Method, device and server for improving face recognition accuracy

Abstract

The present application discloses a method, device and server for improving the accuracy of face recognition. The method includes: obtaining the facial information of the person currently clocking in, and obtaining candidate facial information based on the facial information; determining the probability of clocking in at the current terminal device, the probability of clocking in during the current time period and the probability of correct facial recognition for each candidate facial information based on the historical database; inputting the probability into the Bayesian model to obtain the facial matching accuracy of each candidate; and determining the facial information with the highest accuracy as the target facial information. The present application can make full use of the existing employee attendance clocking-in behavior, match the people who may clock in to specific devices, and in the process of predictive analysis, use the data of the most recent period of time for the user's behavior, and also dynamically count the identified terminal devices, with the function of dynamically updating the data prediction results, thereby improving the accuracy of face recognition and reducing costs.

Description

Technical Field

[0001] The present application relates to the field of face recognition technology, and in particular to a method, device, and server for improving face recognition accuracy. Background Art

[0002] Existing attendance machines using facial recognition technology for industrial production typically pre-load facial database images, extract facial features, and create facial profiles. The attendance machine acquires facial images in real time, compares them with the facial profile, and uses the comparison results as the facial recognition results. However, in actual use, existing technologies have the following problems:

[0003] (1) In existing factories, the number of face databases is relatively large. The larger the database, the more serious the reduction in recognition accuracy.

[0004] (2) In a group, multiple factories or departments share a facial recognition attendance system. In addition to the problem of a large number of databases, the quality of facial database images uploaded by different departments will also be different. Poor quality images will also affect the accuracy of facial comparison. Summary of the Invention

[0005] The purpose of the embodiments of the present application is to provide a method, device and server for improving the accuracy of face recognition, so as to solve the problem in the prior art that the number of face databases is large and the quality of uploaded face database images is different, resulting in low accuracy of face comparison.

[0006] To achieve the above objectives, the present application provides, in a first aspect, a method for improving face recognition accuracy, comprising:

[0007] Get the facial information of the person currently clocking in;

[0008] Obtain candidate facial information based on facial information;

[0009] Determine the probability of each candidate's facial information being clocked in at the current terminal device, the probability of being clocked in during the current time period, and the probability of being correctly recognized based on the historical database;

[0010] The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition are input into the Bayesian model to obtain the facial matching accuracy of each candidate;

[0011] The face with the highest matching accuracy is determined as the target face information.

[0012] In the embodiment of the present application, obtaining candidate facial information based on facial information includes:

[0013] The face matching algorithm is used to obtain the candidate face information that is most similar to the current check-in face information.

[0014] In the embodiment of the present application, determining the probability of clocking in at the current terminal device and the probability of clocking in during the current time period corresponding to each candidate facial information based on the historical database includes:

[0015] Obtain device point data and time data collected by terminal devices in the historical database;

[0016] Determine the probability value of each candidate face information checking in at the current terminal device based on the device point data;

[0017] The probability value of each candidate face information checking in in the current time period is determined based on the time data in the historical database.

[0018] In the embodiment of the present application, determining the probability value of each candidate face information checking in at the current terminal device based on the device point data includes:

[0019] Get the code of the current terminal device;

[0020] According to the code, count the number of times each candidate's facial information is checked in on the current terminal device;

[0021] The probability value of each candidate facial information checking in at the current terminal device is determined according to the number of times each candidate facial information checks in at the current terminal device.

[0022] In the embodiment of the present application, determining the probability of correct face recognition for each candidate face information includes:

[0023] Get the face information entered in the face database;

[0024] Obtain facial information entered by the terminal device;

[0025] The probability of correct face recognition for each candidate face information is calculated based on the face information entered in the face database and the face information entered by the terminal device.

[0026] In the embodiment of the present application, the probability of correct face recognition is expressed by the face similarity obtained by the face matching algorithm.

[0027] In the embodiment of the present application, the Bayesian model satisfies formula (1):

[0028] = ; (1)

[0029] in, The probability of successful matching when a person's facial information is in a certain time period and on a certain terminal device and the similarity of the matching face is obtained; The probability of a person's facial information appearing on a person's facial recognition terminal device; The probability of a person's face appearing in a certain period of time; The probability that a person's face information is correctly recognized by the face; The prior probability of using a face recognition service for someone’s face information; The probability value of each face information checking in on the terminal device; The probability value of each facial information's attendance punch-in in the time period; Face matching similarity.

[0030] In an embodiment of the present application, the method further includes:

[0031] Update the historical database based on the target face information, the corresponding terminal device and time.

[0032] A second aspect of the present application provides a server, including:

[0033] a memory configured to store instructions; and

[0034] The processor is configured to call instructions from a memory and implement the method for improving face recognition accuracy when executing the instructions.

[0035] A third aspect of the present application provides a device for improving face recognition accuracy, comprising:

[0036] multiple terminal devices; and

[0037] Based on the above server.

[0038] A fourth aspect of the present application provides a machine-readable storage medium having stored thereon instructions for enabling a machine to execute the above-mentioned method for improving the accuracy of face recognition.

[0039] Through the above technical solution, the facial information of the person who is currently clocking in is obtained, and the candidate facial information is obtained based on the facial information. The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition corresponding to each candidate facial information are determined respectively based on the historical database. The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition are input into the Bayesian model to obtain the facial matching accuracy of each candidate. The facial information with the highest accuracy is determined as the target facial information. This application can make full use of the existing employee attendance clocking-in behavior, match people who may clock in to specific devices, and in the process of predictive analysis, the user's behavior uses data from the most recent period, and also dynamically counts the identified terminal devices, with the function of dynamically updating data prediction results, thereby improving the accuracy of facial recognition and reducing costs.

[0040] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following detailed description, they are used to explain the embodiments of the present application but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings:

[0042] Figure 1 A flowchart of a method for improving face recognition accuracy according to an embodiment of the present application is schematically shown;

[0043] Figure 2 The following schematically shows a structural block diagram of a server according to an embodiment of the present application;

[0044] Figure 3 The figure schematically shows a structural diagram of a device for improving face recognition accuracy according to an embodiment of the present application.

[0045] Description of Reference Numerals

[0046] DETAILED DESCRIPTION

[0047] To make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0048] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back, etc.), such directional indications are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0049] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the fact that they can be implemented by ordinary technicians in this field. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0050] Figure 1 The following schematically shows a flow chart of a method for improving face recognition accuracy according to an embodiment of the present application. Figure 1 As shown, an embodiment of the present application provides a method for improving face recognition accuracy, which may include the following steps:

[0051] Step 101: Obtain the facial information of the person currently checking in;

[0052] Step 102: Obtain candidate facial information based on the facial information;

[0053] Step 103: Determine, based on the historical database, the probability of clocking in at the current terminal device, the probability of clocking in during the current time period, and the probability of correct facial recognition for each candidate facial information;

[0054] Step 104: Input the probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition into the Bayesian model to obtain the facial matching accuracy of each candidate;

[0055] Step 105: Determine the face with the highest matching accuracy as the target face information.

[0056] In the embodiment of the present application, the attendance machine using facial recognition technology currently used in industrial production usually records facial database images in advance, extracts facial features, and establishes facial files. The attendance machine obtains facial images in real time, compares the facial images with the facial files, and uses the comparison results as the results of facial recognition. However, due to the large number of facial databases, the recognition accuracy is low. At the same time, the quality of images uploaded to the facial database by different departments is different, and images of poor quality will also affect the accuracy of facial comparison. Therefore, the embodiment of the present application comprehensively utilizes the existing departmental organizational data, employee personal behavior data, and facial comparison result data to improve the accuracy of facial recognition.

[0057] In an embodiment of the present application, a server is provided for communicating with multiple terminal devices. The terminal devices may be facial recognition attendance machines installed at multiple locations. The server may include a facial database and a historical database. The facial database stores facial images requiring facial recognition; the historical database may be established based on the time information and terminal device information of historical clock-in faces uploaded by multiple terminal devices.

[0058] When a terminal device detects a clock-in, it first obtains the facial information of the person clocking in and transmits it to the server. After receiving the facial information from the terminal, the server identifies candidate facial information from the facial database based on the acquired facial information. Candidate facial information refers to multiple facial information in the facial database that has a high degree of similarity to the facial information currently clocking in. The number of candidate facial information can be set based on actual circumstances. For example, five highly similar candidate facial information can be set for each facial information. The server then determines the probability of clocking in at the current terminal device, the probability of clocking in during the current time period, and the probability of correct facial recognition for each candidate facial information based on the historical database. These probabilities of clocking in at the current terminal device, the probability of clocking in during the current time period, and the probability of correct facial recognition are input into a Bayesian model to determine the facial matching accuracy for each candidate. A Bayesian model is a prediction method based on Bayesian statistics. It not only utilizes model and data information but also leverages prior information. An empirical analysis compares the Bayesian model with a conventional regression prediction model, demonstrating its significant superiority. After determining the face matching result of each candidate, the server can sort the face matching accuracy of each candidate, determine the candidate face information with the highest face matching accuracy as the target face information, and push the face information to the terminal device.

[0059] The embodiment of the present application obtains the facial information of the current clock-in, and obtains candidate facial information based on the facial information. The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition corresponding to each candidate facial information are determined based on the historical database. The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition are input into the Bayesian model to obtain the facial matching accuracy of each candidate. The facial information with the highest facial matching accuracy is determined as the target facial information. The embodiment of the present application can make full use of the existing employee attendance clock-in behavior, match the personnel who may clock in to a specific device, and in the process of predictive analysis, the user's behavior uses the data of the most recent period, and also dynamically counts the identified terminal devices, with the function of dynamically updating the data prediction results, thereby improving the accuracy of facial recognition and reducing costs.

[0060] In an embodiment of the present application, obtaining candidate facial information based on facial information may include:

[0061] The face matching algorithm is used to obtain the candidate face information that is most similar to the current check-in face information.

[0062] Specifically, since there is a lot of facial information in the facial database, in order to improve recognition efficiency, the facial information of the current clock-in can be matched in the facial database. After matching, the facial information most similar to the current clock-in facial information is obtained, and the candidate facial information is specified in the most similar facial information. The candidate facial information can be represented by a work number or a code. Among them, the number of candidate facial information can be set according to actual conditions. In the actual process of calculating faces, the closest distance d is generally the most similar. Since normalization has been performed during calculation, c=1-d can be used as a similarity judgment. In an example, there are facial information of 20 people in the facial database, and the 5 facial information most similar to the current facial information can be matched as candidate facial information.

[0063] In an embodiment of the present application, determining the probability of clocking in at the current terminal device and the probability of clocking in during the current time period corresponding to each candidate facial information based on the historical database may include:

[0064] Obtain device point data and time data collected by terminal devices in the historical database;

[0065] Determine the probability value of each candidate face information checking in at the current terminal device based on the device point data;

[0066] The probability value of each candidate face information checking in in the current time period is determined based on the time data in the historical database.

[0067] Specifically, the server can leverage existing departmental organizational data, employee behavior data, and facial comparison results to predict and analyze user clock-in behavior, improving recognition efficiency. The server can first obtain the device point and time data for the terminal device, i.e., the facial recognition attendance machine, from a historical database. By analyzing the device point data in the historical database, the server can determine the probability of each candidate face being clocked in at the current facial recognition attendance machine. By analyzing the time data in the historical database, the server can also determine the probability of each candidate face being clocked in during the current time period.

[0068] In an embodiment of the present application, determining the probability value of each candidate face information checking in at the current terminal device based on the device point data may include:

[0069] Get the code of the current terminal device;

[0070] According to the code, count the number of times each candidate's facial information is checked in on the current terminal device;

[0071] The probability value of each candidate facial information checking in at the current terminal device is determined according to the number of times each candidate facial information checks in at the current terminal device.

[0072] Specifically, according to the department where the facial recognition attendance machine is located and the range of faces that need to be recognized, the number of devices that have been logged in by each face in the facial file is counted. Then, according to the device code of the attendance machine that needs to be transmitted to the server when the attendance machine accesses the facial recognition server, the facial recognition attendance machines that employees frequently log in to for attendance are counted separately. Based on the number of times employees log in and out on each facial recognition attendance machine and the total number of times they punch in and out, the probability value of an employee punching in and out on a certain terminal device can be obtained. For example, an employee clocks in and out for 40 workdays over two months. This includes 30 valid clock-ins on terminal a1, 6 valid clock-ins on terminal a2, 4 valid clock-ins on terminal a3, and 0 valid clock-ins on terminal a4. (Due to the scenario, there is no minimum probability of 0. If a terminal has never clocked in, the minimum number of clock-ins for that terminal is set. This can be set based on the actual business scenario. In this case, the employee has never clocked in on terminal a4, but the default value is 4.) Therefore, the probabilities of the employee clocking in and out on terminals a1, a2, a3, and a4 are 30 / 40, 6 / 40, 4 / 40, and 4 / 40, respectively.

[0073] In the embodiment of the present application, determining the probability of correct face recognition for each candidate face information may include:

[0074] Get the face information entered in the face database;

[0075] Obtain facial information entered by the terminal device;

[0076] Calculate the probability of correct face recognition for each candidate face information based on the face information entered in the face database and the face information entered by the terminal device

[0077] Specifically, the server first retrieves the facial information of each candidate pre-stored in the facial database, then retrieves the facial information of the person clocking in at the current terminal. Based on the facial information of each candidate pre-stored in the facial database and the facial information of the person clocking in at the current terminal, it calculates the probability that each candidate's facial information matches the facial information of the person clocking in. For example, there are 20 people in the facial database. Within this database, the five most similar facial information to the currently clocked-in face are matched (the closest facial encoding distance can also be understood as the most similar). In actual face recognition, the closest distance d is generally considered the most similar. Since normalization is already performed during the calculation, c = 1 - d can be used here to determine similarity. Therefore, the probabilities of the currently clocked-in face information matching the five candidate facial information are c1, c2, c3, c4, and c5, respectively.

[0078] In the embodiment of the present application, the probability of correct face recognition is expressed by the face similarity obtained by the face matching algorithm.

[0079] Specifically, a face matching algorithm is pre-stored in the server. When face recognition is required, the face similarity of each candidate face information can be obtained through the face matching algorithm. The face similarity is the probability that the candidate face information is correctly recognized.

[0080] In the embodiment of the present application, the Bayesian model satisfies formula (1):

[0081] = ; (1)

[0082] in, The probability of successful matching when a person's facial information is in a certain time period and on a certain terminal device and the similarity of the matching face is obtained; The probability of a person's facial information appearing on a person's facial recognition terminal device; The probability of a person's face appearing in a certain period of time; The probability that a person's face information is correctly recognized by the face; The prior probability of using a face recognition service for someone’s face information; The probability value of each face information checking in on the terminal device; The probability value of each facial information's attendance punch-in in the time period; Face matching similarity.

[0083] Specifically, for all personnel in the enterprise, the face of a person on a certain device, at a certain time, and that may be matched is the same, so and are all equal, only considering Then weigh , and found that it is only related to the conditional probability of device, time, and matching degree. Therefore, by comparing different faces The size of , determines the real face of the predicted recognition. Through maximum likelihood estimation: The large advantage can be understood as the image that is most similar to the image provided by the terminal has a greater advantage.

[0084] In an embodiment of the present application, the method may further include:

[0085] Update the historical database based on the target face information, the corresponding terminal device and time.

[0086] Specifically, in practical applications, for larger companies and enterprises, they can make full use of the existing employee attendance clocking-in behavior and match people who may clock in to specific devices to improve recognition accuracy. In the process of predictive analysis, the employee's behavior uses data from the most recent period, and also dynamically counts the identified terminal devices to update the historical database, so that the device for improving the accuracy of face recognition in the embodiment of the present application has the function of dynamically updating data prediction results. The frequency of statistics can be set according to actual conditions. In one example, the server iterates the device point information and time information in the historical database of the previous two months every day, and updates the employee photos in the face database every six months.

[0087] In one specific embodiment, the server first collects statistics on terminal devices: Employee A frequently clocks in and out on terminal device a1 (terminal device a1 is Employee A's favorite time clock). For example, if the timeframe is set to two months, with 40 workdays of full attendance, A clocks in and out 30 times on terminal device a1, 6 times on terminal device a2, 4 times on terminal device a3, and 0 times on terminal device a4. (Depending on the scenario, there is no minimum probability of 0. If a terminal device has never been clocked in, the number of times that a terminal has clocked in is set to the minimum number of times. This can be set based on the actual business scenario. In this case, a person has never clocked in on terminal device a4, but the default value is 4 times.) Therefore, the probabilities of Employee A clocking in and out on terminals a1, a2, a3, and a4 are 30 / 40, 6 / 40, 4 / 40, and 4 / 40, respectively.

[0088] The server calculates time statistics: Assuming the normal working hour is 8:00 AM, the clock-in time is divided into 10-minute intervals before work (the specific division can be based on statistics). For example, the statistical interval before 7:30 is t1, 7:30-7:40 is t2, 7:40-7:50 is t3, 7:50-8:00 is t5, and after 8:00 is t6. Assuming that the employee clocks in and out of the five time periods 5, 20, 10, 4, and 1 over 40 perfect workdays, the corresponding probabilities for time periods t1-5 are: 5 / 40, 20 / 40, 10 / 40, 4 / 40, and 1 / 40.

[0089] The server calculates face matching similarity: Assuming there are 20 people in the face database (this is actually a larger set, just an example), to improve computational efficiency, the five faces most similar to the currently checked-in face are matched from the database of 20: A, B, C, D, and E (the closest face encoding distance can also be considered the most similar). In actual face matching, the closest distance d is generally considered the most similar. Since normalization is performed during the calculation, c = 1 - d is used as the similarity criterion. The facial similarity calculated for the five people is c1, c2, c3, c4, and c5 respectively.

[0090] In face recognition, the probability values ​​for each of the five individuals, A, B, C, D, and E, are calculated as follows: Assuming A clocks in at 7:46 on terminal device a1, the probability converted using the Bayesian formula is P1 = 30 / 40 * 10 / 40 * c1. The probability of identifying B is P2 = (conditional probability of B on that terminal device) * (conditional probability of B in that time period) * c2, and so on for C, D, and E. The highest calculated result is determined as the recognition result.

[0091] The embodiment of the present application obtains the facial information of the current clock-in, and obtains candidate facial information based on the facial information. The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition corresponding to each candidate facial information are determined based on the historical database. The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition are input into the Bayesian model to obtain the facial matching accuracy of each candidate. The facial information with the highest facial matching accuracy is determined as the target facial information. The embodiment of the present application can make full use of the existing employee attendance clock-in behavior, match the personnel who may clock in to a specific device, and in the process of predictive analysis, the user's behavior uses the data of the most recent period, and also dynamically counts the identified terminal devices, with the function of dynamically updating the data prediction results, thereby improving the accuracy of facial recognition and reducing costs.

[0092] Figure 2The following schematically shows a structural block diagram of a server according to an embodiment of the present application. Figure 2 As shown, an embodiment of the present application provides a server, which may include:

[0093] Memory 210 configured to store instructions; and

[0094] The processor 220 is configured to call instructions from the memory 210 and implement the above-mentioned method for improving face recognition accuracy when executing the instructions.

[0095] Specifically, in the embodiment of the present application, the processor 220 may be configured to:

[0096] Get the facial information of the person currently clocking in;

[0097] Obtain candidate facial information based on facial information;

[0098] Determine the probability of each candidate's facial information being clocked in at the current terminal device, the probability of being clocked in during the current time period, and the probability of being correctly recognized based on the historical database;

[0099] The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition are input into the Bayesian model to obtain the facial matching accuracy of each candidate;

[0100] The face with the highest matching accuracy is determined as the target face information.

[0101] Furthermore, the processor 220 may be further configured to:

[0102] The face matching algorithm is used to obtain the candidate face information that is most similar to the current check-in face information.

[0103] Furthermore, the processor 220 may be further configured to:

[0104] Obtain device point data and time data collected by terminal devices in the historical database;

[0105] Determine the probability value of each candidate face information checking in at the current terminal device based on the device point data;

[0106] The probability value of each candidate face information checking in in the current time period is determined based on the time data in the historical database.

[0107] Furthermore, the processor 220 may be further configured to:

[0108] Get the code of the current terminal device;

[0109] According to the code, count the number of times each candidate's facial information is checked in on the current terminal device;

[0110] The probability value of each candidate facial information checking in at the current terminal device is determined according to the number of times each candidate facial information checks in at the current terminal device.

[0111] Furthermore, the processor 220 may be further configured to:

[0112] Get the face information entered in the face database;

[0113] Obtain facial information entered by the terminal device;

[0114] The probability of correct face recognition for each candidate face information is calculated based on the face information entered in the face database and the face information entered by the terminal device.

[0115] In the embodiment of the present application, the probability of correct face recognition is expressed by the face similarity obtained by the face matching algorithm.

[0116] In the embodiment of the present application, the Bayesian model satisfies formula (1):

[0117] = ; (1)

[0118] in, The probability of successful matching when a person's facial information is in a certain time period and on a certain terminal device and the similarity of the matching face is obtained; The probability of a person's facial information appearing on a person's facial recognition terminal device; The probability of a person's face appearing in a certain period of time; The probability that a person's face information is correctly recognized by the face; The prior probability of using a face recognition service for someone’s face information; The probability value of each face information checking in on the terminal device; The probability value of each facial information's attendance punch-in in the time period; Face matching similarity.

[0119] Furthermore, the processor 220 may be further configured to:

[0120] Update the historical database based on the target face information, the corresponding terminal device and time.

[0121] Through the above technical solution, the facial information of the person who is currently clocking in is obtained, and the candidate facial information is obtained based on the facial information. The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition corresponding to each candidate facial information are determined respectively based on the historical database. The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition are input into the Bayesian model to obtain the facial matching accuracy of each candidate. The facial information with the highest accuracy is determined as the target facial information. This application can make full use of the existing employee attendance clocking-in behavior, match people who may clock in to specific devices, and in the process of predictive analysis, the user's behavior uses data from the most recent period, and also dynamically counts the identified terminal devices, with the function of dynamically updating data prediction results, thereby improving the accuracy of facial recognition and reducing costs.

[0122] Figure 3 The structure of a device for improving face recognition accuracy according to an embodiment of the present application is schematically shown. Figure 3 As shown, the embodiment of the present application further provides a device for improving face recognition accuracy, which may include:

[0123] multiple terminal devices 310; and

[0124] According to the server 320 described above.

[0125] Specifically, the terminal device obtains the facial information of the person currently clocking in, and the server 320 obtains candidate facial information based on the facial information, and then determines the probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition for each candidate facial information based on the historical database. The probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition are input into the Bayesian model to obtain the facial matching accuracy of each candidate. Finally, the facial matching with the highest accuracy is determined as the target facial information and pushed to the terminal device. This application can make full use of the existing employee attendance clocking-in behavior, match people who may clock in to specific devices, and in the process of predictive analysis, the user's behavior uses data from the most recent period, and also dynamically counts the identified terminal devices, with the function of dynamically updating data prediction results, thereby improving the accuracy of facial recognition and reducing costs.

[0126] An embodiment of the present application further provides a machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the above-mentioned method for controlling an arm.

[0127] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0128] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0129] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0130] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0131] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0132] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0133] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology for information storage. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0134] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0135] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A method for improving face recognition accuracy, characterized in that: The method comprises: Get the facial information of the person currently clocking in; Acquire candidate facial information according to the facial information; Determine, based on the historical database, the probability of clocking in at the current terminal device, the probability of clocking in during the current time period, and the probability of correct facial recognition for each candidate facial information. The probability of correct facial recognition is expressed by the facial similarity obtained by the face matching algorithm. Input the probability of clocking in at the current terminal device, the probability of clocking in at the current time period, and the probability of correct facial recognition into a Bayesian model to obtain the facial matching accuracy of each candidate; The face with the highest matching accuracy is determined as the target face information; The Bayesian model satisfies formula (1): = ；（1） in, The probability of successful matching when a person's facial information is in a certain time period and on a certain terminal device and the similarity of the matching face is obtained; The probability of a person's facial information appearing on a person's facial recognition terminal device; The probability of a person's face appearing in a certain period of time; The probability that a person's face information is correctly recognized by the face; The prior probability of using a face recognition service for someone’s face information; The probability value of each face information checking in on the terminal device; The probability value of each facial information's attendance punch-in in the time period; Face matching similarity.

2. The method according to claim 1, characterized in that The acquiring of candidate facial information according to the facial information includes: The face matching algorithm is used to obtain the candidate face information that is most similar to the current check-in face information.

3. The method according to claim 1, characterized in that Determining the probability of clocking in at the current terminal device and the probability of clocking in at the current time period corresponding to each candidate facial information based on the historical database includes: Obtaining device point data and time data collected by the terminal device in the historical database; Determine the probability value of each candidate facial information checking in at the current terminal device based on the device point data; The probability value of each candidate facial information checking in during the current time period is determined based on the time data in the historical database.

4. The method according to claim 3, characterized in that Determining the probability value of each candidate face information checking in at the current terminal device according to the device point data includes: Obtaining the code of the current terminal device; According to the code, counting the number of times each candidate facial information checks in on the current terminal device; The probability value of each candidate facial information checking in at the current terminal device is determined according to the number of times each candidate facial information checks in at the current terminal device.

5. The method according to claim 1, wherein Determining the probability of correct face recognition for each candidate face information includes: Get the face information entered in the face database; Obtaining facial information entered by the terminal device; The probability of correct face recognition corresponding to each candidate face information is calculated based on the face information recorded in the face database and the face information recorded by the terminal device.

6. The method according to claim 1, wherein The method further comprises: Update the historical database based on the target face information, the corresponding terminal device and time.

7. A server, characterized in that: include: a memory configured to store instructions; as well as A processor is configured to call the instructions from the memory and implement the method for improving face recognition accuracy according to any one of claims 1 to 6 when executing the instructions.

8. A device for improving face recognition accuracy, characterized in that: include: Multiple terminal devices; as well as The server according to claim 7.

9. A machine-readable storage medium, characterized in that The machine-readable storage medium stores instructions for enabling a machine to execute the method for improving face recognition accuracy according to any one of claims 1 to 6.

Images