model 400 may include:

(1) Input the biometric image into the feature extraction network 401, and perform feature extraction processing on the biometric image through the feature extraction network 401 to obtain a feature map.

(2) Input the feature map into the local live body detection network 402, and perform live body detection processing on multiple local areas of the feature map through the local live body detection network 402, and obtain live body detection results in multiple local areas.

Wherein, the above-mentioned plurality of local regions refers to a plurality of regions obtained by region-dividing the feature map according to a certain region-dividing manner. For example, the feature map may be divided into 3 equal parts in the height direction and 3 equal parts in the width direction, so as to divide the feature map into 3*3=9 local regions. It should be understood that, the embodiment of the present specification does not limit the manner of dividing the local areas, for example, the multiple local areas may also be areas obtained by dividing randomly. The living body detection result of each local area indicates the detection result obtained by performing live body detection on the target user based on the image content of the local area. Taking the feature map divided into 9 local areas as an example, the feature map is input into the local living body detection network, and the probability of living body corresponding to the 9 local areas can be obtained.

(3) Input the liveness detection results of the feature map and multiple local regions into the global liveness detection network 403, and perform liveness detection processing on the feature map and the liveness detection results of multiple local regions through the global liveness detection network 403 to obtain the global liveness detection result .

Wherein, the global living body detection result may include: the second probability that the target user is living (that is, the probability P described above). It should be understood that when the global liveness detection network 403 performs global liveness detection, it is not only based on the feature map, but also considers the liveness detection results of multiple local regions, so that the obtained global liveness detection results are more accurate.

(4) Input the living body detection results of multiple local regions into the confidence degree detection network 404 to obtain the confidence degree corresponding to the global living body detection result.

Specifically, the confidence level detection network 404 may determine the confidence levels corresponding to the multiple local areas based on the living body detection results of the multiple local areas. Wherein, the confidence level corresponding to each local area indicates the credibility of the living body detection result of the local area. Furthermore, the confidence level detection network 404 may determine the confidence level corresponding to the global living body detection result (for example, the confidence level C described above) based on the confidence levels corresponding to multiple local regions. For example, a weighted mean operation may be performed on the confidence degrees corresponding to multiple local regions to obtain the confidence degree corresponding to the global living body detection result.

It should be noted that the embodiment of this description does not limit the specific network structure of the above-mentioned feature extraction network 401, local living body detection network 402, global living body detection network 403, and confidence degree detection network 404. For example, CNN (Convolutional Neural Network , convolutional neural network), DNN (Deep Neural Network, deep neural network), RNN (Recurrent Neural Network, cyclic neural network) or any other feasible network structure.

The training process of the living body detection model shown in FIG. 4 will be described below.

The living body detection model can be trained in a staged training manner. In some embodiments, the training process of the living body detection model may include a first training phase and a second training phase. Wherein, the first training stage is configured to train the feature extraction network 401 , the local living body detection network 402 and the global living body detection network 403 . The second training phase is configured to train the confidence detection network 404 . That is to say, the feature extraction network 401 , the local living body detection network 402 and the global living body detection network 403 are trained first, and then the confidence detection network 404 is trained after the three network units converge. It should be understood that by performing training in stages, the amount of training calculations in each stage can be reduced, which is beneficial to improving the overall training efficiency.

Fig. 5 shows a schematic diagram of the training process of the first training stage of the living body detection model provided according to the embodiment of the present specification. As shown in Figure 5, in some embodiments, the training process of the above-mentioned first training phase may include:

(1) Acquiring a sample biometric image and labeling information corresponding to the sample biometric image, where the labeling information includes: global living body labeling results and living body labeling results in multiple local regions.

Wherein, the annotation information corresponding to the sample biometric image is used to indicate whether the sample biometric image is obtained by collecting a living body or by attacking a living body.

(2) Referring to FIG. 5 , the sample biometric image is input into the feature extraction network 401 to obtain a sample feature map, and the sample feature map is input into the local living body detection network 402 to obtain the first live body detection results of multiple local regions, and the The sample feature map and the first liveness detection results of the multiple local regions are input into the global liveness detection network 403 to obtain the first global liveness detection result.

(3) Continue referring to FIG. 5 , based on the first global living body detection result, the first living body detection results of the multiple local regions, the global living body labeling result, and the living body labeling results of the multiple local regions to determine the first A target loss, and aim at minimizing the first target loss to train the feature extraction network 401, the local living body detection network 402 and the global living body detection network 403.

In some embodiments, the calculation method of the first target loss may be as follows: determine the global loss based on the difference between the first global living body detection result and the global living body labeling result, and determine the global loss based on the first living body in the multiple local regions The difference between the detection result and the living body labeling results of the plurality of local regions is determined to determine a local loss, and then determine the first target loss based on the global loss and the local loss. Continuing to refer to FIG. 5, after the first target loss is determined, the network parameters of the feature extraction network 401, the local living body detection network 402 and the global living body detection network 403 can be adjusted to minimize the first target loss until the above three network units reach the convergence condition, or until the preset number of iterations is reached.

In the above first training stage, not only global loss but also local loss are considered when determining the first target loss, so that the determined first target loss is more accurate. After the first training stage, the living body detection model has the ability to perform living body detection based on biometric images.

In some embodiments, the first training stage may further include: performing perturbation processing on the sample biometric image to obtain a disturbed biometric image. Wherein, the above disturbance processing may be global disturbance or local disturbance, for example, noise may be added to all or part of the sample biometric image to form a disturbed biometric image. Furthermore, inputting the disturbance biometric image into the feature extraction network to obtain a disturbance feature map, and inputting the disturbance feature map into the local living body detection network to obtain second living body detection results in multiple local regions.

On this basis, based on the first global living body detection result, the first living body detection results of the multiple local regions, the second living body detection results of the multiple local regions, the global living body labeling result and the The first target loss is determined based on the living body labeling results of the plurality of local regions. In some embodiments, the calculation method of the first target loss may be as follows: determine the first loss based on the difference between the first global living body detection result and the global living body labeling result, and determine the first loss based on the first Determine the second loss based on the difference between the living body detection result and the living body labeling results of the plurality of local regions, based on the first living body detection result of the plurality of local regions and the second living body detection result of the plurality of local regions Determine the third loss based on the difference between them, and then determine the first target loss based on the first loss, the second loss and the third loss. Wherein, the above-mentioned first loss is used to represent the global loss, the above-mentioned second loss is used to represent the local loss, and the above-mentioned third loss is used to represent the perturbation consistency loss, that is, the difference between the local living body detection results of the image before and after the perturbation.

In the above first training stage, not only the original sample biometric image but also the disturbed biometric image are used, and when determining the first target loss, not only the global loss (i.e. the first loss) and the local loss (i.e. the second loss) are considered Loss), also consider the perturbation consistency loss (ie, the third loss), so that the trained liveness detection model is not sensitive to perturbations, that is, the liveness detection results of the perturbed images are consistent with the liveness detection results of the non-perturbed images. In this way, the trained live body detection model has high accuracy for the live body detection results of the perturbed image and the non-perturbed image.

Fig. 6 shows a schematic diagram of the training process of the second training stage of the living body detection model provided according to the embodiment of the present specification. As shown in Figure 6, in some embodiments, the training process of the second training phase may include:

(1) Acquiring a sample biometric image and labeling information corresponding to the sample biometric image, where the labeling information includes: global living body labeling results and living body labeling results in multiple local regions.

(2) Referring to FIG. 6 , the sample biometric image is input into the feature extraction network 401 to obtain a sample feature map, and the sample feature map is input into the local living body detection network 402 to obtain the first living body detection results of multiple local regions, and the multiple live body detection results are obtained. The first living body detection result of a local region is input into the confidence degree detection network 404 to obtain the detection confidence degrees of the plurality of local regions, and based on the detection confidence degrees of the plurality of local regions, the corresponding local regions of the sample feature map are updated to obtain Update the sample feature map, input the updated sample feature map into the local living body detection network 402 to obtain the third live body detection results of multiple local regions, and input the updated sample feature map and the third live body detection results of the multiple local regions into the global living body The detection network 403 obtains a second global living body detection result.

The difference between the second training stage and the first training stage is that in the second training stage, after obtaining the first living body detection results of multiple local regions through the local living body detection network 402, multiple local living body detection results are also obtained through the confidence detection network 404. The detection confidence of the region, and then, the detection confidence of multiple local regions is used as the weight coefficient to update the corresponding local regions of the feature map to obtain the updated feature map. Furthermore, liveness detection is performed on the updated feature map through the local liveness detection network 402 and the global liveness detection network 403 . It should be understood that the liveness detection process of the local liveness detection network 402 and the global liveness detection network 403 for updating the feature map is similar to the first training stage, and will not be repeated here.

(3) Continue referring to FIG. 6 , based on the second global living body detection result, the third living body detection result of the plurality of local regions, the global living body labeling result and the living body labeling results of the multiple local regions to determine the first Two target losses, and the confidence detection network 404 is trained with the training target of minimizing the second target loss.

It should be understood that, the calculation method of the second target loss in the second training stage is similar to the calculation method of the first target loss in the first training stage, and will not be repeated here. After the second target loss is calculated, the network parameters of the confidence detection network 404 are adjusted with the training goal of minimizing the second target loss until the confidence detection network 404 reaches a convergence condition, or until a preset number of iterations is reached.

After the first training stage and the second training stage above, the liveness detection model not only has the ability to detect liveness, but also has the ability to detect the confidence of the model output results, that is, the liveness detection model not only outputs the second probability P, but also outputs Confidence C.

S130: Instruct the target user to input identity verification information of a target length, and perform another biopsy detection on the target user based on the actual input information of the target user, and obtain a target biopsy detection result of the target user, wherein the The target length is correlated with the preliminary liveness test result.

Wherein, the identity verification information refers to information used to verify the identity of the target user. For example, the identity verification information may include at least one of the following: user identification (such as certificate number), user terminal identification (such as mobile phone number), registration account identification (such as the account registered by the user in the app), registration password, etc. .

In the embodiments of this specification, the preliminary biopsy detection result may also be referred to as an intermediate biopsy detection result, and the target biopsy detection result may also be referred to as a final biopsy detection result.

In the embodiment of the present specification, after obtaining the preliminary living body detection result of the target user, an identity verification process is further introduced, that is, the target user is instructed to input identity verification information of a target length. Furthermore, the processor 620 acquires the actual input information input by the target user according to the instruction, and conducts another biopsy detection on the target user based on the actual input information, to obtain a target biopsy detection result of the target user. It can be understood that the above scheme can improve the accuracy of the liveness detection result as a whole through the detection in the two stages of the preliminary liveness detection and the second liveness detection. Furthermore, since the target length is related to the preliminary living body detection result, the target length has certain variability, which can further improve the accuracy of the living body detection result compared with identity verification information of a fixed length.

In some embodiments, the preliminary living body detection result includes: a first probability that the target user is a living body. The aforementioned target length is inversely correlated with the first probability. That is to say, the greater the first probability, the shorter the target length, and the smaller the first probability, the longer the target length. It should be understood that when the first probability is higher, it means that the target user has a higher probability of being alive, that is to say, the security of this biometric identification is higher. Therefore, in this case, the target user can be instructed to enter a shorter identity Verify information, which can reduce the time-consuming interaction of target users and improve the efficiency of liveness detection. When the first probability is small, it means that the target user has a high probability of being non-living, that is to say, the security of this biometric identification is low. Therefore, in this case, the target user can be instructed to input longer identity verification information. In order to improve the security of this biometric identification. It can be seen that the above-mentioned target length has an inverse correlation with the first probability, which can reduce the time-consuming interaction of the target user as much as possible and improve the efficiency of liveness detection under the premise of ensuring the safety of biometric identification.

In some embodiments, before S130, it may further include: determining a target identity verification type, and determining a target length based on the target identity verification type and the first probability. Wherein, the target identity verification type may be one of the following multiple identity verification types to be selected: identity verification through user ID, identity verification through user terminal ID, identity verification through registered account ID, and registration password Authenticate. The above different authentication types have different security guarantees, and one of the authentication types can be determined as the target authentication type according to requirements.

In some embodiments, one of the following methods may be used to determine the target identity authentication type among multiple identity authentication types to be selected:

Method 1: Randomly determine the target authentication type among multiple authentication types to be selected. For example, one of multiple authentication types to be selected may be randomly selected as the target authentication type. The above-mentioned random method prevents a certain type of identity verification from being fixedly used in different biometric identification processes, thereby further improving the security of biometric identification.

Manner 2: Based on the type of the current application scenario, select an identity authentication type that matches the type of the current application scenario from multiple candidate authentication types as the target identity authentication type. Different identity verification types to be selected correspond to different security levels. For example, the security level of "authentication through registration password" is higher than that of "authentication through user terminal identification". In the scenario of , you can use "authentication by registration password" as the target authentication type, and in the scenario with low security requirements, you can use "authentication by user terminal identity" as the target authentication type. This method can meet the security requirements of different application scenarios.

Mode 3: use the authentication type specified by the target user among multiple authentication types to be selected as the target authentication type. For example, the target user can set the target authentication type at the registration stage or any other stage, so as to specify which authentication type is used as the target authentication type. This method can increase the verification flexibility of target users and meet the personalized verification requirements of different users.

In some embodiments, the target length may be determined in the following manner: input the target identity verification type and the first probability into a pre-trained length mapping model to obtain the target length. Wherein, the length mapping model is trained by using multiple sets of training samples, and each set of training samples includes: sample identity verification type, sample probability, and sample length. The above multiple sets of training samples are obtained based on historical data statistics, and the security of the biometric identification corresponding to the above historical data meets the preset requirements. It should be noted that, the embodiment of this specification does not limit the network structure of the length mapping model. For example, a neural network can be used, an MLP (Multi-Layer Perceptron, multi-layer perceptron) network can also be used, or any other feasible network structure. The trained length-mapping model has the ability to map target authentication types and first probabilities to target lengths. In the above embodiments, the target length is determined by using the length mapping model obtained through pre-training. On the one hand, it has high processing efficiency, and on the other hand, the target length has high accuracy.

In some embodiments, the target length can also be determined in the following manner: determine the weight coefficient based on the type of the current application scenario, and the weight coefficient is inversely correlated with the degree of security requirements of the current application scenario for biometric identification; The weighting coefficient performs weighting processing on the first probability to obtain the first probability after scene adaptation; and determine the target length based on the target identity verification type and the first probability after scene adaptation.

Specifically, the higher the security requirements for biometric identification in the current application scenario, the smaller the weight coefficient, and the lower the security requirements for biometric identification in the current application scenario, the larger the weight coefficient. For example, scenarios may be divided into high-security scenarios, medium-security scenarios, and low-security scenarios based on the level of security requirements. If the current application scenario is a high-security scenario, set the weight factor to 0.8; if the current application scenario is a medium-security scenario, set the weight factor to 1; if the current application scenario is a low-security scenario, set the weight factor Determined to be 1.25.

The determined weight coefficient is denoted as α, and the first probability is denoted as S, then the first probability S ^' after scene adaptation can be obtained by the following formula:

S ^′ =*S

After the scene-adapted first probability S ^' is determined, the target identity verification type and the scene-adapted first probability S ^' can be input into a trained length mapping model to obtain the target length.

After the target length is determined, the target user may be instructed to input identity verification information of the target length. For example, assuming that the target length is 4 and the target authentication type is "authentication by user terminal ID", the target user may be instructed to input the last four characters of the terminal ID (such as a mobile phone number). For another example, assuming that the target length is 6, and the target authentication type is "authentication by registration password", the target user may be instructed to input the 2nd to 7th characters of the registration password. It should be noted that the embodiment of this specification does not limit the method of instructing the target user. For example, the target user may be instructed to enter the identity verification information of the target length by text in the interactive interface, or the target user may be instructed by voice Enter a target length of authentication information, or instruct the target user to enter a target length of authentication information by any other means.

After instructing the target user to input identity verification information of a target length, the processor 620 may acquire actual input information input by the target user based on the instruction. It should be noted that the embodiment of this specification does not limit the input method adopted by the target user. For example, the target user may input by text, by voice, or by any other possible input. Further, the processor 620 may perform another biopsy detection on the target user based on the actual input information of the target user, so as to obtain a target biopsy detection result of the target user.

In some embodiments, the target liveness detection result of the target user may be determined in the following manner: the identity information of the first user stored in the database is determined as the target identity information, wherein the first user is the login user who triggers the liveness detection , or, the first user is a user obtained by performing biometric identification on a biometric image. For example, the currently logged-in user may be used as the first user, and the identity information of the first user may be queried in the database as the target identity information. For another example, biometric identification may be performed on the biometric image, the identified user may be used as the first user, and the identity information of the first user may be queried in the database as the target identity information. Furthermore, the actual input information of the target user is matched with the target identity information to obtain a matching result, and one of the first operation and the second operation is executed based on the matching result. Wherein, the first operation includes: determining that the matching result is a successful match, then determining that the target user is a living body, and the second operation includes: determining that the matching result is a matching failure, then determining that the target user is not a living body.

Fig. 7 shows an interactive schematic diagram of a living body detection process provided according to an embodiment of the present specification. As shown in FIG. 7 , taking the facial recognition payment scenario as an example, the client 200 presents an interface 701 , and in the interface 701 , after the target user clicks “Start Paying with Face”, the client 200 presents an interface 702 . As shown in interface 702, the client 200 activates the image collection module to collect the face image of the target user. After the face image is collected, the client 200 or the server 300 (for example, the client 200 sends the biometric image to the server 300) inputs the face image into the living body detection model shown in FIG. Probability P and confidence C, and then calculate the first probability S that the target user is a living body. Further, based on the target identity verification type and the first probability S, the target length of the identity verification information that needs to be input by the target user in the identity verification phase can be determined. Assuming that the determined target length is 4, the client 200 may present an interface 703 in which the target user is instructed to input 4 digits of authentication information (for example, the last 4 digits of the mobile phone number). Referring to interface 704, after the target user enters the 4-digit identity verification information, and clicks OK, the client 200 or server 300 obtains the actual input information of the target user (such as "1234"), and compares the actual input information with the first user's ID stored in the database. Identity information is matched, if the matching is successful, it is determined that the target user is a living body, and if the matching fails, it is determined that the target user is not a living body. When it is determined that the target user is a living body, the client 200 or server 300 can perform face recognition on the target user based on the face image to verify the identity of the target user, and pay when the verification is passed. After the payment is successful An interface 705 is displayed. When it is determined that the target user is not a living body, the client 200 or server 300 does not perform the subsequent face recognition process, and the face payment fails this time, and the target user identity verification failure can be prompted on the display interface.

It should be noted that Figure 7 uses the face payment scenario as an example to illustrate the interaction process. The interaction process of the live detection solution provided in this specification is similar to other scenarios, and this specification does not give examples.

In some embodiments, the preliminary living body detection result includes a first probability that the target user is a living body, and if it is determined that the first probability is greater than or equal to a preset probability, S130 is executed. For example, the preset probability may be 30%, that is, S130 is executed only when the first probability is greater than or equal to 30%. In this way, the situation that the target user is obviously not alive can be filtered out, that is to say, in the case that the target user is obviously not alive, there is no need to perform an identity verification link, which can improve the efficiency of liveness detection.

In some embodiments, after the actual input information of the target user is obtained, the target user may be re-detected based on the actual input information to obtain the result of the re-living detection, and then based on the preliminary living detection result and the re- The liveness detection result determines the target liveness detection result of the target user. For example, assuming that the preliminary liveness detection result indicates that the probability that the target user is alive is 55%, and the probability that the second liveness detection result indicates that the target user is alive is 90%, then the mean (or weighted mean) of the above two probabilities can be used as the target The final probability that the user is a living body, if the final probability is greater than or equal to the preset threshold, it is determined that the target user is a living body, and if the final probability is less than the preset probability, then it is determined that the target user is not a living body. In this embodiment, the target biopsy detection result can be determined by comprehensively analyzing the preliminary biopsy detection result and the re-vitality detection result, which can further improve the accuracy of the target biopsy detection result.

To sum up, the liveness detection method P100 and system 001 provided in this manual obtain a biometric image containing the biological characteristics of the target user, conduct preliminary liveness detection on the target user based on the biometric image, and obtain the preliminary liveness detection result of the target user , furthermore, instructing the target user to input the identity verification information of the target length, and based on the actual input information of the target user, perform another biopsy detection on the target user to obtain the target biopsy detection result of the target user, wherein the target length is the same as the preliminary biopsy The test results are related. The above scheme further introduces an identity verification process on the basis of preliminary liveness detection of the target user based on the biometric image, that is, instructs the target user to input the identity verification information of the target length, and re-inspects the target user based on the actual input information of the target user. Liveness detection, through the above-mentioned two-stage liveness detection process, can improve the accuracy of the liveness detection result as a whole. Furthermore, since the target length corresponding to the identity verification information is related to the preliminary liveness detection result, the target length has certain variability, which can further improve the accuracy of the liveness detection result compared with the fixed-length identity verification information. In addition, the above solution performs a variable-length identity verification process based on the preliminary living body detection result, which also makes the solution have higher application flexibility. For example, when the preliminary liveness detection result indicates that the probability that the target user is alive is high, the target user can be instructed to enter a shorter length of identity verification information to save the interaction time of the target user and improve the efficiency of liveness detection; When the probability that the target user is a living body is low, the target user may be instructed to input a longer length of identity verification information, so as to ensure the accuracy of the living body detection result as much as possible.

In the technical solutions provided in this manual, the collection, storage, use, processing, transmission, provision, and disclosure of users' personal information involved are all in compliance with relevant laws and regulations, and do not violate public order and good customs.

Another aspect of the present specification provides a non-transitory storage medium storing at least one set of executable instructions for liveness detection. When the executable instruction is executed by the processor, the executable instruction instructs the processor to implement the steps of the living body detection method P100 described in this specification. In some possible implementations, various aspects of this specification can also be implemented in the form of a program product, which includes program codes. When the program product runs on the computing device 600, the program code is used to make the computing device 600 execute the steps of the living body detection method P100 described in this specification. A program product for implementing the above methods may include program codes in a portable compact disk read only memory (CD-ROM), and may run on the computing device 600 . However, the program product in this specification is not limited thereto. In this specification, a readable storage medium may be any tangible medium containing or storing a program, and the program may be used by or in combination with the instruction execution system. The program product may reside on any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples of readable storage media include: electrical connections with one or more conductors, portable disks, hard disks, random access memory (RAM), read only memory (ROM), erasable programmable read only memory ( EPROM or flash memory), fiber optics, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. The computer readable storage medium may include a data signal carrying readable program code in baseband or as part of a carrier wave traveling as a data signal. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted by any suitable medium, including but not limited to wireless, cable, optical cable, RF, etc., or any suitable combination of the above. Program code for performing the operations of this specification may be written in any combination of one or more programming languages, including object-oriented programming languages—such as Java, C++, etc., as well as conventional procedural programming languages. Programming language - such as "C" or a similar programming language. The program code may execute entirely on the computing device 600, partly on the computing device 600, as a stand-alone software package, partly on the computing device 600 and partly on a remote computing device, or entirely on the remote computing device implement.

The foregoing describes specific embodiments of this specification. Other implementations are within the scope of the following claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. Multitasking and parallel processing are also possible or advantageous in certain embodiments.

To sum up, after reading this detailed disclosure, those skilled in the art can understand that the foregoing detailed disclosure may be presented by way of example only, and may not be restrictive. Although not explicitly stated here, those skilled in the art can understand that this description needs to cover various reasonable changes, improvements and modifications to the embodiments. Such alterations, improvements and modifications are intended to be suggested by this description, and are within the spirit and scope of the exemplary embodiments of this description.

In addition, certain terms in this specification have been used to describe the embodiments of this specification. For example, "one embodiment," "an embodiment" and/or "some embodiments" means that a particular feature, structure or characteristic described in connection with the embodiment can be included in at least one embodiment of the specification. Thus, it is emphasized and should be understood that two or more references to "an embodiment" or "one embodiment" or "an alternative embodiment" in various parts of this specification are not necessarily all referring to the same embodiment. Also, specific features, structures or characteristics may be properly combined in one or more embodiments of the present specification.

It should be understood that in the foregoing description of the embodiments of the present specification, in order to facilitate the understanding of one feature, for the purpose of simplifying the present specification, the present specification combines various features in a single embodiment, drawing or description thereof. However, this does not mean that the combination of these features is necessary, and it is entirely possible for those skilled in the art to label some of the devices as separate embodiments when reading this specification. That is to say, the embodiments in this specification can also be understood as the integration of multiple sub-embodiments. It is also true that each sub-embodiment lies in less than all features of a single preceding disclosed embodiment.

Every patent, patent application, publication of a patent application, and other material, such as articles, books, specifications, publications, documents, articles, etc., cited herein is hereby incorporated by reference. The entire content for all purposes, except for any prosecution history related thereto, any identical prosecution history that may be inconsistent with or conflict with this document, or any identical prosecution history that may have a restrictive effect on the broadest scope of the claims . now or in the future to be associated with this document. For example, in the event of any inconsistency or conflict in the description, definition, and/or use of terms associated with any contained material in relation to this document, use the term prevails.

Finally, it is to be understood that the embodiments of the application disclosed herein are illustrative of the principles of the embodiments of the specification. Other modified embodiments are also within the scope of this specification. Therefore, the embodiments disclosed in this specification are only illustrative and not restrictive. Those skilled in the art may implement the applications in this specification by adopting alternative configurations according to the embodiments in this specification. Therefore, the embodiments of the present specification are not limited to the embodiments precisely described in the application.

Claims (19)

1. A living body detection method, comprising:

acquiring a biological characteristic image, wherein the biological characteristic image comprises biological characteristics of a target user;

performing preliminary living body detection on the target user based on the biological characteristic image to obtain a preliminary living body detection result of the target user; and

and indicating the target user to input identity verification information of a target length, and performing secondary living detection on the target user based on the actual input information of the target user to obtain a target living detection result of the target user, wherein the target length is related to the primary living detection result.

2. The method of claim 1, wherein the preliminary in-vivo detection result comprises: the target user is a first probability of a living body; and

The target length is inversely related to the first probability.

3. The method of claim 2, wherein prior to instructing the target user to enter authentication information of a target length, further comprising:

determining a target identity verification type; and

the target length is determined based on the target authentication type and the first probability.

4. A method according to claim 3, wherein the determining the target length based on the target authentication type and the first probability comprises:

determining a weight coefficient based on the type of a current application scene, wherein the weight coefficient and the requirement degree of the current application scene on the biological feature recognition safety are in an anti-correlation relationship;

weighting the first probability based on the weight coefficient to obtain a first probability after scene adaptation; and

and determining the target length based on the target identity verification type and the first probability after scene adaptation.

5. A method according to claim 3, wherein the determining the target length based on the target authentication type and the first probability comprises:

inputting the target identity verification type and the first probability into a pre-trained length mapping model to obtain the target length,

The length mapping model is obtained by training a plurality of groups of training samples, and each group of training samples comprises: sample authentication type, sample probability, and sample length.

6. A method according to claim 3, wherein said determining a target authentication type comprises:

randomly determining the target identity verification type in a plurality of identity verification types to be selected; or alternatively

Determining the type of a current application scene, and selecting an identity verification type matched with the type of the current application scene from a plurality of identity verification types to be selected as the target identity verification type; or alternatively

And taking the identity verification type appointed by the target user in a plurality of identity verification types to be selected as the target identity verification type.

7. The method of claim 6, wherein the plurality of candidate authentication types comprises at least two of:

identity verification is carried out through the user identification;

identity verification is carried out through the identification of the user terminal;

identity verification is carried out through a registration password; and

and carrying out identity verification through the registration account number identification.

8. The method of claim 1, wherein the performing the re-living detection on the target user based on the actual input information of the target user to obtain the target living detection result of the target user comprises:

Determining the identity information of a first user stored in a database as target identity information, wherein the first user is a login user triggering living body detection, or the first user is a user obtained by performing biological feature recognition on the biological feature image; and

matching the actual input information with the target identity information, and selecting one of the first operation or the second operation to execute based on the matching result, wherein,

the first operation includes: determining that the matching result is successful, determining that the target user is a living body, and

the second operation includes: and if the matching result is determined to be the matching failure, determining that the target user is a non-living body.

9. The method of claim 1, wherein the preliminary in-vivo detection result comprises: the target user is a first probability of a living body; and

the preliminary living body detection is carried out on the target user based on the biological characteristic image, and a preliminary living body detection result of the target user is obtained, and the method comprises the following steps:

inputting the biological characteristic image into a trained living body detection model, and performing living body detection processing on the biological characteristic image through the living body detection model to obtain a second probability that the target user is living body and a confidence level corresponding to the second probability; and

And determining the first probability based on the second probability and the confidence corresponding to the second probability.

10. The method of claim 9, wherein determining the first probability based on the second probability and a confidence level corresponding to the second probability comprises:

and determining the product of the second probability and the confidence corresponding to the second probability as the first probability.

11. The method of claim 9, wherein the living detection model comprises: a feature extraction network, a local living detection network, a global living detection network, and a confidence detection network; and

performing living body detection processing on the biological feature image through the living body detection model to obtain a second probability that the target user is living body and a confidence level corresponding to the second probability, wherein the living body detection processing comprises the following steps:

performing feature extraction processing on the biological feature image through the feature extraction network to obtain a feature map;

respectively performing living body detection processing on a plurality of local areas of the characteristic map through the local living body detection network to obtain living body detection results of the plurality of local areas;

performing living body detection processing on the characteristic map and living body detection results of the plurality of local areas through the global living body detection network to obtain a second probability that the target user is living body; and

And determining the confidence corresponding to the second probability based on the living body detection results of the plurality of local areas through the confidence detection network.

12. The method of claim 11, wherein the training process of the living body detection model includes a first training phase and a second training phase, wherein,

the first training stage is configured to train the feature extraction network, the local biopsy network and the global biopsy network,

the second training stage is configured to train the confidence detection network.

13. The method of claim 12, wherein the first training phase comprises:

acquiring a sample biological feature image and labeling information corresponding to the sample biological feature image, wherein the labeling information comprises: a global living body labeling result and living body labeling results of a plurality of local areas;

inputting the sample biological feature image into the feature extraction network to obtain a sample feature map, inputting the sample feature map into the local living body detection network to obtain first living body detection results of a plurality of local areas, and inputting the sample feature map and the first living body detection results of the plurality of local areas into the global living body detection network to obtain first global living body detection results; and

And determining a first target loss based on the first global living detection result, the first living detection results of the plurality of local areas, the global living labeling result and the living labeling results of the plurality of local areas, and training the feature extraction network, the local living detection network and the global living detection network with the aim of minimizing the first target loss.

14. The method of claim 13, wherein,

the first training phase further comprises: performing disturbance processing on the sample biological feature image to obtain a disturbance biological feature image, inputting the disturbance biological feature image into the feature extraction network to obtain a disturbance feature map, and inputting the disturbance feature map into the local living body detection network to obtain a second living body detection result of a plurality of local areas; and

determining a first target loss based on the first global biopsy result, the first biopsy result of the plurality of local regions, the global biopsy result, and the biopsy result of the plurality of local regions, comprising:

the first target loss is determined based on the first global biopsy result, the first biopsy result of the plurality of local regions, the second biopsy result of the plurality of local regions, the global biopsy result, and the biopsy result of the plurality of local regions.

15. The method of claim 14, wherein determining the first target loss based on the first global biopsy result, the first biopsy result for the plurality of local regions, the second biopsy result for the plurality of local regions, the global biopsy result, and the biopsy result for the plurality of local regions comprises:

determining a first loss based on a difference between the first global living detection result and the global living annotation result;

determining a second loss based on a difference between a first living detection result of the plurality of local regions and a living labeling result of the plurality of local regions;

determining a third loss based on a difference between a first living detection result of the plurality of local areas and a second living detection result of the plurality of local areas; and

the first target loss is determined based on the first loss, the second loss, and the third loss.

16. The method of claim 12, wherein the second training phase comprises:

acquiring a sample biological feature image and labeling information corresponding to the sample biological feature image, wherein the labeling information comprises: a global living body labeling result and living body labeling results of a plurality of local areas;

Inputting the sample biological feature image into the feature extraction network to obtain a sample feature map, inputting the sample feature map into the local living network to obtain first living detection results of a plurality of local areas, inputting the first living detection results of the plurality of local areas into the confidence detection network to obtain detection confidence of the plurality of local areas, updating the corresponding local areas of the sample feature map based on the detection confidence of the plurality of local areas to obtain an updated feature map, inputting the updated feature map into the local living network to obtain third living detection results of the plurality of local areas, and inputting the updated feature map and the third living detection results of the plurality of local areas into the global living detection network to obtain second global living detection results; and

and determining a second target loss based on the second global living detection result, the third living detection result of the plurality of local areas, the global living labeling result and the living labeling results of the plurality of local areas, and training the confidence detection network by taking the second target loss as a training target.

17. The method of claim 1, wherein the re-living detection of the target user based on the actual input information of the target user, to obtain a target living detection result of the target user, comprises:

performing secondary living body detection on the target user based on the actual input information to obtain a secondary living body detection result; and

the target living body detection result is determined based on the preliminary living body detection result and the re-living body detection result.

18. The method of claim 1, wherein the biometric features comprise at least one of: at least one of a face, iris, fingerprint, and palmprint.

19. A living body detection system, characterized by comprising:

at least one storage medium storing at least one set of instructions for performing a living organism detection; and

at least one processor communicatively coupled to the at least one storage medium,

wherein the at least one processor reads the at least one instruction set and performs the in-vivo detection method of any one of claims 1 to 18 as directed by the at least one instruction set when the in-vivo detection system is running.

Images