KR102625814B1 - DUI zero solution - Google Patents

Abstract

The present invention installs an AI module that recognizes the driver's face and a breathalyzer module that can perform a breathalyzer test in front of the driver's seat inside the vehicle, so that the driver engaged in the transportation industry uses the AI facial recognition module after boarding the bus. After recognizing the driver's face, it is compared with the driver information data of the vehicle already stored on the server to determine whether the driver on board is the same person, and then the driver's drinking is measured using a breathalyzer. It provides a vehicle-type breathalyzer system that recognizes the driver's face and determines whether it is the same person as the dispatched driver. In particular, it uses AI algorithm-based CNN (Convolutional neural network) or deep learning-based YOLO. We provide an AI facial recognition-based zero drunk driving solution that uses one facial recognition measurement method selected from the (You Only Look Once) algorithm.

Description

AI facial recognition-based drunk driving zero solution {DUI zero solution}

The present invention relates to a zero solution for drunk driving (DUI: Drive under the influence). More specifically, an AI module that recognizes the driver's face and a breathalyzer module that can perform a breathalyzer test are installed in front of the driver's seat inside the vehicle. Once installed and a driver engaged in the transportation industry boards the bus, the driver's face is recognized using the AI facial recognition module, and the driver's face is compared with the vehicle's driver information data already stored on the server to determine the driver's status. This is about an AI facial recognition-based drunk driving zero solution that determines whether the driver is the same person, then uses a breathalyzer to measure whether the driver is drunk or not, and then operates the vehicle if there is no problem.

Traffic accidents caused by drunk driving lead to major accidents, so the seriousness of accidents and damage caused by drunk driving is continuously promoted, but drunk driving is not decreasing. After drinking, you must leave your car and use public transportation, but drunk drivers mistakenly think they can drive and end up driving drunk.

In particular, drivers who drive public transportation such as buses, taxis, trains, and subways are responsible not only for their own lives but also for the lives of public transportation users, so drunk driving should be further prevented.

With the recent revision of the Passenger Transport Service Act, transportation workers are checked to see if they are drinking before boarding a vehicle, and if they are judged to be unable to drive safely due to drinking, they are restricted from operating the vehicle.

In particular, drunk driving accidents by bus drivers can put the lives of many passengers at risk, and as there is a high possibility of major casualties occurring, the government passed an amendment to the Passenger Transport Business Act at the Cabinet meeting in April 2020 to ban drunk driving. A plan should be developed to strengthen management responsibility for transportation companies to check and record drinking before driving.

Drivers manage the breathalyzer test of transport workers before boarding the vehicle, but this is insufficient to prevent surrogate measurements and false records, so the need for a real-time breathalyzer test system to solve this problem is increasing. Currently, it is being used in the transportation business field, but the part to prevent proxy measurement is insufficient and needs improvement. In the future, the application fields are expected to gradually diversify, not only in the transportation business but also in industrial sites, medical fields, and personal measurement purposes.

Meanwhile, an example of prior art regarding a system for measuring alcohol consumption in relation to the transportation industry is the Republic of Korea Intellectual Property Office Patent Publication No. 10-2085489 (name: Hybrid alcohol consumption measurement management system).

The prior art is a hybrid breathalyzer management system (1) including an office breathalyzer management terminal 100, a network 200, and a breathalyzer management server 300, wherein the office breathalyzer management terminal 100 is Installed within the union office, transport workers check the information of transport workers through fingerprint recognition using a fingerprint reader (120) before driving the vehicle, and measure the transport worker's blood alcohol concentration using a breathalyzer (110) to analyze suitability and inadequacy. Afterwards, the appropriate and unsuitable information is stored on the storage unit 138 and transmitted to the breathalyzer management server 300 through the network 200 for management to prevent accidents caused by drinking.

However, the prior art uses fingerprints to identify the driver, so problems such as illegal acts where a drunk driver only verifies his identity and is measured by another person on his behalf during the subsequent alcohol test may occur.

Republic of Korea Intellectual Property Office Registered Patent Publication No. 10-2085489

Therefore, the purpose of the present invention to solve the problems described above is to install an AI module that recognizes the driver's face and a breathalyzer module that can perform a breathalyzer test in front of the driver's seat inside the vehicle, so that the driver engaged in the transportation industry can After boarding the bus, the driver's face is recognized using the AI facial recognition module, and then compared with the vehicle's driver information data already stored on the server to determine whether the driver on board is the same person. We provide a zero drunk driving solution that uses a measuring device to measure whether the driver is drunk and allows the vehicle to be driven if there is no problem.

Another object of the present invention is to recognize the driver's face to determine whether it is the same person as the dispatched driver, and in particular, to use AI algorithm-based CNN (Convolutional neural network) or deep learning-based YOLO (You Only Look Once) We provide an AI facial recognition-based zero drunk driving solution that uses one facial recognition measurement method selected from among the algorithms.

The present invention to achieve the above-described object is a vehicle-type alcohol breathalyzer system that measures the driver's drinking status when the driver engaged in a transportation business gets on the vehicle. It is installed in front of the driver's seat of the vehicle and has a display. A mounted terminal, an AI facial recognition module installed at the top of one front side of the terminal and capturing facial recognition information of the driver, a camera module installed at the other front side of the terminal and capturing a video of the driver seated in the driver's seat, the above A breathalyzer module installed at the bottom of one side of the front of the terminal to measure the blood alcohol concentration of a driver whose face recognition has been completed, and a control module 60 that receives and controls information transmitted from the facial recognition module, camera module, and breathalyzer module. It includes, and the control module compares the driver's facial image information transmitted from the facial recognition module with the driver's information stored in the database server and determines that the driver who boarded the vehicle is the same as the driver who received the dispatch. The control module compares and determines whether a person is a person, and the control module compares and determines the driver's face stored in the database server when the face of the driver seated in the driver's seat is photographed and transmitted from the camera module. The presence or absence of a breathalyzer test is determined by comparing the driver's blood alcohol concentration measurement value measured in the breathalyzer module with the allowable breathalyzer test value, and the facial recognition module uses AI algorithm for CNN (Convolutional neural network) or grid. It uses a facial recognition measurement method selected from the deep learning-based YOLO (You Only Look Once) algorithm, which predicts the number of bounding boxes designated in a predefined form centered on the center of the grid and calculates reliability based on this. CNN using AI algorithm collects each image feature detected through a detection unit that detects the entire outline of the driver's face, a facial feature detection unit that detects features on the driver's face, and a facial asymmetry detection unit that detects facial asymmetry of the driver. By collecting and transmitting the information to the control module 60, the control module can accurately compare and determine the transmitted facial image information of the driver with the driver and facial images stored in the database server, and the facial recognition module From the point where the driver starts boarding the dispatched vehicle, the driver's movements are detected and an image of the driver's face is taken, and the camera module measures breath alcohol from the breathalyzer module after the facial recognition module operates. Video recording of the driver continues until this is completed. A remote control module is connected to the control module, and the remote control module can be remotely controlled using the manager's terminal if a malfunction occurs in the terminal. It is characterized by being able to do so.

The present invention installs an AI module that recognizes the driver's face and a breathalyzer module that can perform a breathalyzer test in front of the driver's seat inside the vehicle, so that the driver engaged in the transportation industry uses the AI facial recognition module after boarding the bus. After recognizing the driver's face, it is compared with the driver information data of the vehicle already stored in the server to determine whether the driver on board is the same person, and then the driver's drinking is measured using a breathalyzer. It has the effect of preventing drunk driving by transportation workers, which can cause serious accidents, by allowing the vehicle to be driven when there is no problem.

In particular, a facial recognition measurement selected from CNN (Convolutional neural network) or deep learning-based YOLO (You Only Look Once) algorithm using an AI algorithm that recognizes the driver's face and determines whether the driver is the same person as the dispatched driver. By using this method and transmitting face information, it has the effect of transmitting accurate data without error.

Figure 1a is a diagram showing the state of the driver's seat to which the AI facial recognition-based zero drunk driving solution has been applied.
FIG. 1B is a diagram illustrating the tablet and breathalyzer module of FIG. 1A.
Figure 2 is a diagram showing the configuration of Figure 1.
Figure 3 is a diagram showing the operational flow of the AI facial recognition-based drunk driving zero solution.
Figure 4 is a diagram showing the configuration of a control module.
Figure 5 is a flowchart showing the operational flow of the AI facial recognition-based drunk driving zero solution.
Figure 6 is a flowchart sequentially showing the administrator's actions in the AI facial recognition-based drunk driving zero solution.

Hereinafter, embodiments of the present invention will be described with reference to the attached drawings. In the detailed description to be described later, representative embodiments of the present invention will be presented to achieve the above-described technical problem. And other embodiments that can be presented with the present invention are replaced with descriptions in the configuration of the present invention.

In the present invention, the existing transportation company measures whether all drivers are drinking and prepares and stores a breathalyzer management ledger with related contents. This is converted to digital to determine whether the driver is drinking alcohol, measurement time, result, and time of measurement. The purpose is to transmit and monitor images to a web server in real time to help solve drunk driving. In other words, since the existing breathalyzer test system is performed manually, there is a difficulty in detecting non-measurements or proxy testers, but the present invention can solve the problems mentioned through user authentication and recording through facial recognition. In addition, through the development of the product's own technology for the face recognition unit and body temperature measurement unit, it is expected that the expected economic and technical effects will be maximized by reducing waste and upgrading performance.

The facial recognition unit recognizes the driver through AI facial recognition, but currently, proxy operation and proxy measurement management are insufficient in measuring and driving the drinking of all drivers. To begin operation, attach a "drinking measurement system" to the driver's seat. It allows for proxy driving through facial recognition, breathalyzer testing, and real-time facial recognition during driving. In particular, the present invention predicts the number of bounding boxes designated in a predefined form centered on the center of the grid for a CNN or grid using an AI algorithm and based on this, the facial recognition module is used for face recognition. By using one facial recognition measurement method selected from the deep learning-based YOLO (You Only Look Once) algorithm that calculates It was allowed to happen.

Hereinafter, specific examples of the AI facial recognition-based drunk driving zero solution of the present invention will be described with reference to the attached drawings.

FIG. 1A is a diagram showing the state of the driver's seat to which the AI facial recognition-based drunk driving zero solution has been applied, FIG. 1B is a diagram showing the tablet and breathalyzer measurement module of FIG. 1A, and FIG. 2 is a diagram showing the configuration of FIG. 1. It is a drawing, and Figure 3 is a diagram showing the operational flow of the zero drunk driving solution.

As shown in Figures 1 to 3, the AI facial recognition-based drunk driving zero solution includes a tablet 10 equipped with a display 12, and a facial recognition system provided on the tablet 10 to recognize the driver's face. The module 20, the camera module 30, and the tablet 10 are connected via Bluetooth, a wireless transmission and reception method, and the breathalyzer module 40, which measures whether the driver has been drinking, the remote control module 50, and the driver's A control module 60 that compares the face with pre-set data to identify the driver and determines the drunk state based on breathalyzer test values, and is linked to the control module 60 to identify drivers engaged in the transportation industry. It consists of a database server (70) that converts and stores the face into data, and also converts and stores breathalyzer test values into data, and a management center (80) that monitors the driver's situation.

The tablet 10 is installed in the driver's seat of the vehicle, and from the time the driver, who is employed in the transportation industry, begins boarding the vehicle, after the driver sits in the driver's seat, and during actual driving, the driver's identity is checked and the driver is driven. The driver's drinking or other situations are determined in real time and transmitted to the management center (80).

The tablet 10 is installed in a vehicle, and the operating system must be Android 10 or higher and Windows 10 or higher. The breathalyzer module, which is a device for performing a breathalyzer test, is also separately installed in the vehicle, and the breathalyzer module and tablet are installed separately in the vehicle. By connecting in Bluetooth format, alcohol consumption can be measured immediately in the vehicle.

The facial recognition module 20 is installed on one side of the front of the tablet 10 and detects the driver's movements from the time of boarding the vehicle after going to work, and uses CNN, an Ai algorithm, or YOLO algorithm to detect the driver's movements. After facial recognition is recorded as a video, it is transmitted to the control module (60). The control module 60 compares and determines the driver's facial recognition information transmitted from the facial recognition module 20 and the driver's facial information stored in the database server 70, and then determines whether the driver who performed facial recognition Determine whether the person is the same as the driver assigned to operate the vehicle.

The facial recognition module 20 operates using a video recording method. Preferably, the facial recognition module 20 of the present invention can utilize a convolutional neural network (CNN) model.

Typically, face recognition technology basically detects the face area that exists in an input still image or video like an image and performs preprocessing such as face alignment, luminance correction, and image normalization using landmark information extracted from the face area. . Next, features are extracted from the preprocessed face image, a classification model is learned in the training stage using these features, and this model is used in the verification stage to determine what kind of person the person is. Face recognition technology is divided into face verification and face identification technology depending on application.

Face verification is a 1:1 verification problem that determines whether two input face images are the same person, and face identification determines which of the N pre-registered faces a single input face image corresponds to. It can be seen as a 1:N problem.

Usually, face identification technology is performed by comparing probe images, which are the subject of verification, using a learned model, and gallery images of N people registered in advance.

In existing traditional face recognition technologies, technologies for extracting distinguishable features from face images (e.g., Local Binary Patterns, Handcrafted features such as Gabor features) and classification models to determine what kind of person the extracted features are were used. .

Methods using these existing computer vision and pattern recognition technologies are known to have very poor performance due to inconsistencies between the data used for learning and the input test images when various changes occur in facial images in a real environment. In contrast, in deep learning-based face recognition technology, feature extraction and classification model learning are performed end-to-end using large amounts of face data constructed in various environments, so very high-dimensional features are learned on their own. Accordingly, face recognition performance in wild environments is greatly improving, and research cases that surpass human recognition performance are also appearing.

The facial recognition module 20 provided in the present invention utilizes a convolutional neural network (CNN) using an AI algorithm.

A convolutional neural network (CNN) using AI algorithms is a type of multi-layer feed-forward artificial neural network used to analyze visual images. It is a deep neural network technique that can effectively process images by applying filtering techniques to artificial neural networks. It is a technique that classifies images through a process in which each element of the filter expressed in a matrix is automatically learned to be suitable for data processing.

CNN is made up of connections between layers. Specifically, you can see that the convolution layer and pooling layers are connected. However, a CNN is not composed of only a convolution layer and a pooling layer. CNN is largely divided into two types. One consists of feature extraction, which extracts features, and the other consists of classification, which derives the result after passing feature extraction. At this time, feature extraction is a mixture of a convolution layer and a pooling layer, and classification is a fully-connected layer.

When using CNN, the driver's facial image information is acquired, converted to the specifications of the CNN structure, then converted into data that can be classified into preset classes using a 3D convolutional neural network, and the accuracy of each class is calculated. Afterwards, the class with the highest value is determined and learned, and the learned facial image data information is stored and processed on the database server.

For example, the facial recognition module consists of a detection unit that detects the entire outline of the driver's face, a facial feature detection unit that detects features on the driver's face, and a facial asymmetry detection unit that detects the driver's facial asymmetry. By collecting image features, the driver's facial information is converted into the most complete form of image data and then transmitted to the control module. And the control module compares the driver's information stored in the database server with the driver's facial recognition information transmitted from the facial recognition module to determine whether the driver is the same person. In other words, after using a convolutional neural network (CNN) to store the facial image data of all drivers belonging to the company in the database server, when the driver whose dispatch has been confirmed begins to board the vehicle, the facial recognition module is activated for the corresponding driver. The driver's facial features are captured by detecting the driver's movements, then collected and transmitted in real time to the control module. The control module combines the driver's facial image data information stored in the database server with the driver's facial features stored in the vehicle. By comparing facial images, it is determined whether the driver on board is the same person as the driver driving the assigned vehicle.

Additionally, the facial recognition module 20 provided in the present invention can utilize the deep learning-based YOLO (You Only Look Once) algorithm. The YOLO algorithm divides the original image into grids of equal size. For each grid, the number of bounding boxes designated in a predefined form centered on the center of the grid is predicted and reliability is calculated based on this. This includes whether the image contains an object or only the background, and identifies the object category by selecting a location with high object confidence.

Like most image detection models, YOLO is based on the backbone model. The role of this model is to extract meaningful features from the image to be used in the final layer. For this reason, the backbone model is also called a feature extractor.

In other words, the facial recognition module 20 using the YOLO (You Only Look Once) algorithm detects the driver's movements when a driver belonging to the company begins to board the vehicle and controls the driver's video image information. It is transmitted in real time to the module 60, and the control module 60 compares the facial image data information of the driver stored in the database server with the facial image of the driver riding in the vehicle to determine whether the driver is dispatched. Determine whether the person is the same as the driver operating the vehicle.

The camera module 30 is installed on the other front side of the terminal 10, photographs the driver, and transmits the captured image to the control module 60. The control module 60 compares the image transmitted from the camera module 30 and the driver's face stored in the database server 70 to determine whether the driver seated in the driver's seat is the same person as the driver who received the dispatch. do.

In particular, the camera module 30 records a video of the driver from the time the facial recognition module 20 operates until the breathalyzer test is completed in the breathalyzer module 40 to prevent other drivers from performing a breathalyzer test on their behalf. Make sure filming continues.

The breathalyzer module 40 is installed at the bottom of one front side of the terminal 10, measures the blood alcohol concentration of the driver whose face has been recognized by the face recognition module 20, and sends the measured value to the control module 60. ) and send it to

The breathalyzer module 40 is a high-precision electrochemical sensor-based breathalyzer. The displayable range is 0.000~0.450%BAC, and the error range is ±0.005%BAC (when measuring 0.100%BAC). The operating temperature is -5~40℃, and the input voltage uses 3 AA alkaline batteries or USB 5V. The main body of the breathalyzer module 40 is molded from synthetic resin capable of preventing moisture, heat, and static electricity.

A breathalyzer module mouthpiece is installed in the breathalyzer module 40, and the breathalyzer module mouthpiece is formed to protrude in the shape of a truncated cone (funnel) on the front of the breathalyzer module.

The breathalyzer module mouthpiece is formed so that the breathalyzer blows into the main body of the breathalyzer module for a breathalyzer test. Unlike the cylindrical (straw-type) mouthpiece, which is generally put in the mouth and exhaled, it is not placed in the mouth. It has a hygienic effect because there is no need to ask questions to measure it.

Additionally, the breathalyzer module mouthpiece is detachable so that it can be replaced periodically.

The breathalyzer module wirelessly connects to the tablet via Bluetooth to transmit and receive breathalyzer data.

The control module 60 is provided with a driver identification unit 62, an image management unit 64, a drunkenness determination unit 66, an attendance processing unit 68, and a breathalyzer management unit 69.

As shown in FIG. 4, the driver identification unit 62 receives the driver's facial recognition information captured by the facial recognition module 20 and compares it with the driver's facial recognition information previously stored in the database server 70. The facial recognition information is received and the driver corresponding to it is identified.

The database server 70 stores and manages employee information including the driver's employee number, affiliation, name, date of birth, vehicle number, etc., and the driver's facial recognition information.

The video management unit 64 receives images captured from the camera module 30, stores them in a database, and manages them.

The drunkenness determination unit 66 determines that the blood alcohol concentration level received from the breathalyzer measurement module 40 is in a drunk state if it exceeds the preset allowable blood alcohol concentration level, and if it is lower than the preset allowable blood alcohol concentration level, it is judged to be in a non-drinking state.

Based on the judgment result of the drunkenness determination unit 66, the attendance processing unit 68 recognizes the relevant transportation worker as attending work only if he or she is not drunk, and recognizes the employee as not attending work if he or she is drunk.

The breathalyzer management unit 69 stores and manages the employee information and measurement information of the driver who was breathalyzed in the database server 70.

The measurement information includes date and time, location, and measurement result, where the measurement result may mean the measurement value.

If the measurement result is a drunk state, the drinking water measurement management unit 69 may be configured to separately manage employee information and measurement information of a drunk transportation worker.

The remote control module 50 remotely controls the vehicle breathalyzer system using Ethernet communication.

This means that immediate recovery can be achieved without an on-site visit through a remote control module using a management center or administrator terminal. Here, the management center may mean a remote location or a network management system (Network Management System, NFS).

This allows immediate recovery by enabling a remote reboot through the remote control module 50 in the event of a failure of the tablet 10 of the vehicle-type breathalyzer system, thereby reducing recovery delay time due to on-site dispatch of failure inspection personnel and Problems that incur costs can be resolved. Therefore, there is an effect of reducing civil complaints due to unusability of the system.

Additionally, the remote control module 50 may be configured to enable not only rebooting but also resetting. Here, rebooting means completely turning the power off and then turning it back on using the power button when there is a problem running a program while the computer is running. Reset means turning the switch off while the computer is turned on when there is a problem running a program while the computer is running. This means restarting the electrical signal. Additionally, the reset may mean returning all or part of the data processing mechanism to a predetermined state.

Meanwhile, the manager's operation management is from the start of the operation until the end of the operation. If the driver exhibits abnormal behavior (such as drowsy driving), it is transmitted in real time through the camera module and a warning is issued through sound, and the current location information is sent to the manager. Make sure it is sent immediately. And from the moment the breathalyzer test is performed, all screens are sent to the manager in real time, and if the driver's abnormal behavior is discovered, it is displayed on the screen for confirmation.

Meanwhile, various other functions can be implemented in tablets. For example, while the driver is driving, the tablet's camera module takes pictures of the driver in real time. While filming is in progress, the driver is still working, so the driver's working hours are automatically calculated. can do.

And to ensure rest time when driving for long periods of time, the tablet sets off an alarm at certain intervals so that the driver can take a break at set times. For example, a 15-minute break is guaranteed when driving for 2 hours, or a 20-minute break is guaranteed when driving for 2 hours and 30 minutes.

In addition, if the driver's abnormal behavior, such as drowsiness or lack of attention to the front, is detected while driving, the driver is notified of this. Instead, the tablet sounds its own alarm sound, allowing the driver to be immediately warned.

Hereinafter, a method of performing a breathalyzer test using the AI facial recognition-based drunk driving zero solution according to the present invention and a method of managing attendance based on this will be described.

The breathalyzer test method using the AI facial recognition-based drunk driving zero solution of the present invention performs the tasks of attendance management when not drinking and when drinking alcohol separately.

When not drinking, the driver gets on the vehicle → Detects the driver's movements → Face recognition → Checks the driver's data → Compares with stored data → Measures alcohol intake → Checks drinking results → If there are no abnormalities → Processing of attendance → Saving and transmitting collected data → End of operation The process consists of checking measurement data and submitting a report → leaving work.

When drinking, the driver gets into the vehicle → Detects the driver's movements → Face recognition → Checks the driver's data → Compares with stored data → Breathe alcohol test → Checks the results of drinking → Detected drunk → Sends a drunk detection alarm to the manager → Instruction to get out of the vehicle → Relevant employee The process consists of contacting and writing a report → returning home.

Below, we will describe in more detail the breathalyzer-based attendance management method using the AI facial recognition-based drunk driving zero solution.

Figure 5 is a block diagram showing the operational flow of the AI facial recognition-based drunk driving zero solution. As shown in Figure 5, when a driver engaged in the transportation industry goes to work and gets on a dispatched vehicle, from this point on, the facial recognition module installed in the driver's seat of the vehicle detects the driver's movements and displays the driver's face as a video. Take a picture and transmit it to the control module (S10).

Next, the control module 60 receives facial recognition information from the facial recognition module 20, compares the facial image data information of the driver stored in the database server 70, and determines whether the driver in the vehicle receives the dispatched vehicle. Determine whether the driver is the same person (S20).

Next, once it is determined to be the same person, the camera module 30 operates to capture a video of the driver in the driver's seat, and transmits the captured video to the control module 60 (S30). The control module 60 compares the video transmitted from the camera module 30 with the driver's face stored in the database server 70 to determine whether the driver is the same person as the driver who received the dispatch. In addition, the camera module 30 records a video of the driver from the time the facial recognition module 20 operates until the breathalyzer test is completed by the breathalyzer module 40 to prevent other drivers from performing a breathalyzer test. This must be done continuously to prevent others from performing breathalyzer tests on your behalf.

Next, before driving, the driver performs a breathalyzer test using the breathalyzer module 40 mounted on the vehicle (S40).

The value measured by the breathalyzer module 40 is transmitted to the control module 60. At this time, if the video information transmitted from the camera module 30 during the breathalyzer test detects someone other than the driver, the breathalyzer test is performed. This is stopped immediately (S50).

Next, the control module 60 determines that the blood alcohol concentration level received from the breathalyzer measurement module 40 is in a drunk state if it exceeds the preset allowable blood alcohol concentration level, and if it is less than the preset allowable blood alcohol concentration level, the control module 60 determines the blood alcohol concentration level in a non-drinking state (S60). .

If it is determined that the user is drunk by exceeding the permitted blood alcohol concentration, a notification of detection of drinking is displayed through the speaker, and a notification message is sent to the administrator (S70).

Additionally, if the allowable blood alcohol concentration is below the standard, the control module 60 stores measurement information including the measurement results in the database server 70 and performs attendance processing (S80). In other words, the control module 60 determines that driving is impossible when the driver is drunk and does not allow attendance, and only recognizes attendance when the driver is not drunk.

Next, the manager checks the breathalyzer data of the drunk driver, prepares a report, and prints it out (S90).

The attached Figure 6 is a block diagram showing the operation flow performed by the manager in the AI facial recognition-based drunk driving zero solution.

When the facial recognition module 20 of the tablet 10 mounted on the vehicle operates and recognizes the driver's face, a notification arrives at the manager's terminal (S110).

The manager checks the terminal to which the notification has arrived (S120) and then checks the notification received on the terminal (S130).

After the driver's breathalyzer test is performed on the breathalyzer module installed in the vehicle, the breathalyzer test result is checked (S140).

If a drunk detection alert sounds on the terminal (S150), the relevant driver and manager are notified and are instructed to get off the vehicle (S160).

Conversely, if the drunk detection notification does not sound (S150), normal driving is instructed (S170), the breathalyzer data is stored in the database server (S180), and the measurement data is viewed and a report is output (S190).

10: Tablet 20: Facial recognition module
30: Camera module 40: Breathalyzer module
50: remote control module 60: control module
70: Database server 80: Management center

Claims (1)

In the AI facial recognition-based drunk driving zero solution that measures whether a driver working in the transportation industry is drinking when he or she gets on the vehicle,
A tablet (10) installed in front of the driver's seat of the vehicle and equipped with a display (12);
A facial recognition module (20) installed at the top of one front side of the tablet (10) and capturing facial recognition information of the driver;
A camera module 30 is installed on the other front side of the tablet 10 and captures a video of the driver seated in the driver's seat;
A breathalyzer module 40 that is mounted on the driver's seat of the vehicle, transmits and receives wirelessly with the tablet 10, and measures the blood alcohol concentration of the driver whose face has been recognized;
It includes a control module 60 that receives and controls information transmitted from the facial recognition module 20, the camera module 30, and the breathalyzer module 40;
The control module 60 compares and determines the driver's facial image information transmitted from the facial recognition module 20 and the driver's information stored in the database server 70 to determine whether the driver riding in the vehicle will be dispatched. Compare and determine whether the driver you received is the same person;
The control module 60 compares and determines the driver's face stored in the database server 70 with the face of the driver seated in the driver's seat, which is photographed and transmitted from the camera module 30;
The control module 60 compares the driver's blood alcohol level measured by the breathalyzer module 40 with the allowable breathalyzer test value to compare and determine whether a breathalyzer test is present or not;
The face recognition module 20 is a CNN (Convolutional neural network) using an AI algorithm or a deep learning system that predicts the number of bounding boxes specified in a predefined form centered on the center of the grid and calculates reliability based on this. It uses one facial recognition measurement method selected from the YOLO (You Only Look Once)-based algorithm;
CNN using the above AI algorithm detects each image feature through a detection unit that detects the entire outline of the driver's face, a facial feature detection unit that detects features on the driver's face, and a facial asymmetry detection unit that detects facial asymmetry of the driver. By collecting and transmitting the information to the control module 60, the control module 60 can accurately compare and determine the transmitted facial image information of the driver with the driver and facial images stored in the database server 70. There is;
The facial recognition module 20 detects the driver's movements from the point where the driver starts boarding the dispatched vehicle and captures the driver's facial image;
The camera module 30 continues to record video of the driver from the time the facial recognition module 20 operates until the breathalyzer test is completed by the breathalyzer module 40;
A remote control module 50 is connected to the control module 60, and the remote control module 50 allows remote control using the administrator's terminal when a malfunction occurs in the tablet 10. Zero drunk driving solution based on AI facial recognition.