RU2777267C1 - Method for working time recording and the system for its implementation - Google Patents

Abstract

FIELD: biometrics.

SUBSTANCE: invention relates to the field of workplace monitoring using biometrics. The method for working time recording is that the camera continuously transmits a stream of images of a person’s face at the workplace, or images of the workplace to a computer of the workplace, which analyzes the received stream of images using software and determines whether an event has occurred, records the time of the event, and in the case of positive result issuance the computer installed at the workplace transmits an image of a person’s face to a server containing a database of reference images of a person’s face and software, which verifies the image of a person’s face with low resolution, the server records the time and verifies the image of a person’s face from its database of images of human faces, in case of detection of a high verification error, the server transmits an image of a person’s face made at the workplace.

EFFECT: increase in the accuracy of determining the time spent at the workplace.

16 cl, 2 dwg

Description

The invention relates to the field of recording working hours using biometrics, can be used, for example, to record the time of driving a vehicle, the time of training of athletes on simulators, the time of preparation of cadets or students, the time of work of operators of machines, mechanisms and technological equipment, etc.

Further in the text, the applicant gives the terms that are necessary to facilitate an unambiguous understanding of the essence of the claimed materials and to exclude contradictions and / or controversial interpretations when performing an examination on the merits.

Biometrics is a system for recognizing people by one or more physical or behavioral features (photograph of the face and/or body, voice pattern, fingerprints, hand vein pattern, blood group, special photo of the cornea of the eye, etc.). [https://ru.wikipedia.org/wiki/Biometry].

Identification - establishing the identity of an unknown object to a known one based on the coincidence of signs; recognition [https://ru.wikipedia.org/wiki/Identification].

Verification is a comparison in which the biometric system attempts to verify the person's identity. In this case, the new biometric sample is compared with the previously stored sample. By comparing these two samples, the system confirms that this person is really who he claims to be [http://izi.vlsu.ru/files/teach/books/910/theory.html]. Verification is a confirmation of the correctness of any action, object or event after verification [https://myrouble.ru/verification/].

EVMRM - computer of the workplace, for example, a computer,

ILC - an image of a person's face.

HBMI is an image of a person's face obtained at the workplace.

RM - workplace.

EILF is a reference image of a person's face.

DB - database.

BDEILCH is a database containing reference images of a person's face.

In the world there is a problem of accounting for the targeted use of working time in various areas of industry, the economy and the social sphere, which can be solved, for example, by expanding the arsenal of methods and systems for recording working time.

The claimed technical solution is aimed at improving the accounting (control) of working hours and can be used in any area that requires accounting of working hours, provided there is a workplace limited in space with the possibility of placing a video camera and a computer.

For example, the claimed technical solution can be fully used to account for:

• driving time to improve road safety;

• time of work of operators of machines, mechanisms and technological equipment, in order to control their presence at the workplace, which, accordingly, will increase the safety of the operation of machines, mechanisms and technological equipment;

• the time of preparation of cadets and students in order to control their presence at the training place in order to improve the training indicators as planned;

• training time for athletes in order to control sports performance;

and etc.

From the studied prior art, the applicant identified technical solutions aimed at controlling working time in various fields of activity.

So, there are known technical solutions for recording the working time of employees of organizations. From the studied state of the art, the Anviz AIM standard program [https://anviz.ru/support/help/aimstandard/item/reports-variants.html] was identified. The regular mode of attendance in the Anviz AIM standard program is the accounting of attendance by the first and last event of the day. In this case, the program considers the time between the first and last event to be working time. The event type is not taken into account. To track working time for the first and last event, you need to create a shift, a schedule, and assign a schedule to employees. Further, when calculating the working time, the program will offer to delete the events that are between the first and last event, the first event will be the arrival, and the last event the departure. In this case, the total working time will be calculated as the time of the working shift specified for the shift in the schedule (for example, 480 minutes) minus "late" and "leaving early". Anviz AIM standard provides time tracking capabilities to improve discipline. However, for more complete accounting with the possibility of using breaks, 1s shifts, night work, part-time work, payroll, etc., additional modules must be used.

The disadvantage of the known technical solution is the lack of the possibility of verifying a person from a photograph, as a result of which it is possible to carry out an event (marking arrival - departure) by another person.

The claimed technical solution fully relates to taking into account the training time of vehicle drivers, which is relevant in relation to obtaining the appropriate qualifications for them in order to reduce the number of road accidents (RTA), as a result of which, in addition to death and injury of people, material damage is caused to the economy . Accidents occur for various reasons, including due to insufficient preparedness of drivers, as well as their tired state as a result of processing, which, in particular, may be the result of:

- incomplete implementation of training driving programs in driving schools and, accordingly, low quality of driver training;

- insufficient control of the working regime of drivers in commercial road transport and taxis, which leads to the processing of drivers (tired state);

- insufficient control over the work regime of drivers in carsharing cars, which can lead to forgery of accounts, lack of timely information about violations of traffic rules (SDA) by the driver;

- and etc.

Further, the applicant considers these problems in more detail.

The lack of control over the implementation of approved driving instruction programs in driving schools leads to the fact that insufficiently trained drivers become road users and increase the risk of accidents.

As a result, the most important task is to expand the arsenal of methods for recording driving time.

Known methods for recording driving time in driving schools based on digital technologies, electronic and conventional document management [Driving School-Control System http://dscontrol.ru/], [Information System for Driving Schools Zebra https://scienceforum.ru/ 2015/article/2015014963] for verification use:

• magnetic cards and a student's mobile phone;

• a journal in which the instructor writes information about the implementation of the driving instruction program (DRP);

• a paper Certificate of completion by the student of the PPV to enter the exam.

The disadvantage of known systems is that they can allow fraud and create corruption risks.

There is also a problem in the world of controlling the working regime of drivers in commercial road transport , namely, the processing of drivers (tired state), which often leads to accidents

[https://rnsinfo.ru/news/tachograph/2051/],

[https://www.rbc.ru/society/23/12/2019/5e011b119a7947951c1b9a84?from=from_main].

It is known that tachographs are installed in commercial motor vehicles to record driving time [Tachographs: Decree of the Government of the Russian Federation No. 1213 of November 23, 2012 and Order of the Ministry of Transport No. 36 of February 13, 2013].

и трудозатраты, поэтому контролем охватывается небольшая часть коммерческого автомобильного транспорта. The disadvantage of using tachographs is the possible forgery of the driver card. Also, to remove information from the tachograph, employees of the GUOBDD (Main Directorate for Road Safety of the Ministry of Internal Affairs of Russia) require

and labor costs, so a small proportion of commercial road transport is covered by controls.

There is also a problem of controlling the identity of the driver in carsharing and taxi cars in the world, namely:

- forgery of accounts [https://www.pnp.ru/social/karshering-predlagayut-otregulirovat-zakonom.html];

- Lack of timely information about traffic violations by the driver, his qualifications and processing in a taxi.

Known technical solutions for driver verification use:

- comparison of a photo (selfie) of the driver and a driver's license, passport with those available in the database;

– verification of the geolocation of the car with the driver’s phone.

The disadvantage of the known technical solutions is that they do not verify drivers, allowing for the possibility of forgery. This often allows drivers of carsharing cars to drive aggressively, commit vandalism, and taxi drivers, after working out the allowed driving time, connect to another aggregator and continue working. Aggressive driving and tired drivers often lead to traffic accidents 2051/].

From the level of technology studied by the applicant, it was revealed that to solve this problem, systems of biometric verification of a person by facial features are known, which are used in various fields: security systems, control and recording of the working time of specialists, marketing, entertainment, etc.

Biometric systems are easy to use - they work contactlessly, continuously, without distracting a person from his activities.

Thus, the invention is known according to the patent of the Russian Federation No. 2587408 "Video monitoring of the functional state", the essence is a device for video monitoring of the functional state, which contains a 4.3 V power supply connected by a power input to the output of the connector for connecting the on-board power supply 12-24 V, and the output to the input power supply of a single-board microcomputer, which is connected to the accelerometer, to the first USB input, to the second USB input, to the photodiode, to the driver card slot, to the memory output, to the output of the mixer and the panic button. The single-board microcomputer is connected to the input-output port of the standard UART interface of the GSM/GPRS modem, which is connected to the GSM antenna through the GSM connector, the GSM/GPRS modem is connected to the SIM card, which is necessary for device authentication in the GSM network. The single-board microcomputer 8 is connected to the input-output port of the standard UART interface of the satellite navigation receiver, which is connected to the GLONASS/GPS/GALILEO antenna via the GLONASS/GPS/GALILEO antenna connector. The 4.3 V power supply is loaded on the supply inputs of devices 3, 5, 6, 7, 9, 13, 15, 18, 21, 22, 23, 24. The voltage regulator supplies the supply voltage through the GLONASS / GPS / GALILEO connector to the antenna GLONASS/GPS/GALILEO, which transmits signals from visible satellites of the GLONASS/GPS/GALILEO constellation to the GLONASS/GPS/GALILEO receiver. The GLONASS / GPS / GALILEO receiver, according to the data from the standard NMEA protocol, calculates the geographical coordinates of the object and transmits information about the location, speed and time, through the I / O port of the standard UART interface to a single-board microcomputer, in which information is accumulated and synchronized with time UTC. From the accelerometer to a single-board microcomputer comes information that is converted in the form of the distance traveled, speed and, in fact, is a duplicate of the information received from the GLONASS / GPS / GALILEO satellite navigation receiver. Through the input-output port of the standard UART interface, the navigation signal processed in a single-board microcomputer is fed to the GSM / GPRS modem, from where, through the GSM connector, it enters the GSM antenna. The GSM antenna transmits signals to the GSM network, while the SIM card determines the GSM network operator. The navigation signal formed according to the data from the GLONASS/GPS/GALILEO satellite constellation is transmitted via the Era GLONASS or eCall transport security system channels in the following cases:

A) if it is impossible to identify the identity of the driver, caused by alcohol or drug intoxication or the absence of his electronic photo in the authorized database;

B) upon the onset of a borderline or close to sleep state of the driver;

B) when a single-board microcomputer receives a signal from an accelerometer that exceeds the critical, “permissible” impact level in a traffic accident (RTA).

The driver's card slot can be used by a police officer when the vehicle is evacuated, while he is also subjected to a state determination procedure, and in addition, his photo (frame) is transmitted through the channels of the Era GLONASS or eCall transport security system, as a photo of a potential hijacker. The state detection system used in the device also performs an additional function of the security system that does not allow a person who is not included in the electronic database of authorized persons to drive this vehicle to drive a vehicle.

Known source "Real-time vehicle security system using face recognition" / K. Nandakumar, G. Muralidaran and N. Tarani / International Review of Applied Engineering Research. ISSN 2248-9967 Volume 4, Number 4 (2014), pp. 371-378 © Research India Publications [http://www.ripublication.com/iraer.htm].

The entity is a microcomputer system including an image processing control unit and a microprocessor for preventing theft of a parked vehicle. The face detection and recognition system uses an advanced authentication algorithm. The ARM microprocessor is used as the control unit in the system. A passive infrared sensor attached to the driver's seat activates a hidden camera installed inside the vehicle. using microprocessor control of the ARM microcomputer. after the intruder enters the vehicle. The camera captures an image of a person inside the car, fixed in the appropriate position in front of the driver's seat. After capturing an image of a person, the system now tries to detect the face. The microcomputer, which has an image processing unit built into it, performs face detection and authorizes the person. Image processing includes two parts: face detection and face recognition.

The resulting image is processed for face detection using the Viola Jones algorithm (Viola and Jones, 2002), which makes efficient use of cascading object detection. The cascade detector detects the face of the received image and highlights the face area. Authentication based security. The system has a database that stores images of authorized persons in various environments. Face images are enhanced by normalizing them to remove unwanted information due to lighting limitations upon image acquisition and stored in a database.

Face recognition can be performed using various algorithms based on features or models. Basically, feature-based algorithms are used in real-time security systems. Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) algorithms are efficient in terms of feature extraction to perform recognition. Both algorithms are compared (SK Hese and MR Banwaskar, 2013) and they find that they have similar features, but Linear Discriminant Analysis (LDA) outperforms Principal Component Analysis (PCA) when large training sets are involved in recognition. In addition, LDA efficiently recognizes most of the information presented in an image by calculating scatter matrices within a class and between classes. Using a database that contains normalized face images, recognition is performed in the vehicle's security system using the LDA algorithm. ms, feature extraction from stored images in the database, called training images, and a face image captured by the camera, called test image. The test image is compared with images from the database and the classifier used in the algorithm decides the image as known or unknown using Euclidean distance and a threshold. The Euclidean distance is calculated between the respective feature weights, and the image that gives the minimum distance best fits the test. A person is classified as known or authorized when the Euclidean distance is less than a threshold value, and a person is classified as unknown or unauthorized when the distance value is greater than the threshold.

An invention is known according to RF patent No. 2717878 "System and method for recognizing a vehicle user", the essence is a system for recognizing a vehicle user, containing a computer programmed to capture an image of a vehicle user by a camera in the vehicle; extracting a plurality of user profiles; comparing the image with the data stored in each user profile; in determining that the image does not match any user profile, identifying a plurality of user devices, each user device being associated with stored contact information for each user profile; and transmitting a message including the image to each identified user device. A method for recognizing a vehicle user, comprising capturing an image of the vehicle user by a camera in the vehicle; extracting a plurality of user profiles; comparing the image with data stored in each user profile; in determining that the image does not match any user profile, identifying a plurality of user devices, each user device being associated with stored contact information for each user profile; and transmitting a message including the image to each identified user device.

A common disadvantage of the analogs described above is that their database is not created on the server, but on the car's computer, which has less computing power, as a result of which a lower resolution is achieved in identifying a person, which leads to excessive false positives or tolerance errors. This limits the use of the known technical solution for its intended purpose.

The closest set of essential features, chosen by the applicant as a prototype, is the invention according to the patent of the Russian Federation No. real-time monitoring of events occurring in the vehicle cabin, which consists in performing the steps at which, using a computing device installed in the vehicle, the following is performed: obtaining data from a camera installed in the vehicle cabin, characterizing the image of the vehicle driver; extracting from the received data at least one image containing a display of at least the face of the driver of the vehicle; performing image preprocessing; determining the position of key points on the image; determining at least one area of interest of the image based on the position of the key points; determining, based on the analysis of each identified image area of interest, at least one type of incident, the analysis including processing the image area with a predictor for the presence of each possible incident type from a given set of incidents; fixing at least the time of detection of each mentioned incident; formation of a data packet containing at least information about each detected incident and the time of its fixation; and transmitting said data packet to the server part. The method according to claim 1, characterized in that the pre-processing includes converting the image to grayscale and normalizing the image. The method according to claim 2, characterized in that additionally at least one type of processing is performed, selected from the group: image alignment, image rescaling, determination of additional key points, mirror image of the face image relative to the nose point. The method according to claim 3, characterized in that the alignment of the image is carried out relative to the vertical axis and/or relative to the location of the eyes on the face image. The method according to claim 1, characterized in that the computing device additionally captures the GNSS coordinates of the vehicle and/or the speed of the vehicle and/or the parameters of the technical condition of the vehicle. The method according to claim 1, characterized in that the vehicle contains an additional camera that captures and transmits images of the environment of the vehicle to a computing device for fixing traffic violations. The method according to claim 1, characterized in that the set of types of incidents includes: talking on a mobile phone, smoking, obstruction of the system, distraction from the road, falling asleep, yawning, the driver is not wearing a seat belt. The method according to claim 7, characterized in that the authorized access to the vehicle is analyzed on the server based on the identification of the driver of the vehicle according to the images received from the computing device. A system for video analysis and monitoring of events occurring in the vehicle (V) cabin in real time, comprising: - a camera capable of recording a video data stream in the vehicle's cabin; - a computing device installed in the cabin of the vehicle and associated with said camera, and said device is configured to perform the method according to any one of paragraphs. 1-7; - a server connected by a data transmission channel to a computing device. The system according to claim 9, characterized in that it additionally contains installed on the vehicle: a GNSS antenna, a vehicle status data unit (BDTS), a GSM/GPRS/LTE/5G modem, or combinations thereof. The system according to claim 9, characterized in that it additionally includes an workstation of an employee of the Situation Center, configured to receive data on fixing incidents from the server. The system according to claim 9, characterized in that the server provides analysis of authorized access to the vehicle based on the identification of the driver of the vehicle from the images received from the computing device. The system according to claim 9, characterized in that it contains an additional camera installed on the vehicle, which ensures the capture and transmission of images of the environment of the vehicle to a computing device for fixing traffic violations by both the vehicle and other road users.

The disadvantages of the prototype are:

• the use of a more complex method of recognition - identification compared to a simpler method - verification in the claimed technical solution;

• lower DFI recognition accuracy due to the limited computing power of both the computing device installed in the vehicle and its own server, compared to the claimed technical solution, which uses a third-party DFI verification server to ensure high verification accuracy;

• greater labor costs and resources for the development of its own accurate high-quality neural network with high resolution, compared to the claimed technical solution, in which the server requests verification from a third-party server, the development of which was provided by a large number of invested resources (volume of training samples; hours spent on server development; number of machine hours spent on neural network training);

• lower accuracy of DFI recognition due to the fact that there is no way to replenish the database with reference images (from the workplace in different angles of the face, in different environments and lighting conditions), including those obtained by verifying the DFI on a third-party server, in contrast to the declared technical solutions where such a method is described.

At the same time, the applicant believes that the claimed technical solution is not obvious to a specialist in this field of technology and is not a summation of known technical solutions both in relation to the method and in relation to the system, for the following reasons.

Regarding the method:

• the identified technical solutions, including the prototype, do not match the declared technical solution for its intended purpose, since they are mainly aimed at identifying incidents, therefore, verification is not declared in them and there is no working time accounting algorithm;

• the identified analogs, including the prototype, do not describe the method of identifying the DST, while it follows from the prior art that databases containing reference DSTs are used for identification, which may differ greatly from the DST performed at the workplace (including including in the vehicle), but in the declared technical solution of the ILCRM, made in various angles, appearance (beard, mask, glasses, clothes, etc.) and environment (lighting, background), verified on a third-party server with high resolution, are entered into the database as a reference, which significantly increases the accuracy of recognition;

• in the identified analogues, including the prototype, the method of identifying the driver of the vehicle (V) is not described, and in the claimed method, the verification method is specified by indicating the actions to be performed if a high verification error is detected: if a high verification error is detected, server C1 refers to a third-party C2 server for accurate verification; wherein C1 transmits to C2 a sample of the reference DFI and DFI and receives a verification result;

• the identified analogues, including the prototype, do not indicate the possibility of replenishing the database with verified HFI, and in the claimed method, with a positive result of the verification of the HDLHRM on C2, the verified HILCH is added to the OBEHMI C1, thus the OBEHMI C1 is constantly replenished with the verified on the C2 HLI, which become EIFI, which increases the accuracy of recognition, as well as speeds up and reduces the cost of verification of subsequent DFI.

Regarding the system.

The identified sources do not contain such features of the claimed system as communication with a third-party server capable of performing high-resolution DFI verification, which allows:

• Significantly improve the accuracy of verification or identification compared to known analogues, including the prototype;

• allows you to manage less labor costs, resources, computing power of the system during the next deployment and operation of the claimed system in comparison with known analogues, including the prototype.

• increase the accuracy of system verification by increasing the comparison sample.

Based on the foregoing, the claimed technical solution in relation to the method and in relation to the system for its implementation, according to the applicant, is not obvious to specialists from this field of technology.

The objective and technical result of the claimed technical solution is to develop a method for recording working time and a system for its implementation, allowing to eliminate the shortcomings of the prototype, namely:

- the use of a simpler method of recognition - verification compared to a more complex method - identification in the prototype;

– increasing the accuracy of DFI recognition due to the expansion of computing power by using a third-party DFI server for verification, without increasing the complexity of your own server;

– reduction of the required resources and labor costs during the development, deployment and operation of the system, as well as improving the quality of verification by adding a third-party C2 server to the system with a much higher computing power and resolution of the DLI verification, the development of which was provided by a large number of invested resources (the volume of training samples ; the number of man-hours spent on the development of the server; the number of machine hours spent on training the neural network);

– increasing the accuracy and reducing the required resources necessary for high-quality verification of the DFI by developing a method for replenishing the database with reference images of a human face (from a workplace in different angles of the face, in different settings and lighting conditions), including those obtained by verifying the DFI with high accuracy on a third-party C2 server.

The development of the claimed technical solution, according to the applicant, will improve the quality of training by accounting (controlling) working time (for example, by taking into account the driving time of cadets in driving schools, training time for athletes on simulators, training time for cadets and students), the quality of car maintenance, mechanisms and technological equipment (for example, accounting for the time of work of operators of machines, mechanisms and technological equipment).

The claimed technical solution is not limited to the above list of specialties, since it is clear to a specialist in this field of technology that the claimed technical result can be achieved taking into account the working time of any specialist located at a stationary workplace with the possibility of placing a camera and a computer.

The essence of the claimed technical solution is a method for recording working time, which consists in the fact that the camera continuously transmits a stream of images of a person's face at the workplace, or images of the workplace, to an electronic computer of the workplace, which analyzes the received image stream using software and determines , whether the event occurred, fixes the time of the event, while a positive result is issued if the result of determining the number of faces in the frame has changed or is not defined; if a positive result is issued, an electronic computer installed at the workplace transmits the image of a person’s face to a server containing a database of reference images of a person’s face and software that verifies the image of a person’s face with a low resolution, the server fixes the time and verifies the image of the face a person in their database of images of human faces; if a high verification error is detected, the server transmits the image of a person's face made at the workplace and a selection of reference images of a person's face from its own database to a third-party server running third-party software that provides high-resolution verification; a third-party server verifies a person's face and returns the result to the server; if the verification result is positive, the database of reference images of a person's face is replenished with a verified photo; in case of coincidence, _{accounting of} working time is carried _out by the difference between the time of _{disappearance} from the frame and the time of _appearance in the _frame according to the formula t _{disappearance} – time of disappearance from the frame, n – number of disappearances from the frame, ∑ _n – n-th number of summations; in case of a discrepancy, the accounting of working hours is not carried out. The method according to claim 1, characterized in that the workplace is a vehicle cabin. The method according to claim 2, characterized in that the cabin of the vehicle is a taxi cabin. The method according to claim 2, characterized in that the cabin of the vehicle is the interior of a car sharing car. The method according to claim 2, characterized in that the cabin of the vehicle is the cabin of a commercial vehicle. The method according to claim 1, characterized in that the workplace is a training simulator. The method according to claim 1 , characterized in that the workplace is a training place for a cadet or student. The method according to claim 1, characterized in that the workplace is the workplace of the operator of machines, mechanisms and technological equipment. A system for implementing the method according to claim 1, comprising a camera capable of capturing a video data stream from a workplace, an electronic computer of a workplace with software associated with said camera, a server configured with a database of reference images of a human face from various angles, and with software capable of carrying out low-resolution verification, wherein the server is connected by a data transmission channel to an electronic computer of the workplace, characterized in that the server is connected to a third-party server on which third-party software is executed with the possibility of carrying out high-resolution verification. The system according to claim 9, characterized in that the workplace is a vehicle cabin. The system according to claim 10, characterized in that the cabin of the vehicle is a taxi cabin. The system according to claim 10, characterized in that the cabin of the vehicle is the interior of a car sharing car. The system according to claim 10, characterized in that the cabin of the vehicle is the cabin of a commercial vehicle. The system according to claim 9, characterized in that the workplace is a training simulator. The system according to claim 9, characterized in that the workplace is a training place for a cadet or student. The system according to claim 9, characterized in that the workplace is the workplace of the operator of machines, mechanisms and technological equipment.

The claimed technical solution is illustrated in Fig.1, Fig.2

Figure 1 shows a diagram of the claimed system for implementing the claimed method, where:

1 - camera.

2 – workplace computer with software developed by the applicant.

C1 - a server containing a database of reference images of a human face (BDEHFI) from various angles and software developed by the applicant, with the possibility of low-resolution verification.

C2 is a third-party server containing third-party software, with the ability to carry out high-resolution verification.

Figure 2 shows a Table that presents the results of accounting for the working time of various categories of users according to the formula t _r = ∑ _n (t _{disappearance} - t _appearance ), where:

t _r - working time taken into account,

t _appearance - the time of appearance in the frame,

t _{disappearance} - the time of disappearance from the frame,

n is the number of drops from the frame,

∑ _n – nth number of summations

Further, the applicant shows the implementation of the claimed technical solution.

Biometric systems for recording working time are easy to use - they work contactlessly, continuously, without distracting a person from his activities. However, the accuracy and cost of one recognition, the amount of data transmitted from the computer to the server C1, are quite large. At the same time, to create own software on the C1 server that would perform high-resolution verification, significant material and technical resources are required, including, for example, the volume of training samples, the number of man-hours spent on server development, the number of machine hours, spent on neural network training, etc.

The claimed method, implemented on the claimed system, was developed on the idea of connecting a third-party server C2 to its own server C1 with the possibility of carrying out high-resolution biometric verification of reference images of a person's face, since the third-party server C2 contains its own software that allows high-resolution verification.

So, for example, if the cost of developing software for the C1 server is about several million rubles, then the cost of developing software for a third-party server C2 is about several tens of millions of dollars. As a result, the material and technical costs of connecting a third-party C2 server turn out to be disproportionately lower than the development, deployment and operation of a similar extended system on your own C1 server. At the same time, when using a third-party server C2, the accuracy of DST verification increases sharply as a result of the possibility of carrying out verification with high resolution.

In addition, there is a replenishment of the ODEILCH of the C1 server, since the verified DFI (performed at different angles of the face, in different settings and lighting conditions) are entered into the C1 database as reference ones, while the recognition accuracy is significantly increased, which leads to the achievement of the stated technical results. .

The claimed system consists of the following parts (Figure 1):

– Camera 1, aimed at the workplace and made with the possibility of capturing a video stream of data (images) from the workplace.

- EVMRM 2 with software developed by the applicant and a channel for communication via the Internet, GSM or GLONASS with server C1, for example, a radio channel.

– Server C1 containing a database of reference images of a human face (BDEHFI) from various angles and software developed by the applicant, with the possibility of low-resolution verification.

– Third party C2 server with third party software capable of high resolution verification.

Wherein:

- EVMRM 2 is connected to the camera 1, for example, by a cord connection or a radio channel;

- server C1 is connected to the computer, for example, by a two-way radio data transmission channel,

- server C1 is connected to a third-party server C2, for example, via the Internet.

The claimed method is carried out as follows (see Fig.1).

At the workplace, the camera 1 is installed so that it is directed to the workplace. The camera continuously captures a stream of images of a person's face at the workplace or images of the workplace.

The camera 1 is connected to the EVMRM 2, for example, by a cable connection or a radio channel. The software developed by the applicant is pre-installed on the EVMRM with the ability to:

• counting the number of persons in the frame;

• detection of events when the number of faces in the frame has changed;

• identifying events where the number of persons cannot be accurately determined;

• fixing the time of the above events;

• transmission including other data, such as sensor readings.

The camera 1 continuously transmits a stream of HDLM or PM images to the computer.

The EVMRM analyzes the received image stream using software and determines whether an event has occurred, for example, a significant change in the image in the frame, fixes the time of the event, while a positive result is issued if the result of determining the number of faces in the frame has changed or is not defined.

In case of issuing a positive result, the EVMRM transmits the DFI to the C1 server containing the ODEILCH database and software that verifies the image of a person's face with a low resolution.

Server C1 fixes the time, verifies the image of a person's face against its database of images of a person's face.

If a high verification error is detected, the C1 server transmits the ILCHRM and the EILF sample from its own database to the third-party server C2, which runs third-party software that provides high-resolution verification.

Third-party server C2 performs verification of a person's face, and returns the result to server C1.

If the verification result is positive (in case of a match), the server C1 ODEILCH is replenished with the verified DLI.

In case of a match, the accounting of working time is carried out according to the difference in the time of disappearance and appearance in the frame. If we assume that a person has disappeared from the frame n times, the working time is recorded according to the formula (1):

t _r = ∑ _n (t _{disappearance} - t _appearance ), (1)

where:

t _r - working time taken into account,

t _appearance - the time of appearance in the frame,

t _{disappearance} - the time of disappearance from the frame,

n is the number of times out of the frame.

∑ _n is the nth number of summations.

If there is a discrepancy, the working hours are not recorded.

Further, the applicant provides examples of the implementation of the claimed technical solution.

The applicant conducted accounting of working hours of various categories of users according to the claimed method using the claimed system. The results are shown in the Table in Fig.2.

Example 1. Accounting for driving time of a vehicle.

1.1. Accounting for driving time in training cadets of driving schools.

The claimed system is installed in the cab of the training vehicle (TC), namely:

- camera 1, so that it was directed to the workplace of a driving school cadet, for example, fixed on the windshield;

- EVMRM 2 with software associated with camera 1, for example, a cord connection or a radio channel, is installed, for example, on the torpedo of the vehicle.

Server C1 is installed, for example, in the operating organization.

Server C2 is, for example, a Google or Microsoft server.

Camera 1 continuously captures a stream of images of a person's face at the workplace or images of the workplace.

The face of a person (driver) appears in the field of view of camera 1.

The stream of images of a person's face from the camera 1 is continuously transmitted to the computer.

The computer analyzes the received image stream using software developed by the applicant, determines whether an event has occurred, for example, a significant change in the image in the frame, fixes the time of the event.

If a person's face appears in the frame, a positive result is issued and the EVMRM 2 transmits the DFC to the server C1, for example, using a radio channel.

Server C1 receives data from the EVMRM 2 fixes the time t _{of occurrence} , for example, 14.00, performs verification of the low resolution of the DFI by comparing the DFI with the BDEILCH it has.

If a high verification error is detected, C1 transfers the ILCHRM and the EILF sample from its own database to a third-party server C2, which runs third-party software that provides high-resolution verification.

Third-party server C2 performs verification of a person's face, and returns the result to server C1.

At the same time, in case of a match, the C1 server database is updated with a verified photo.

In this case, in case of a match, the time of appearance t _{of appearance} is fixed, for example, 14.00.

If there is a discrepancy, the working hours are not recorded.

When a person's face disappears from the frame (for example, at the end of the driving time), a sequence of actions similar to that described above is performed. At the same time, the time of the disappearance of the person's face from the frame t of the _{disappearance} is fixed, for example, 15.30.

In this way:

t _{of appearance} = 14.00, t of _{disappearance} = 15.30.

The vehicle driving time t _r is taken into account by the difference between the time of disappearance from the frame t of the _{disappearance} and the time of appearance in the frame t _{of the appearance} (n = 1) according to the formula (1):

t _r = 15.30 - 14.00 = 1 hour 30 min.

Thus, by the claimed method using the claimed system, it was reliably established that the driving time of the vehicle was 1 hour. 30 minutes.

If a person's face disappears from the frame during the training time with its subsequent appearance (n > 1), the accounting of working time t _r is calculated by summing the difference in the time of disappearances and appearances in the frame.

So, if a person's face disappears from the frame during the training time 2 times (n = 2), for example:

1. t _{of appearance} = 14.00, t of _{disappearance} = 14.45,

2. t _{of appearance} = 15.00, t of _{disappearance} = 15.30,

t _rv \u003d (14.45 - 14.00) + (15.30 - 15.00) \u003d 45 min. + 30 min. = 1 hour.15 min.

Thus, by the claimed method using the claimed system, it was reliably established that the time of driving a training vehicle by a cadet of a driving school was 1 hour. 15 minutes.

Accounting for working time with an increase in the number n of disappearances of a person from the frame is carried out in a similar way, increasing the number of summations.

Examples 1.2 - 4. Accounting for working hours of various categories of users.

The applicant conducted accounting of working hours of various categories of users according to the claimed method using the claimed system.

For this, a sequence of actions was carried out according to example 1.1., which differs in the type of workplace, type of working time, type of control, indicators n, t _{of appearance,} t t _{of disappearance.}

The results are shown in the Table in Fig.2.

From the results given in the Table, it can be seen that according to the claimed method, using the claimed system, the operating time of various categories of users was reliably established.

Example 5. A method of accounting for working time in the event of the appearance of unauthorized persons in the frame.

If, after the start of work, at least one unauthorized person appeared in the frame, a sequence of actions is carried out according to Example 1.

At the same time, a positive result is also issued and the EVMRM transmits to server C1 images of all faces in the frame.

The software time stamps images and performs low resolution verification on C1.

In the event of a negative result for a controlled person, the accounting of working time is not carried out for him.

In the case of a positive result, the server C1 transmits to the third-party server C2 the photographs of faces received from the computer 2 and images from its own database.

Server C2 performs high resolution verification.

In the event of a negative result for a controlled person, the accounting of working time is not carried out for him.

In the case of a positive result, server C2 sends server C1 a match/no match conclusion.

Next, work hours are recorded according to Example 1 - formula (1).

Example 6. A method of recording working time in case of impossibility to identify a person.

If the face is not detected or is not clearly defined when shooting with the camera, the calculation of working time is not carried out and the working time is not counted.

Thus, from the above, we can conclude that the applicant has achieved the tasks and the claimed technical result , namely: a method for recording working time and a system for its implementation have been developed, which allow eliminating the shortcomings of the prototype, namely:

- a simpler method of recognition was used - verification compared to a more complex method - identification in the prototype;

– increased accuracy of DFI recognition due to the expansion of computing power by using a third-party DFI server for verification, without increasing the complexity of its own server;

– the required resources and labor costs during the development, deployment and operation of the system have been reduced, as well as the quality of verification has been improved by adding a third-party C2 server to the system with a much higher computing power and resolution of the DCI verification;

– increased accuracy and reduced the required resources required for high-quality verification of the HID by developing a method for replenishing the BDEHIL with images made from different angles and verified with high accuracy on a third-party server.

The development of the claimed technical solution, according to the applicant, has improved the quality of training by accounting for working time (for example, by accounting for the time of driving cadets in driving schools, the time for training athletes, the time for training cadets in military specialties), the quality of maintenance of technological equipment (accounting for working hours of process equipment operators).

The claimed technical solution is not limited to the above list of specialties, since it is clear to a specialist in this field of technology that the claimed technical result can be achieved taking into account the working hours of any specialist located at a stationary workplace with the possibility of placing a camera and a computer.

The claimed technical solution complies with the " novelty " patentability condition for inventions, since at the date of submission of the application materials by the applicant from the studied prior art, no sources were identified that have a set of features identical to the set of features of the claimed technical solution.

The claimed technical solution complies with the " inventive step " patentability condition for inventions, since the totality of the claimed features provides technical results that are not obvious to a specialist and exceed the technical result of a prototype. From the studied state of the art, the Applicant has not identified technical solutions that would simultaneously use the claimed features in the claimed purpose, the identified sources lack the sequence of actions of the claimed method and the cumulative features of the claimed system, therefore, the claimed technical solution is not obvious to a specialist in the analyzed field of technology.

The claimed technical solution meets the condition of patentability " industrial applicability " for inventions, because the claimed method can be carried out using standard equipment and known techniques.

Claims (16)

1. A method of recording working time, which consists in the fact that the camera continuously transmits a stream of images of a person's face at the workplace, or images of the workplace to an electronic computer of the workplace, which analyzes the received image stream using software and determines whether event, fixes the time of the event, while a positive result is issued if the result of determining the number of faces in the frame has changed or is not defined; if a positive result is issued, an electronic computer installed at the workplace transmits the image of a person’s face to a server containing a database of reference images of a person’s face and software that verifies the image of a person’s face with a low resolution, the server fixes the time and verifies the image of the face a person in their database of images of human faces; if a high verification error is detected, the server transmits the image of a person's face made at the workplace and a selection of reference images of a person's face from its own database to a third-party server running third-party software that provides high-resolution verification; a third-party server verifies a person's face and returns the result to the server; if the verification result is positive, the database of reference images of a person's face is replenished with a verified photo; in case of coincidence, _{accounting of} working time is carried _out by the difference between the time of _{disappearance} from the frame and the time of _appearance in the _frame according to the formula t _{disappearance} – time of disappearance from the frame, n – number of disappearances from the frame, ∑ _n – n-th number of summations; in case of a discrepancy, the accounting of working hours is not carried out. 2. The method according to claim 1, characterized in that the workplace is a vehicle cabin. 3. The method according to claim 2, characterized in that the cabin of the vehicle is a taxi cabin. 4. The method according to claim 2, characterized in that the cabin of the vehicle is the interior of a car sharing car. 5. Method according to claim 2, characterized in that the cabin of the vehicle is the cabin of a commercial vehicle. 6. The method according to claim 1, characterized in that the workplace is a training simulator. 7. The method according to claim 1, characterized in that the workplace is a training place for a cadet or student. 8. The method according to claim 1, characterized in that the workplace is the workplace of the operator of machines, mechanisms and technological equipment. 9. A system for implementing the method according to claim 1, comprising a camera configured to capture a video data stream from a workplace, an electronic computer of a workplace with software associated with said camera, a server configured with a database of reference images of a person's face under various angles and with software with the possibility of low-resolution verification, while the server is connected by a data transmission channel with the electronic computer of the workplace, characterized in that the server is connected to a third-party server, on which third-party software is executed with the possibility of high-resolution verification . 10. The system according to claim 9, characterized in that the workplace is a vehicle cabin. 11. The system according to claim 10, characterized in that the cabin of the vehicle is a taxi cabin. 12. The system according to claim 10, characterized in that the cabin of the vehicle is the interior of a carsharing car. 13. The system according to claim 10, characterized in that the cab of the vehicle is the cab of a commercial vehicle. 14. The system according to claim 9, characterized in that the workplace is a training simulator. 15. The system according to claim 9, characterized in that the workplace is a training place for a cadet or student. 16. The system according to claim 9, characterized in that the workplace is the workplace of the operator of machines, mechanisms and technological equipment.

Images