A face recognition system, method, apparatus, electronic device and program product

Abstract

The present disclosure belongs to the technical field of face recognition, and particularly relates to a face recognition system, method, device, electronic device and computer program product. The method comprises: obtaining a similarity between each face sample in a base library and a face to be recognized based on each face recognition algorithm in a plurality of face recognition algorithms respectively, wherein the base library comprises a plurality of face samples of a preset number for each person; calculating a first similarity sum of each person in the base library corresponding to the face recognition algorithm; obtaining a weight of the face recognition algorithm, and determining a person corresponding to the face to be recognized based on the weight of the face recognition algorithm and the first similarity sum of each person. The present disclosure can reduce the false recognition rate of face recognition of face data collected in different application scenarios.

Description

Technical Field

[0001] This disclosure belongs to the field of facial recognition technology, specifically relating to a facial recognition system, method, device, electronic device, and computer program product. Background Technology

[0002] Facial recognition is a biometric technology that acquires a person's facial features and compares these features with pre-recorded facial data in a database to identify the person.

[0003] In a face recognition algorithm, the probability of identifying person A from the database based on A's facial features is called the recognition rate; the probability of identifying other people from the database who are not A based on A's facial features is called the false recognition rate.

[0004] Facial recognition algorithms can only determine the degree of similarity between two faces; determining whether two faces belong to the same person requires a business-level assessment. However, different facial recognition algorithms are trained on different business models for different application scenarios, and the pre-recorded facial data in the database is often obtained based on a specific application scenario. Therefore, each facial recognition algorithm has a high false recognition rate when recognizing faces obtained from other application scenarios. Summary of the Invention

[0005] This disclosure proposes a face recognition scheme to address the problem of high false recognition rates in existing face recognition schemes when using face data from different application scenarios.

[0006] A first aspect of this disclosure provides a face recognition system, including a database and a recognition module, wherein:

[0007] The database contains a preset number of face samples for each person;

[0008] The recognition module includes multiple face recognition algorithms;

[0009] The face recognition system uses each of the face recognition algorithms to obtain the similarity between each face sample in the database and the face to be recognized, and determines the person in the database corresponding to the face to be recognized based on the similarity and the weight of each of the face recognition algorithms.

[0010] In some embodiments, different facial samples of the same person are obtained by the person in different preset scenarios, wherein one facial sample corresponds to one preset scenario.

[0011] In some embodiments, determining the person in the database corresponding to the face to be identified based on the similarity and the weight of each of the face recognition algorithms includes:

[0012] Based on the similarity, calculate the sum of the first similarity scores for each person in the database corresponding to the face recognition algorithm;

[0013] The second similarity sum for each person is calculated based on the weights of the face recognition algorithm and the first similarity sum for each person.

[0014] The person with the highest sum of second similarity is the person corresponding to the face to be identified.

[0015] A second aspect of this disclosure provides a face recognition method, applied to the face recognition system described in the first aspect of this disclosure, comprising:

[0016] The similarity between each face sample in the database and the face to be identified is obtained based on each of the various face recognition algorithms. The database contains a preset number of face samples for each person.

[0017] Calculate the sum of the first similarity scores for each person in the database corresponding to the face recognition algorithm;

[0018] The weights of the face recognition algorithm are obtained, and the person corresponding to the face to be identified is determined based on the weights of the face recognition algorithm and the sum of the first similarity of each person.

[0019] In some embodiments, calculating the sum of the first similarity scores for each person in the database corresponding to the face recognition algorithm includes:

[0020] The similarity scores of each face sample of the person in the database obtained by the face recognition algorithm are summed to obtain the first similarity score of the person corresponding to the face recognition algorithm.

[0021] In some embodiments, determining the person corresponding to the face to be identified based on the weights of the face recognition algorithm and the sum of the first similarities for each person includes:

[0022] The second similarity of the person is obtained by adding the product of the weight of each face recognition algorithm and the first similarity of the person corresponding to the face recognition algorithm;

[0023] The person with the highest sum of second similarity is the person corresponding to the face to be identified.

[0024] In some embodiments, determining the person corresponding to the face to be identified based on the weights of the face recognition algorithm and the sum of the first similarities for each person includes:

[0025] Identify the top N individuals with the highest sum of first similarities corresponding to each of the aforementioned face recognition algorithms, where N is a natural number;

[0026] The candidate set is composed of the top N individuals whose sum of the first similarity scores corresponding to each of the aforementioned face recognition algorithms is used.

[0027] For each person in the candidate set, the weight of each face recognition algorithm and the product of the first similarity sum of the person corresponding to the face recognition algorithm are added together to obtain the second similarity sum of the person. The person with the highest second similarity sum is the person corresponding to the face to be identified.

[0028] A third aspect of this disclosure provides a face recognition device, applied to the face recognition system of claim 1, comprising:

[0029] The acquisition module is used to acquire the similarity between each face sample in the base database and the face to be identified based on each of the various face recognition algorithms. The base database contains a preset number of face samples for each person.

[0030] The calculation module is used to calculate the sum of the first similarity scores for each person in the database corresponding to the face recognition algorithm;

[0031] The determination module is used to obtain the weights of the face recognition algorithm and determine the person corresponding to the face to be identified based on the weights of the face recognition algorithm and the sum of the first similarity of each person.

[0032] A fourth aspect of this disclosure provides an electronic device, including a memory and a processor.

[0033] The memory is used to store computer programs;

[0034] The processor is configured to implement the method described in the second aspect of this disclosure when executing the computer program.

[0035] A fifth aspect of this disclosure provides a computer program product, including a computer program and instructions, which, when executed by a processor, implement the method described in the second aspect of this disclosure.

[0036] In summary, the face recognition systems, methods, devices, electronic devices, and computer program products provided in the embodiments of this disclosure can obtain face feature data from multiple application scenarios by pre-entering face samples of each person in a base database for different application scenarios. Then, by using multiple face recognition algorithms to recognize the face samples from the multiple application scenarios, it can cover more application scenarios compared to a single face recognition algorithm, thereby reducing the false recognition rate when performing face recognition on face data from different application scenarios. Attached Figure Description

[0037] The features and advantages of this disclosure will be more clearly understood by referring to the accompanying drawings, which are schematic and should not be construed as limiting the scope of this disclosure in any way.

[0038] Figure 1 This is a schematic diagram of a computer system to which this disclosure applies;

[0039] Figure 2 This is a flowchart illustrating a face recognition method according to some embodiments of the present disclosure;

[0040] Figure 3 This is a detailed flowchart illustrating a face recognition method according to some embodiments of the present disclosure;

[0041] Figure 4 A schematic diagram of a face recognition device shown according to some embodiments of the present disclosure;

[0042] Figure 5 This is a schematic diagram of an electronic device shown in some embodiments of this disclosure. Detailed Implementation

[0043] In the following detailed description, numerous specific details of this disclosure are set forth by way of example in order to provide a thorough understanding of the relevant disclosure. However, it will be apparent to those skilled in the art that this disclosure may be practiced without these details. It should be understood that the terms “system,” “apparatus,” “unit,” and / or “module” used in this disclosure are a method of distinguishing different parts, elements, sections, or components at different levels in a sequential arrangement. However, these terms may be replaced by other expressions if they can achieve the same purpose.

[0044] It should be understood that when a device, unit, or module is referred to as being "on," "connected to," or "coupled to" another device, unit, or module, it may be directly connected to or coupled to, or communicate with, other devices, units, or modules, or there may be intermediate devices, units, or modules present, unless the context explicitly indicates otherwise. For example, the term "and / or" as used in this disclosure includes any one and all combinations of one or more of the associated listed items.

[0045] The terminology used in this disclosure is for the purpose of describing particular embodiments only and is not intended to limit the scope of this disclosure. As shown in this specification and claims, unless the context clearly indicates otherwise, words such as "a," "an," "an," and / or "the" do not specifically refer to the singular and may include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified features, integrals, steps, operations, elements, and / or components, and such expressions do not constitute an exclusive list, in which other features, integrals, steps, operations, elements, and / or components may also be included.

[0046] Referring to the following description and accompanying drawings, these and other features and characteristics, operating methods, functions of related structural elements, combinations of parts, and economics of manufacture of this disclosure can be better understood, wherein the description and drawings form part of the specification. However, it is clearly understood that the drawings are for illustrative and descriptive purposes only and are not intended to limit the scope of protection of this disclosure. It is understood that the drawings are not drawn to scale.

[0047] Various structural diagrams are used in this disclosure to illustrate various variations of embodiments according to this disclosure. It should be understood that the preceding or following structures are not intended to limit this disclosure. The scope of protection of this disclosure is defined by the claims.

[0048] Figure 1 This is a schematic diagram of a computer system to which this disclosure applies. For example... Figure 1 The system shown has a face recognition server connected to an image acquisition device and a database network. The face recognition server acquires the face image of the person to be identified through the image acquisition device and compares the face image with the face images pre-recorded in the database. When the similarity exceeds a preset threshold, it is determined that the person to be identified is the person corresponding to the face image in the database.

[0049] in:

[0050] The image acquisition device can be a camera, a surveillance camera, or a camera built into other devices. Specifically, the image acquisition device can be an embedded camera in a face recognition device. The image acquisition device is oriented towards a preset face recognition area to acquire a facial image of the person to be identified located within the face recognition area.

[0051] The face recognition server can be any of a standalone, clustered, or distributed server, or it can be a smart device with computing capabilities, such as a mobile phone, personal digital assistant (PDA), or tablet computer. The face recognition server can be a standalone device or a network server connected to a network. Specifically, the face recognition server can be a face recognition software module deployed on the face recognition device.

[0052] The base database can be any of a single-machine, cluster, or distributed database. In particular, the base database can be a database deployed on the face recognition device.

[0053] This disclosure provides embodiments of a face recognition system. The hardware architecture of the face recognition system is as follows: Figure 1 The computer system shown includes a database, a face recognition server, and an image acquisition device. The database, the face recognition server, and the image acquisition device can be independent devices connected to a network, or they can be functional modules of a face recognition device. In the face recognition device, the face recognition server can be a recognition module deployed therein.

[0054] In the facial recognition system, the base database contains a preset number of facial samples for each person. In some embodiments, different facial samples are obtained by the person in different preset scenarios, wherein one facial sample corresponds to one preset scenario.

[0055] The recognition module includes a variety of face recognition algorithms.

[0056] The facial recognition system uses each of the aforementioned facial recognition algorithms to obtain the similarity between each facial sample in the database and the face to be identified. Then, based on the similarity scores, it calculates a first similarity sum for each person in the database corresponding to the facial recognition algorithm. Based on the weights of the facial recognition algorithms and the first similarity sum for each person, it calculates a second similarity sum for each person. The person with the highest second similarity sum is the person corresponding to the face to be identified.

[0057] Figure 2 This is a flowchart illustrating a face recognition method according to some embodiments of the present disclosure. In some embodiments, the face recognition method is... Figure 1 The face recognition server in the system shown executes the face recognition method, which includes the following steps:

[0058] S210, obtain the similarity between each face sample in the base database and the face to be identified based on each of the various face recognition algorithms, wherein the base database contains a preset number of face samples for each person.

[0059] Facial recognition extracts the facial features of the person to be identified using facial recognition algorithms and compares them with the features of faces pre-entered in a database. If the similarity is greater than a preset threshold, the person to be identified is considered to be the person corresponding to the face in the database.

[0060] This application's database contains pre-recorded facial photos of the same person in different scenarios, such as wearing glasses, not wearing glasses, wearing makeup, and not wearing makeup. Only one qualified facial photo is recorded for each scenario.

[0061] This application's recognition module incorporates multiple face recognition algorithms, such as SenseTime's algorithm and ArcSoft's algorithm. Each algorithm is trained on different datasets, covering various business scenarios. The system assigns default recognition weights to each algorithm.

[0062] In this step, the face recognition system uses each face recognition algorithm to compare the features of the person to be identified with each face data in the database and calculates the similarity.

[0063] S220, calculate the sum of the first similarity scores for each person in the database corresponding to the face recognition algorithm.

[0064] Specifically, after obtaining the similarity between each face data in the base database and the face to be identified, the similarity of all face data belonging to the same person is added together to obtain the first similarity sum of the person.

[0065] S230, obtain the weight of the face recognition algorithm, and determine the person corresponding to the face to be identified based on the weight of the face recognition algorithm and the sum of the first similarity of each person.

[0066] Specifically, at the start of each face recognition operation, the operator can input the weight of each face recognition algorithm through the system configuration page. If no weight is input, the system default weights will be used.

[0067] In some embodiments of this disclosure, the weight of each face recognition algorithm is multiplied by the product of the first similarity sum of the person corresponding to the face recognition algorithm to obtain the second similarity sum of the person, and the person with the highest second similarity sum is the person corresponding to the face to be identified.

[0068] Since the facial data of most people in the database has a very low similarity to the face to be identified, they are obviously not the people to be identified. Therefore, in order to reduce the amount of computation, these people can be directly excluded.

[0069] Therefore, in some other embodiments of this disclosure, the top 5 individuals with the highest sum of first similarity to each of the face recognition algorithms are determined, and a candidate set is formed from the top 5 individuals with the highest sum of first similarity to each of the face recognition algorithms. For each individual in the candidate set, the weight of each face recognition algorithm and the product of the first sum of first similarity to the individual corresponding to the face recognition algorithm are added together to obtain the second sum of similarity to the individual. The individual with the highest second sum of similarity is the individual corresponding to the face to be identified.

[0070] Figure 3 This is a detailed flowchart illustrating a face recognition method according to an embodiment of the present disclosure.

[0071] like Figure 3 The method shown is optimized by combining a multi-base-database approach and a multi-algorithm approach to reduce false identification:

[0072] (1) The business requires each person to input a fixed number of faces, such as 3 or 5 faces. Each face must be the person in the photos, but they can be slightly different, such as wearing glasses, not wearing glasses, wearing makeup, or not wearing makeup. The business uses a sum of similarity calculations. For example, among the top 5 faces with the highest similarity, the face with the highest sum of similarity for the same person ID is considered the person in the photos.

[0073] (2) Select multiple reliable algorithms in business, such as SenseTime and ArcSoft. Then, SenseTime algorithm accounts for 70% weight and ArcSoft algorithm accounts for 30% weight. When performing face recognition, the two algorithms are used to extract feature values ​​and compare them 1:N. When calculating the final similarity corresponding to a certain person ID, it is SenseTime similarity x 0.7 + ArcSoft similarity x 0.3.

[0074] (3) The calculation method after combining the two schemes is as follows: assuming the person ID is a, the total similarity of person a is obtained by SenseTime's algorithm as A, and the total similarity of person a is obtained by ArcSoft's algorithm as B. The final total similarity is C = A x 0.7 + B x 0.3. If C is the person with the largest total similarity among the appearing persons, then a can be identified as that person.

[0075] The multi-database approach encompasses multiple characteristics of an individual, providing a more accurate representation compared to a single-database approach. The multi-algorithm approach, with each algorithm having its own trained business model, can cover a wider range of data models. Combining these two approaches results in higher recognition reliability and a lower false recognition rate.

[0076] Figure 4 This is a schematic diagram of a face recognition device according to some embodiments of the present disclosure. Figure 4As shown, the face recognition device 400 includes an acquisition module 410, a calculation module 420, and a determination module 430. The device's interactive functions can be provided by... Figure 1 The face recognition server in the system shown performs the following:

[0077] The acquisition module 410 is used to acquire the similarity between each face sample in the base database and the face to be identified based on each of the multiple face recognition algorithms. The base database contains a preset number of face samples for each person.

[0078] Calculation module 420 is used to calculate the sum of the first similarity of each person in the base database corresponding to the face recognition algorithm;

[0079] The determination module 430 is used to obtain the weight of the face recognition algorithm and determine the person corresponding to the face to be identified based on the weight of the face recognition algorithm and the sum of the first similarity of each person.

[0080] One embodiment of this disclosure provides an electronic device. For example... Figure 5 As shown, the electronic device 500 includes a memory 520 and a processor 510. The memory 520 is used to store computer programs; the processor 510 is used to implement, when executing the computer program, [the following is unclear and requires context: "to implement"]. Figure 2 The method described in S210-S230.

[0081] One embodiment of this disclosure provides a computer program product, including a computer program and instructions, which, when executed by a processor, implement... Figure 2 The method described in S210-S230.

[0082] In summary, the face recognition systems, methods, devices, electronic devices, and computer program products provided in the embodiments of this disclosure can obtain face feature data from multiple application scenarios by pre-entering face samples of each person in a base database for different application scenarios. Then, by using multiple face recognition algorithms to recognize the face samples from the multiple application scenarios, it can cover more application scenarios compared to a single face recognition algorithm, thereby reducing the false recognition rate when performing face recognition on face data from different application scenarios.

[0083] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the devices and modules described above can be referred to the corresponding descriptions in the foregoing device embodiments, and will not be repeated here.

[0084] Although the subject matter described herein is provided in the general context of execution on a computer system in conjunction with an operating system and applications, those skilled in the art will recognize that other implementations can also be executed in conjunction with other types of program modules. Generally, program modules include routines, programs, components, data structures, and other types of structures that perform specific tasks or implement specific abstract data types. Those skilled in the art will understand that the subject matter described herein can be practiced using other computer system configurations, including handheld devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, minicomputers, mainframes, etc., and can also be used in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In a distributed computing environment, program modules may reside on both local and remote memory storage devices.

[0085] Those skilled in the art will recognize that the units and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0086] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of this disclosure and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of this disclosure should be included within the protection scope of this disclosure. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A face recognition system, comprising a base database and a recognition module, wherein: The database contains a preset number of face samples for each person; The recognition module includes multiple face recognition algorithms; The face recognition system uses each of the face recognition algorithms to obtain the similarity between each face sample in the database and the face to be recognized, and determines the person in the database corresponding to the face to be recognized based on the similarity and the weight of each of the face recognition algorithms. The step of determining the person in the database corresponding to the face to be identified based on the similarity and the weight of each of the face recognition algorithms includes: Based on the similarity, calculate the sum of the first similarity scores for each person in the database corresponding to the face recognition algorithm; The second similarity sum for each person is calculated based on the weights of the face recognition algorithm and the first similarity sum for each person; The person with the highest sum of second similarity is the person corresponding to the face to be identified.

2. The system according to claim 1, characterized in that: Different facial samples of the same person are obtained by the person in different preset scenarios, wherein one facial sample corresponds to one preset scenario.

3. A face recognition method, applied to the face recognition system of claim 1, characterized in that, include: The similarity between each face sample in the database and the face to be identified is obtained based on each of the various face recognition algorithms. The database contains a preset number of face samples for each person. Calculate the sum of the first similarity scores for each person in the database corresponding to the face recognition algorithm; Obtain the weights of the face recognition algorithm, and determine the person corresponding to the face to be identified based on the weights of the face recognition algorithm and the sum of the first similarity of each person. The calculation of the first similarity sum of each person in the database corresponding to the face recognition algorithm includes: adding the similarity of each face sample of the person in the database obtained by the face recognition algorithm to obtain the first similarity sum of the person corresponding to the face recognition algorithm; The step of determining the person corresponding to the face to be identified based on the weights of the face recognition algorithm and the sum of the first similarities of each person includes: adding the products of the weights of each face recognition algorithm and the sum of the first similarities of the people corresponding to the face recognition algorithm to obtain the second similarity sum of the people; the person with the highest second similarity sum is the person corresponding to the face to be identified.

4. The method according to claim 3, characterized in that, The process of determining the person corresponding to the face to be identified based on the weights of the face recognition algorithm and the sum of the first similarities of each person includes: Identify the top N individuals with the first sum of similarities corresponding to each of the aforementioned face recognition algorithms, where N is a natural number; The candidate set is composed of the top N individuals whose sum of the first similarity scores corresponding to each of the aforementioned face recognition algorithms. For each person in the candidate set, the weight of each face recognition algorithm and the product of the first similarity sum of the person corresponding to the face recognition algorithm are added together to obtain the second similarity sum of the person. The person with the highest second similarity sum is the person corresponding to the face to be identified.

5. A face recognition device, applied to the face recognition system of claim 1, characterized in that, include: The acquisition module is used to acquire the similarity between each face sample in the base database and the face to be identified based on each of the various face recognition algorithms. The base database contains a preset number of face samples for each person. The calculation module is used to calculate the sum of the first similarity scores for each person in the database corresponding to the face recognition algorithm; The determination module is used to obtain the weights of the face recognition algorithm and determine the person corresponding to the face to be identified based on the weights of the face recognition algorithm and the sum of the first similarity of each person. The calculation module is specifically used to: add the similarity of each face sample of the person in the base database obtained by the face recognition algorithm to obtain the first similarity sum of the person corresponding to the face recognition algorithm; The determining module is specifically used to: add the product of the weight of each face recognition algorithm and the first similarity sum of the person corresponding to the face recognition algorithm to obtain the second similarity sum of the person; the person with the highest second similarity sum is the person corresponding to the face to be identified.

6. An electronic device, comprising a memory and a processor, The memory is used to store computer programs; The processor is configured to implement the method of claim 3 or 4 when executing the computer program.

7. A computer program product comprising a computer program and instructions, wherein when the computer program and instructions are executed by a processor, the method of claim 3 or 4 is implemented.

Images