Patent Application
20240027426
The subject of the invention is a method of remotely evaluating the alcohol content in the exhaled air, applicable for conditioning access to workplaces using a computer application. The creation of an appropriate model allows the development of a virtual assistant supporting the work of a person supervising sobriety at the workplace.
So far, methods of remote monitoring and verifying the sobriety of employees are known, including remote measurement of alcohol content in exhaled air and informing the employer about its result based on telecommunications systems, with the possibility of remotely blocking access to the workplace. Known solutions do not include technology that enables the detection of counterfeits, as well as mistakes in the identification of the person for whom the measurement is made, with the possibility of self-learning.
As part of the solution, according to the invention, a set of activities was developed, including data preparation, transformation, and analysis to create an optimal device and user identification model.
From the Canadian description of the invention CA2906116 there is known a method of monitoring sobriety by means of a handheld device for breath testing, which, after receiving the alcohol concentration value in the exhaled air, generates a signal containing data on the content of the breath.
The substance and user identification data are sent wirelessly via the cellular network to the receiving station.
The American invention US2020191769A1 discloses a method of remotely monitoring sobriety by means of wireless devices connected via a communication network, providing users with various forms of alerts in the form of electronic messages.
From another US description of the invention, US2020256848, a user sobriety monitoring system is known. The system may include a test device that generates a signal on the content of the substance. The test device may further include a mouthpiece and a user identification device. The user identification device may generate user identification data in response to the user's breath and may transmit it from the testing device to the monitoring station. The test device may further include at least one LCD screen or a light emitting diode (“LED”). At least one of the LCD screens or the LEDs may display at least one randomly generated visible identification mark.
On the other hand, from also the American description of the invention US2020388117 there is known a method of monitoring sobriety by means of a handheld breath testing device which, after receiving a user's breath, generates a signal containing the substance content data and user identification data and wirelessly transmits the signal to the receiving station, in which the breath testing device includes a fingerprint reader.
The aim of the invention is to develop a method for measuring the alcohol content in the exhaled air of users, leveling the possibility of a false and incorrect measurement as to the device and user.
Solving the problem of automatic identification according to the proposed method allows for continuous evaluation of the recognition model.
As used herein, the word control refers to verification and taking into account a factor, in this case, the factor being the level of exhaled alcohol.
A method of remotely monitoring the alcohol content in the exhaled air, in which the alcohol concentration in the exhaled air is measured by means of a control device (UK) with radio data transmission technology, characterized in that the electronic traces of the control device (UK) and user (U) are saved in the web application data collection module (MGAW) on the server contained within the data analysis network infrastructure (ISAD), which are then used to train the deep convolutional neural network model, whose task is classification and identification of the characteristic visual features of users' faces (U), individual features identifying the control device (UK), such as dimensions and an individual QR code or bar code, and this information is compared with the alcohol concentration value obtained from the control device (UK) in exhaled air, combined in the mobile device (UM) with the date, time, geographical location, and with said defined QR code or barcode individualizing the electronic trace of the control device (UK) and visual characteristics of the user obtained from the camera of said mobile device (UM) (U), additionally with fingerprints, an individual code personalizing the user (U) manually entered into a mobile device (UM), then the obtained information is digitized and, via telecommunications links, saved in any format on the above-mentioned server contained within the data analysis network infrastructure (ISAD) in the web application data collection module (MGAW), for the purpose of comparison with model information, whereby via telecommunications links from said server, push or text messages are sent to the mobile device (UM) about the need to make a test, at any time intervals, preferably at the same time of the day and time, obliging the user (U) to perform measurement activities, provided that access to the workplace is allowed or prevented, and if the measured result of the alcohol concentration measurement in the exhaled air (U) by the user (U) is within the tolerance limits set in (BGAW), then the data analysis network infrastructure (ISAD) server sends the information unlocking the access to the workplace,
Preferably, in the web application data collection module (BGAW), individual user accounts (U) are created in which historical data, including date, individual device trace (UK) and user (U), and measured blood alcohol concentration results are stored.
Preferably, the web application data collector (BGAW) establishes a measurement time interval, including time of day, date, time, or time between preset measurement times, or random measurement at any time period, with the breath alcohol tolerance limit.
Preferably, the workstation is a means of transport or an element of production infrastructure, in particular a car, a production machine, a warehouse cart, an airplane, a production machine, or entrance gates, the server is connected to an electronic ignition or access control device.
Preferably, the deep convolutional neural network model for user face recognition (U) as part of the data analysis network infrastructure (ISAD) is continuously trained, that is, when new data arrives in the data block
Preferably, in the web application data collection module (BGAW), a model image of the user's face (U) is created on the basis of graphic files, presumably obtained from the camera of the mobile device (UM), and in time intervals it is compared with the image obtained within the individualizing electronic trace of the control device (UK) and the user (U), conditioning its compliance with the access to the workstation of the said user (U), the compliance being determined by the percentage of the characteristic visual elements of the user's face (U), which is stored in the template data storage module (DW) which can then be used to train a deep convolutional neural network model that determines the characteristics of the user's face (U) in percentage ranges.
Preferably, the email addresses or telephone numbers are added to the web application data collection module (BGAW), which will be notified if the breath alcohol concentration measurement result is above the tolerance set in the web application data collection module (BGAW), or if in the set over time, there is no feedback or the information obtained within the individualizing electronic trace of the control device (UK) and the user (U) is incomplete or inconsistent with the model information, in particular, the percentage of the characteristic visual elements of the user's face (U) is below the assumed value.
Preferably, the percentage of the visual characteristic of the user's face (U) is established within the framework of biometric 3D authentication of the personality of the user's image (U).
Preferably, if the user (U) ignores the command of the data analysis network infrastructure (ISAD) server regarding the measurement of alcohol concentration in the breath and positive or negative facial recognition, then along with the blocking of the workstation, said server sends an email notification or a text message to the addresses defined in the module web application data collection (BGAW).
The subject of the invention has been presented in the embodiment in the drawing, which shows a block diagram illustrating the method.
The use of the invention allows for a very accurate correlation of the user, the measuring device and the result of measuring the alcohol concentration in the exhaled air, with a very short time of at most a dozen or so seconds. The “learned” models are based on model data, then they are used to identify the features of individual faces and the features of individual devices. The selection of models is performed with the use of appropriate evaluation criteria and the selection of the best model obtained so far. The entire process of optimizing new models is automated and requires no human intervention.
A method of remotely checking the alcohol content in the exhaled air, in which the alcohol concentration in the exhaled air is measured by means of a control device (UK) with radio data transmission technology, characterized by the fact that the created reference data set (DW) containing the individualizing electronic traces of the control device (UK) and the user (U) are saved in the web application data collection module (MGAW) on the server contained within the data analysis network infrastructure (ISAD), which are then used to train the deep convolutional neural network model, which the task is to classify and identify the characteristic visual features of the users' faces (U), individual features identifying the control device (UK), such as dimensions and an individual QR code or bar code, and this information is compared with the concentration value obtained from the control device (UK) breath alcohol, combined in the mobile device (UM) with the date, time, geographic location, and with said defined QR code or barcode individualizing being the electronic trace of the control device (UK) and features obtained from the camera of said mobile device (UM) with the user's data (U), additionally with fingerprints, an individual code personalizing the user (U) manually entered into the mobile device (UM), then the obtained information is digitized and via telecommunications links are saved in any format on the above-mentioned server contained within the network infrastructure data analysis (ISAD) in the web application data collection module (MGAW), in order to compare them with the model information, whereby using telecommunications links from the said server, push or text messages are sent to the mobile device (UM) about the need to take a test, at any time intervals, preferably at the same time of day and time, obliging the user (U) to
In the web application data collection module (BGAW), individual user accounts (U) are created in which historical data including date, individual device trace (UK) and user (U), the result of the measured blood alcohol concentration are saved. In the web application data collection module (BGAW), the time interval of the measurement is set, including the time of day, date, time, or the time between the preset measurement times, or the randomness of the measurement at any time, with the limit value of alcohol tolerance in the exhaled air. A workstation is a means of transport or an element of production infrastructure, in particular a car, production machine, warehouse cart, plane, production machine or entrance gates, where the server is connected to an electronic ignition or access control device. The deep convolutional neural network model for user face recognition (U), which is part of the data analysis network infrastructure (ISAD), is continuously trained, which means that if new data appears in the template data storage (DW) block, in the next iteration training, the neural network model is trained using more examples. In the web application data collection module (BGAW), a model face image is created on the basis of graphic files obtained by default from the camera of a mobile device (UM) user (U) and in time intervals it is compared with the image obtained as part of the individualizing electronic trace of the control device (UK) and the user (U), conditioning access to the workstation of said user (U), the compliance being determined by a percentage characteristic visual elements of the user's face (U), which is saved in the template data storage (DW) module, which can then be used to train a deep convolutional neural network model, whose task is to determine the characteristics of the user's face (U) in percentage ranges. To the web application data collection module (BGAW), email addresses or telephone numbers are added, which will be notified if the result of the breath alcohol concentration measurement is above the tolerance set in the web application data collection module (BGAW), or if it does not take place within a given period of time. has feedback or the information obtained within the individualizing electronic footprint of the control device (UK) and the user (U) is incomplete or inconsistent with the model information, in particular, the percentage of the characteristic visual elements of the user's face (U) is below the assumed value. The percentage of the characteristic visual elements of the user's face (U) is determined as part of the biometric 3D authentication of the personal features of the user's image (U).
The subject of the invention is a method of remote control of alcohol content in the exhaled air, applicable when conditioning access to workplaces using a computer application. Creating an appropriate model allows you to develop a virtual assistant supporting the work of a person supervising sobriety at the workplace. The method of remote control for the value of users' alcohol, in which the concentration of alcohol in the exhaled air is measured by means of a control device (UK) implicitly with radio data transmission technology, is characterized by the fact that the created data set the reference (DW) comprising the individualizing electronic traces of the control device (UK) and the user (U) is stored in the web application data collection module (MGAW) on a server contained inside the network data analysis (ISAD) rastructure, which are then used to train a model of a deep convolutional neural network, whose task is to classify and identify the characteristic visual features of the individual face of the user (U), individual features identifying the device control (UK) such as dimensions and an individual QR code or barcode, this information being compared with the value of the alcohol concentration in the exhaled air obtained from the control device (UK), combined in the device mobile (UM) with date, time, geographical location, and with the said
