Drunk driving identification method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention provides a drunk driving identification method, which comprises the following steps: acquiring a road shot image of a driving vehicle, and determining a driver of the driving vehicle according to the road shot image; according to preset personnel marking information, whether the driver is a marked personnel is determined, and the personnel marking information is obtained by marking the wine-related personnel in the wine-related place; if the driver is the marked person, obtaining driving data of the driver in a preset time and space; and determining whether the driver is in drunk driving according to the driving data. Whether a driver of a driving vehicle is a marked person or not is determined by obtaining a road shot image of the driving vehicle and preset person marking information, and if the driver is the marked person, whether the driver is in drunk driving or not is determined according to driving data of the driver in a preset time and space. In the whole drunk driving recognition process, comprehensive analysis is carried out through face and behavior recognition of a wine-related place and a driving scene, and the recognition efficiency can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of traffic management, and in particular to a drunk driving identification method, device, electronic equipment and storage medium. Background Art

[0002] With the increase in road mileage and the popularization of motor vehicles, the importance of automobiles to human life is increasing. Automobiles have become an indispensable means of transportation in human life, but traffic safety problems have arisen in an endless stream. Among them, traffic accidents caused by drunk driving and drunk driving account for the highest proportion, far higher than the proportion of traffic accidents caused by speeding, driving in the wrong direction and other problems. Therefore, it is necessary to strictly control and investigate drunk driving and drunk driving to avoid traffic accidents caused by drunk driving and drunk driving. However, drunk driving and drunk driving are extremely difficult to investigate. The existing investigation method is often based on comprehensive analysis of historical experience and statistical data by traffic managers, setting up checkpoints in advance at intersections and other places, and manually checking drunk driving and drunk driving with handheld alcohol detection instruments. However, this method requires a lot of manpower and material resources, and the investigation scope is limited and the efficiency is extremely low. Summary of the invention

[0003] The embodiment of the present invention provides a method for identifying drunk driving, aiming to solve the problem that the existing drunk driving screening requires a lot of manpower, is inefficient, and has a limited screening scope. By acquiring the road shooting image of the moving vehicle and the preset personnel marking information, it is determined whether the driver of the moving vehicle is a marked person. If the driver is a marked person, it is determined whether the driver is drunk driving based on the driver's driving data in a preset time and space. The entire drunk driving identification process is comprehensively analyzed through face and behavior recognition of alcohol-related places and driving scenes, which can reduce manpower investment while realizing a large range of intelligent drunk driving analysis and improving identification efficiency.

[0004] In a first aspect, an embodiment of the present invention provides a method for identifying drunk driving, the method comprising:

[0005] Acquire a road shot image of a moving vehicle, and determine the driver of the moving vehicle based on the road shot image;

[0006] Determine whether the driver is a marked person according to preset person marking information, wherein the person marking information is obtained by marking the person involved in alcohol in the alcohol-related place;

[0007] If the driver is a marked person, obtaining the driving data of the driver in a preset time and space;

[0008] Determine whether the driver is driving under the influence of alcohol based on the driving data.

[0009] Optionally, before acquiring a road shot image of a moving vehicle and determining a driver of the moving vehicle according to the road shot image, the method further includes:

[0010] Acquire a sequence of photographed images of target persons in the alcohol-related venue;

[0011] Determining the drinking confidence level corresponding to the target person based on the captured image sequence;

[0012] When the drinking confidence is greater than or equal to a first confidence threshold, determining the target person as a marked person;

[0013] Obtain the personnel marking information corresponding to the marked personnel.

[0014] Optionally, determining the drinking confidence level corresponding to the target person based on the captured image sequence includes:

[0015] Based on the captured image sequence, determining facial features and behavioral features of the target person;

[0016] The drinking confidence of the target person is determined based on the facial features and behavioral features.

[0017] Optionally, the person marking information includes an identity of the marked person, and determining whether the driver is the marked person according to the preset person marking information includes:

[0018] Obtaining the identity of the driver;

[0019] Comparing the driver's identity with the marked person's identity to obtain a comparison result;

[0020] According to the comparison result, it is determined whether the driver is a marked person.

[0021] Optionally, if the driver is a marked person, obtaining driving data of the driver in a preset time and space includes:

[0022] Performing road collection according to the time information and spatial information of the traveling vehicle to obtain a road collection result;

[0023] Determining the driver's behavior data and the driving data of the driving vehicle according to the road collection result;

[0024] The driving data is determined based on the behavior data of the driver and the driving data of the driving vehicle.

[0025] Optionally, determining whether the driver is driving under the influence of alcohol based on the driving data includes:

[0026] Determining a drunk driving confidence level corresponding to the driver based on the driving data;

[0027] When the drunk driving confidence level is greater than or equal to a second confidence level threshold, determining that the driver is drunk driving;

[0028] When the drunk driving confidence level is less than a second confidence level threshold, it is determined that the driver is not drunk driving.

[0029] Optionally, the driving data includes behavior data of the driver and driving data of the driving vehicle, and determining the drunk driving confidence level corresponding to the driver based on the driving data includes:

[0030] Determining the behavior characteristics of the driver based on the behavior data of the driver, and determining the driving characteristics of the driving vehicle based on the driving data of the driving vehicle;

[0031] The drunk driving confidence level corresponding to the driver is determined according to the behavior characteristics and the driving characteristics.

[0032] In a second aspect, an embodiment of the present invention further provides a drunk driving identification device, the drunk driving identification device comprising:

[0033] A first acquisition module is used to acquire a road shot image of a moving vehicle and determine the driver of the moving vehicle according to the road shot image;

[0034] A first determination module is used to determine whether the driver is a marked person according to preset person marking information, wherein the person marking information is obtained by marking alcohol-related persons in alcohol-related places;

[0035] A second acquisition module is used to acquire driving data of the driver within a preset time and space if the driver is a marked person;

[0036] The second determination module is used to determine whether the driver is driving under the influence of alcohol based on the driving data.

[0037] In a third aspect, an embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the drunk driving identification method provided in an embodiment of the present invention when executing the computer program.

[0038] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the drunk driving identification method provided in the embodiment of the invention are implemented.

[0039] In an embodiment of the present invention, a road shot image of a moving vehicle is obtained, and the driver of the moving vehicle is determined based on the road shot image; based on preset personnel marking information, it is determined whether the driver is a marked person, and the personnel marking information is obtained by marking the alcohol-related person in the alcohol-related place; if the driver is a marked person, the driving data of the driver in a preset time and space is obtained; and based on the driving data, it is determined whether the driver is driving under the influence. By obtaining a road shot image of a moving vehicle and preset personnel marking information, it is determined whether the driver of the moving vehicle is a marked person. If the driver is a marked person, it is determined whether the driver is driving under the influence based on the driving data of the driver in the preset time and space. The entire drunk driving identification process is comprehensively analyzed through face and behavior recognition of alcohol-related places and driving scenes, which can achieve a large range of intelligent drunk driving analysis while reducing manpower input and improving identification efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 is a flow chart of a drunk driving identification method provided by an embodiment of the present invention;

[0042] Figure 2 is a structural schematic diagram of a drunk driving identification device provided in an embodiment of the present invention;

[0043] Figure 3 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] like Figure 1 As shown, Figure 1 : is a flow chart of a drunk driving identification method provided by an embodiment of the present invention. The drunk driving identification method comprises the steps of:

[0046] 101. Obtain a road shot image of a moving vehicle, and determine a driver of the moving vehicle according to the road shot image.

[0047] In an embodiment of the present invention, the above-mentioned drunk driving identification method can be deployed in a traffic management platform. The above-mentioned traffic management platform can obtain the positioning and navigation data of the above-mentioned moving vehicle. The positioning and navigation data of the above-mentioned moving vehicle can be the positioning and navigation data of the above-mentioned moving vehicle determined based on GPS or Beidou satellite navigation. The above-mentioned vehicle management platform stores vehicle registration information and personnel identity identification. The above-mentioned vehicle registration information may include vehicle identification, vehicle owner identification, and vehicle parameters. The above-mentioned personnel identity identification may be a personnel identity identification and a personnel facial image, wherein each personnel identity identification corresponds to a personnel facial image, and the corresponding personnel identity identification can be determined based on the personnel facial image.

[0048] The above-mentioned moving vehicle can be any vehicle traveling on the road, and the above-mentioned any vehicle can be any vehicle that can be driven by a person, for example, it can be a small motor vehicle, a large passenger car, a large truck, and other vehicles that can be driven by a person. The above-mentioned road can be any road, for example, it can be any road installed with a road camera, such as a highway, a national road, a provincial road, an urban road, a suburban road, a rural road, etc. The above-mentioned road shot image includes the shooting time and the shooting location. Specifically, each frame of the above-mentioned road shot image can include the corresponding shooting time and shooting location. The above-mentioned traffic management platform can be an electronic device with image processing, image storage, image analysis and other functions, for example, it can be a server or server cluster with image processing, image storage, image analysis and other functions.

[0049] The vehicle identification may be a mainland China license plate number or a Hong Kong license plate number, the vehicle owner identification may be a vehicle owner identification, and the vehicle parameters may be vehicle color, vehicle model, vehicle age, vehicle size, etc. The road shooting image may be obtained by shooting an image by the road camera in any of the roads installed with the road camera, and uploading the image to the vehicle management platform. The image upload may be uploaded to the vehicle management platform via a data transmission protocol such as SFTP (Secure File Transfer Protocol) or SCP (Secure Copy Protocol).

[0050] Before uploading the above-mentioned images to the above-mentioned vehicle management platform, it is also necessary to identify in real time through the above-mentioned road camera whether there is an object approaching. If there is an object approaching, it is further identified whether the above-mentioned object is a moving vehicle. If the above-mentioned object is a moving vehicle, the start and end times of the above-mentioned road shooting images that need to be uploaded are determined according to the start time and end time of identifying the above-mentioned moving vehicle.

[0051] The above-mentioned determination of the driver of the moving vehicle based on the road shot image can be performed by comparing the image of the person located at the main driving position in the above-mentioned road shot image with the person's face image in the above-mentioned vehicle management platform, and determining the identity of the person in the main driving position based on the person's identity corresponding to the above-mentioned person's face image, and determining the driver of the above-mentioned moving vehicle based on the identity of the person in the main driving position.

[0052] In a possible embodiment, when the road camera recognizes that an object is approaching, the start and end times of the road images that need to be uploaded to the vehicle management platform are determined according to the start time and end time when the road camera recognizes the object. Specifically, the start time of the road images that need to be uploaded to the vehicle management platform can be determined according to the start time when the road camera recognizes the object, and the end time of the road images that need to be uploaded to the vehicle management platform can be determined according to the end time when the road camera recognizes the object.

[0053] After determining the road shooting image that needs to be uploaded to the vehicle management platform according to the start time and end time when the road camera detects an approaching object, the road shooting image is transmitted to the vehicle management platform through the data transmission protocol. After the road shooting image is transmitted to the vehicle management platform through the data transmission protocol, the road shooting image is compared with the facial image of the person stored in the vehicle management platform. If the road shooting image is successfully compared with any of the facial images of the person stored in the vehicle management platform, the person identity corresponding to the facial image is determined according to the facial image, and the driver of the moving vehicle is determined according to the facial image.

[0054] 102. Determine whether the driver is a marked person based on preset person marking information.

[0055] In an embodiment of the present invention, the above-mentioned personnel marking information is obtained by marking the alcohol-related personnel in the alcohol-related places. The above-mentioned personnel marking information may be marking information such as the identity identification of the alcohol-related personnel, the facial image of the alcohol-related personnel, etc. The facial image of the alcohol-related personnel may also include the time information of the facial image of the alcohol-related personnel and the location information of the facial image of the alcohol-related personnel. The location information corresponds to the alcohol-related places. The above-mentioned preset personnel marking information may be personnel marking information within twenty-four hours or forty-eight hours. The above-mentioned alcohol-related places may be key alcohol-related places such as restaurants, bars, pubs, KTVs, etc. The method for determining the above-mentioned alcohol-related personnel may be to determine the behavioral characteristics and facial characteristics of the alcohol-related personnel in the alcohol-related places. The above-mentioned behavioral characteristics may be behavioral characteristics such as staggering when walking, the person needs to be supported by others, vomiting, slow movement, etc. The above-mentioned facial characteristics may be facial characteristics such as flushed face, bloodshot eyes, stiff facial expression or dull eyes.

[0056] Specifically, the above-mentioned marking can be obtained by collecting the alcohol-related images of the above-mentioned alcohol-related persons through cameras arranged in public places such as the above-mentioned alcohol-related places, and transmitting the above-mentioned alcohol-related images to the above-mentioned traffic management platform through a data transmission protocol. The above-mentioned traffic management platform can be deployed with a trained alcohol-related detection model, and the above-mentioned alcohol-related images are detected by the above-mentioned trained alcohol-related detection model, and the alcohol-related persons in the above-mentioned alcohol-related images are marked to obtain the personnel marking information. The above-mentioned marking can also be obtained by any person shooting and collecting the alcohol-related images of the above-mentioned alcohol-related persons through any terminal or shooting device, and applying for access to the above-mentioned traffic management platform to obtain upload authority for the above-mentioned alcohol-related images through any of the above-mentioned terminals or shooting devices, and transmitting the above-mentioned alcohol-related images to the above-mentioned traffic management platform through a preset data transmission protocol to realize reporting, and the above-mentioned alcohol-related images are detected by the trained alcohol-related detection model deployed in the above-mentioned traffic management platform, and the alcohol-related persons in the above-mentioned alcohol-related images are marked to obtain the above-mentioned personnel marking information, and the above-mentioned personnel marking information is stored in the above-mentioned traffic management platform.

[0057] For example, the alcohol-related detection model can be a deep convolutional neural network. Specifically, the alcohol-related detection model can be a model constructed based on deep convolutional neural networks such as ResNet, YOLOV, Faster R-CNN, SSD, and RetinaNet. The trained alcohol-related detection model can be obtained through supervised training of a large number of sample image sequences. The sample image sequences can be image sequences with behavioral characteristics of people involved in alcohol, image sequences with facial characteristics of people involved in alcohol, and image sequences without behavioral characteristics of people involved in alcohol, and image sequences without facial characteristics of people involved in alcohol. The label data corresponding to the image sequences with behavioral characteristics of people involved in alcohol are alcohol-related, the label data corresponding to the image sequences with facial characteristics of people involved in alcohol are alcohol-related, the label data corresponding to the image sequences without behavioral characteristics of people involved in alcohol are not alcohol-related, and the label data corresponding to the image sequences without facial characteristics of people involved in alcohol are not alcohol-related. During the training process, the sample image sequence is input into the detection model to be trained, and the loss is calculated between the output results of the detection model to be trained and the labels corresponding to the sample image sequence to obtain the loss function. The minimum loss function is taken as the optimization goal, and the parameters of the detection model to be trained are adjusted through the back propagation algorithm. The above parameter adjustment process is iterated until the detection model to be trained converges at the minimum loss function, or the number of iterations reaches a preset number, the training is stopped, and the trained detection model is obtained, and the trained detection model is deployed to the above traffic management platform.

[0058] In a possible embodiment, after the driver of the moving vehicle is determined based on the road shooting image, the above-mentioned personnel marking information and the identity of the above-mentioned driver or the facial image of the driver are obtained, and the identity of the above-mentioned driver is compared with the identity of the alcohol-related person in all the personnel marking information. If the identity comparison is successful, the above-mentioned driver is judged as a marked person, and if the identity comparison fails, the above-mentioned driver is judged as a non-marked person. Alternatively, the facial image of the above-mentioned driver can be compared with the facial images of the alcohol-related persons in all the personnel marking information. If the facial comparison is successful, the above-mentioned driver is judged as a marked person, and if the facial comparison fails, the above-mentioned driver is judged as a non-marked person.

[0059] In another possible embodiment, due to the different alcohol tolerance and alcohol metabolism rates of different people, it generally takes twenty-four hours or forty-eight hours to completely decompose the alcohol. Therefore, the above-mentioned marking information needs to be time-effective. After the driver of the moving vehicle is determined based on the road shooting image, the driver of the moving vehicle is compared with the personnel marking information within the above twenty-four hours or forty-eight hours. If the facial image of the drunk person in the personnel marking information within the above twenty-four hours or forty-eight hours is successfully compared with the facial image of the driver of the driving vehicle, then the driver of the driving vehicle is judged to be the marked person. If the facial image of the drunk person in the personnel marking information within the above twenty-four hours or forty-eight hours is successfully compared with the facial image of the driver of the driving vehicle, then the driver of the driving vehicle is judged to be the marked person. If the facial image comparison of the driver of the vehicle fails, it is judged that the driver of the above-mentioned vehicle is not the marked person. The above-mentioned comparison can also be to compare the identity identification of the driver of the above-mentioned vehicle with the identity identification of the alcohol-related person in the above-mentioned personnel marking information within twenty-four hours or forty-eight hours. If the identity identification of the driver of the above-mentioned vehicle is successfully compared with the identity identification of the alcohol-related person in the above-mentioned personnel marking information within twenty-four hours or forty-eight hours, it is judged that the driver of the above-mentioned vehicle is the marked person. If the identity identification of the driver of the above-mentioned vehicle is successfully compared with the identity identification of the alcohol-related person in the above-mentioned personnel marking information within twenty-four hours or forty-eight hours, it is judged that the driver of the above-mentioned vehicle is not the marked person.

[0060] 103. If the driver is a marked person, the driving data of the driver within a preset time and space is obtained.

[0061] In an embodiment of the present invention, the driving data may be driving behavior characteristics and driver behavior characteristics. The driving behavior characteristics may be speeding, driving a vehicle off the lane, and driving a vehicle changing lanes back and forth. The driver behavior characteristics may be yawning, closing eyes, and other characteristics. The driving data may be acquired by calling the road camera image of the road management platform. The driving data may also be acquired by a person taking a road image of the driving vehicle through any terminal or shooting device, and applying for access to the traffic management platform to obtain upload permission for the road image through any terminal or shooting device, and transmitting the road image to the traffic management platform through a preset data transmission protocol. The speeding of the driving vehicle may also be measured and determined based on the speed sensor or the positioning navigation data. The driving data in the preset time and space may be determined based on the location of the alcohol-related venue and the location of the road camera that photographed the driving vehicle to determine the spatial range of the driving data, and the time range of the driving data may be determined based on the time when the camera arranged in the public places such as the alcohol-related venue photographed the marked person and the time when the road camera photographed the driving vehicle.

[0062] For example, when the location of the above-mentioned alcohol-related venue is ten kilometers away from the location of the road camera that photographed the above-mentioned moving vehicle, the location of the above-mentioned alcohol-related venue can be used as the first central location, and the first spatial range can be determined based on the ten-kilometer radius of the above-mentioned first central location. The location of the road camera that photographed the above-mentioned moving vehicle can be used as the second central location, and the second spatial range can be determined based on the ten-kilometer radius of the second central location. The first spatial range is added to the second spatial range to determine the spatial range of the above-mentioned driving data. When the time when the cameras arranged in public places such as the above-mentioned alcohol-related venues photograph the above-mentioned marked persons is four hours away from the time when the above-mentioned road cameras photographed the above-mentioned moving vehicle. Then, the time when the marked person is photographed by the cameras arranged in the above-mentioned public places such as the alcohol-related places can be used as the first central time, and the first time range can be determined based on the four hours before and after the time when the marked person is photographed by the cameras arranged in the above-mentioned public places such as the alcohol-related places. The time when the moving vehicle is photographed by the above-mentioned road camera is used as the second central time, and the second time range can be determined based on the four hours before and after the time when the moving vehicle is photographed by the above-mentioned road camera. The time range of the above-mentioned driving data is determined by adding the above-mentioned first time range to the above-mentioned second time range, and the driving data within the above-mentioned preset time and space is determined based on the time range of the above-mentioned driving data and the spatial range of the above-mentioned driving data.

[0063] 104. Determine whether the driver is driving under the influence of alcohol based on driving data.

[0064] In an embodiment of the present invention, a drunk driving judgment can be made based on the driving behavior characteristics and the driver behavior characteristics in the driving data to determine whether the driver is drunk driving. Specifically, the driving data includes vehicle driving data and driving behavior data, and corresponding weights can be set based on the behavior characteristics of the vehicle driving data such as speeding, driving the vehicle off the lane, and driving the vehicle back and forth, and the driver yawning, closing the eyes of the driver in the driving behavior data to determine whether the driver is drunk driving. The weights include driving behavior feature weights and driver behavior feature weights. The driving behavior feature weights include speeding weights, lane deviation weights, and back and forth lane changing weights. The driver behavior feature weights include driver yawning weights and driver eye closing weights.

[0065] The above driving behavior features such as speeding, driving off the lane, driving back and forth lane change and the above driver behavior features such as driver yawning and driver closing eyes can be represented by string label values, which is convenient for the final drunk driving judgment. The above string includes feature labels and result labels, wherein the above feature labels include 000, 001, 010, 011, 100, the above 000 is the feature label of speeding, the above 001 is the label of driving off the lane, the above 010 is the label of driving back and forth lane change, the above 011 is the label of driver yawning, and the above 100 is the label of driver closing eyes. The above result labels include 0 and 1. If the label value corresponding to the result label in the above driving data is 0, it means that the driving behavior feature or driver behavior feature corresponding to the above label value does not appear in the above driving data. If the label value corresponding to the result label in the above driving data is 1, it means that the driving behavior feature or driver behavior feature corresponding to the above label value appears in the above driving data.

[0066] The speeding of the driving vehicle can be determined according to the speed limit regulations of the driving section where the driving vehicle is located. For example, the speed limit of the driving section where the driving vehicle is located is 100 kilometers per hour, and according to the above-mentioned positioning navigation data, it is determined that the speed of the driving vehicle on the driving section is 110 kilometers per hour, then the above-mentioned driving data contains the feature of speeding of the driving vehicle. The above-mentioned driving vehicle deviating from the lane can be determined according to the lane width of the driving section where the driving vehicle is located and the distance of the driving section where the driving vehicle is located. For example, the body width of the driving vehicle is 2 meters, the lane width of the driving section where the driving vehicle belongs is 2.5 meters, the driving vehicle occupies 1 meter of one lane of the driving section and 1.5 meters of another lane of the driving section, and the driving vehicle occupies 1 meter of one lane of the driving section and 1.5 meters of another lane of the driving section, and the driving distance exceeds 20 meters, then the above-mentioned driving data contains the driving vehicle deviating from the lane. The lane-off-lane driving feature, the lane-changing of the driving vehicle back and forth can be determined based on the lane and driving distance of the driving section where the driving vehicle is located. For example, when the driving vehicle is driving on the driving section, the driving distance is within 100 meters, and the lane where the driving vehicle is located changes back and forth more than 3 times, then it is determined that the driving data contains the lane-changing feature of the driving vehicle back and forth. The yawning of the driver can be determined based on the facial image of the driver in the driving behavior data. For example, when it is identified that the driver yawns more than 15 times continuously within 1 hour in the facial image of the driver, then it is determined that the driving data contains the yawning feature of the driver. The closing of the eyes of the driver can be determined based on the facial image of the driver in the driving behavior data. For example, when it is identified that the driver's eyes are closed for more than 5 seconds in the facial image of the driver and the number of times the eyes are closed for more than 5 seconds within 1 hour exceeds 15 times, then it is determined that the driving data contains the closing of the eyes of the driver.

[0067] In a possible embodiment, the drunk driving determination may be based on the driving behavior characteristics of speeding, driving the vehicle off the lane, driving the vehicle changing lanes back and forth and the driver behavior characteristics of yawning, closing the eyes, flushing of the driver and other characteristics to set corresponding drunk driving weights, and determine whether the driver is drunk driving according to the corresponding drunk driving weights. The drunk driving weight may be when the driving vehicle has a speeding characteristic value, the drunk driving weight is 30%, the drunk driving weight may be when the driving vehicle has a lane deviation characteristic, the drunk driving weight is 15%, the drunk driving weight may be when the driving vehicle has a lane change characteristic, the drunk driving weight is 15%, and the The drunk driving weight may be 10% when the driver yawns, 10% when the driver flushes, and 20% when the driver closes his eyes. A corresponding drunk driving weight threshold is set based on the drunk driving weight. If the driving behavior characteristics are added to the weights corresponding to the driver behavior characteristics, a total weight value is obtained. The total weight value is compared with the drunk driving weight threshold. If the total weight value is less than the drunk driving weight threshold, it is determined that the driver is not drunk driving. If the total weight value is greater than the drunk driving weight threshold, it is determined that the driver is drunk driving.

[0068] In an embodiment of the present invention, a road shot image of a moving vehicle is obtained, and the driver of the moving vehicle is determined based on the road shot image; based on preset personnel marking information, it is determined whether the driver is a marked person, and the personnel marking information is obtained by marking the alcohol-related person in the alcohol-related place; if the driver is a marked person, the driving data of the driver in the preset time and space is obtained; and it is determined whether the driver is driving under the influence of alcohol based on the driving data. By obtaining a road shot image of a moving vehicle and preset personnel marking information, it is determined whether the driver of the moving vehicle is a marked person. If the driver is a marked person, it is determined whether the driver is driving under the influence of alcohol based on the driving data of the driver in the preset time and space. The entire drunk driving identification process is comprehensively analyzed through face and behavior recognition of alcohol-related places and driving scenes, which can achieve a large range of intelligent drunk driving analysis while reducing manpower input and improving identification efficiency.

[0069] Optionally, before obtaining road-shot images of a moving vehicle and determining the driver of the moving vehicle based on the road-shot images, a sequence of shot images of a target person in a drinking venue may be obtained; based on the shot image sequence, a drinking confidence level corresponding to the target person may be determined; when the drinking confidence level is greater than or equal to a first confidence threshold, the target person may be determined as a marked person; and the person marking information corresponding to the marked person may be obtained.

[0070] In an embodiment of the present invention, the target person may be any person in the alcohol-related venue, the captured image sequence may be images of the alcohol-related person collected by cameras arranged in public places such as the alcohol-related venue, the captured image sequence may also be images of the alcohol-related person collected by a person using any terminal or shooting device, the first confidence threshold may be set according to the drinking confidence, the person characteristic behavior confidence and the person facial feature confidence may be obtained by output of the alcohol-related detection model, the drinking confidence may be determined according to the person behavior characteristic confidence corresponding to the person behavior characteristic and the person facial feature confidence corresponding to the person facial feature, and the drinking confidence may be determined according to the behavior characteristic confidence and the drunk driving weight.

[0071] For example, the target person can be determined as a marked person based on the drinking confidence level. If the target person is a marked person, the identity identification or facial image of the marked person is obtained, and the target person is determined as a marked person and stored in a marked person database. The validity period of the marked persons in the marked person database can be a preset validity period, such as 24 hours, 12 hours, etc. The validity period can also be determined based on the drinking confidence level. The greater the drinking confidence level, the longer the validity period.

[0072] In a possible embodiment, after a sequence of captured images of at least one of the target persons is collected in the alcohol-related venue through the alcohol-related camera or any of the terminals or shooting devices, the drinking confidence is determined based on the confidence of the behavioral characteristics and facial features of the target person, and the drinking confidence is compared with the first confidence threshold. When the drinking confidence is greater than or equal to the first confidence threshold, the target person is determined as a marked person, and the person marking information corresponding to the marked person is obtained through the traffic management platform.

[0073] Optionally, in the step of determining the drinking confidence level corresponding to the target person based on the captured image sequence, the facial features and behavioral features of the target person may be determined based on the captured image sequence; and the drinking confidence level of the target person may be determined based on the facial features and behavioral features.

[0074] In an embodiment of the present invention, the facial features of the target person may be flushed face, bloodshot eyes, stiff facial expression or dull eyes, and the behavioral features of the target person may be unsteady walking, needing help from others, vomiting, slow movement, and the like.

[0075] In a possible embodiment, after a sequence of captured images of the target person is collected in the alcohol-related venue through the alcohol-related camera or any of the terminals or shooting devices, the facial features of the target person and the behavioral features of the target person are determined based on the sequence of captured images, the facial feature confidence level and the behavioral feature confidence level of the target person are determined based on the facial features of the target person and the behavioral features of the target person, and the drinking confidence level of the target person is determined based on the facial feature confidence level and the behavioral feature confidence level of the target person.

[0076] Optionally, in the step of determining whether the driver is the marked person based on preset personnel marking information, where the personnel marking information includes the identity of the marked person, the identity of the driver can be obtained; the identity of the driver is compared with the identity of the marked person to obtain a comparison result; and based on the comparison result, it is determined whether the driver is the marked person.

[0077] In an embodiment of the present invention, the above-mentioned identity identification may be pre-stored in the above-mentioned traffic management platform, the above-mentioned driver identity identification may be obtained by extracting the facial features of the driver in the above-mentioned road shooting image and the facial features of the ID card facial image corresponding to the above-mentioned person through a face recognition algorithm based on the above-mentioned road shooting image, and comparing the facial features of the driver in the above-mentioned road shooting image with the facial features of the ID card facial image corresponding to the driver to determine the identity identification of the above-mentioned driver, and the identity identification of the above-mentioned marked person may be obtained by extracting the facial features of the marked person in the above-mentioned shot image sequence and the facial features of the ID card facial image corresponding to the above-mentioned person through a face recognition algorithm based on the above-mentioned shot image sequence, and comparing the facial features of the marked person in the above-mentioned shot image sequence with the facial features of the ID card facial image corresponding to the above-mentioned person to determine the identity identification of the above-mentioned marked person.

[0078] In a possible embodiment, after obtaining a road-shot image of the moving vehicle and determining the driver of the moving vehicle based on the road-shot image, the facial features of the ID card facial image corresponding to the identity identifier pre-stored in the traffic management platform are compared with the facial features of the driver and the facial features of the marked person, respectively, and the identity identifier whose facial features of the driver are consistent with the facial features of the ID card facial image is determined as the identity identifier of the driver, and the identity identifier whose facial features of the marked person are consistent with the facial features of the ID card facial image is determined as the identity identifier of the marked person, and the ID card number corresponding to the identity identifier of the marked person is compared with the ID card number of the driver. If the ID card number of the marked person is consistent with the ID card number of the driver, the driver is determined to be the marked person, otherwise the driver is determined to be the unmarked person.

[0079] Optionally, if the driver is a marked person, in the step of obtaining the driver's driving data within a preset time and space, road collection can be performed based on the time information and spatial information of the moving vehicle to obtain a road collection result; based on the road collection result, the driver's behavior data and the driving data of the moving vehicle are determined; based on the driver's behavior data and the driving data of the moving vehicle, the driving data is determined.

[0080] In an embodiment of the present invention, the above-mentioned time information can be determined based on the time zone where the above-mentioned moving vehicle is located, the above-mentioned spatial information can be determined based on the displayed longitude and latitude position where the above-mentioned moving vehicle is located, and the above-mentioned road acquisition results can include the image sequence captured by the above-mentioned moving vehicle and the above-mentioned moving vehicle speed data.

[0081] For example, when the driver is determined to be a marked person, the shooting time and shooting location of the road shooting image are determined based on the road shooting image corresponding to the traveling vehicle, and the acquisition range of the road acquisition is determined based on the shooting time and shooting location of the road shooting image. The acquisition range can be based on the shooting time of the road shooting image, and the acquisition time of the acquisition range is determined to be one hour before and after the shooting time corresponding to the road shooting image. The acquisition area of ​​the acquisition range can be the area within a radius of five kilometers with the shooting location of the road shooting image as the center of the acquisition area. Based on the acquisition area and the acquisition time, image acquisition is performed through the road camera to obtain the driving image of the traveling vehicle in the acquisition time and acquisition area.

[0082] Based on the above-mentioned driving images, the above-mentioned driving images are recognized and processed by a preset behavior recognition algorithm to obtain the behavior data of the above-mentioned driver and the driving data of the above-mentioned driving vehicle, and the above-mentioned driving data is determined by the above-mentioned driver's behavior data and the driving data of the above-mentioned driving vehicle. The above-mentioned preset behavior recognition algorithm can be LeNet-5 algorithm, AlexNet algorithm, VGGNet algorithm, GoogLeNet (Inception algorithm) and other algorithms that can be used for vehicle and personnel behavior recognition.

[0083] Optionally, in the step of determining whether the driver is driving under the influence of alcohol based on the driving data, the drunk driving confidence level corresponding to the driver can be determined based on the driving data; when the drunk driving confidence level is greater than or equal to a second confidence threshold, the driver is determined to be driving under the influence of alcohol; when the drunk driving confidence level is less than the second confidence threshold, the driver is determined to be driving under the influence of alcohol.

[0084] In an embodiment of the present invention, the above-mentioned drunk driving confidence level can be obtained by outputting the above-mentioned alcohol-related detection model, and the above-mentioned second confidence threshold can be set according to the above-mentioned drunk driving confidence level. If the above-mentioned drunk driving confidence level is greater than or equal to the second confidence threshold level, it is determined that the above-mentioned driver is drunk driving; if the above-mentioned drunk driving confidence level is less than the above-mentioned second confidence threshold level, it is determined that the above-mentioned driver is not drunk driving.

[0085] Specifically, whether the above-mentioned driver is driving under the influence of alcohol can also be calculated according to the following formula:

[0086]

[0087] The above P is the confidence level of drunk driving, Q i is the drinking confidence, W i is the weight of drunk driving, and i is the above-mentioned behavioral characteristics.

[0088] Optionally, in the step of determining the drunk driving confidence level corresponding to the driver based on the driving data, where the driving data includes the driver's behavior data and the driving data of the moving vehicle, the driver's behavior characteristics can be determined based on the driver's behavior data, and the driving characteristics of the moving vehicle can be determined based on the driving data of the moving vehicle; the driver's drunk driving confidence level corresponding to the driver is determined based on the behavior characteristics and the driving characteristics.

[0089] In an embodiment of the present invention, the above-mentioned behavior data and the above-mentioned driving data can be extracted based on the above-mentioned road shooting image to perform behavior data extraction and driving data extraction to obtain a behavior data image corresponding to the behavior data of the above-mentioned driver and a driving data image corresponding to the driving data of the above-mentioned driving vehicle, and feature extraction is performed on the above-mentioned driving data image and the above-mentioned behavior data image through a feature extraction algorithm to obtain the behavior characteristics of the above-mentioned driver and the driving characteristics of the above-mentioned driving vehicle. The above-mentioned feature extraction algorithm can be a MobileNet algorithm, a ResNeXt algorithm, a DenseNet algorithm, and an EfficientNet algorithm.

[0090] In a possible embodiment, after the driving data is determined based on the driver's behavior data and the driving data of the moving vehicle, a behavior data image corresponding to the driver's behavior data can be determined based on the driving data, and a driving data image corresponding to the driving data of the moving vehicle can be determined based on the driving data. Feature extraction is performed based on the behavior data image and the driving data image to obtain the behavior characteristics and the driving characteristics, and the driver's drunk driving confidence is determined based on the behavior characteristics and the driving characteristics.

[0091] It should be noted that the drunk driving identification method provided in the embodiment of the present invention can be applied to devices such as photographing equipment, smart phones, computers, servers, etc. that can perform drunk driving identification.

[0092] like Figure 2 As shown, an embodiment of the present invention provides a drunk driving identification device, the drunk driving identification device comprising:

[0093] A first acquisition module 201 is used to acquire a road shot image of a moving vehicle and determine the driver of the moving vehicle according to the road shot image;

[0094] A first determination module 202 is used to determine whether the driver is a marked person according to preset person marking information, wherein the person marking information is obtained by marking a person involved in alcohol in a place involving alcohol;

[0095] A second acquisition module 203 is used to acquire driving data of the driver in a preset time and space if the driver is a marked person;

[0096] The second determination module 204 is used to determine whether the driver is driving under the influence of alcohol according to the driving data.

[0097] Optionally, the first acquisition module 201 further includes:

[0098] The first acquisition submodule is used to acquire a sequence of photographed images of the target person in the alcohol-related venue;

[0099] A first determination submodule, configured to determine the drinking confidence level corresponding to the target person based on the captured image sequence;

[0100] A second determination submodule, configured to determine the target person as a marked person when the drinking confidence is greater than or equal to a first confidence threshold;

[0101] Obtain the personnel marking information corresponding to the marked personnel.

[0102] Optionally, the first determining submodule includes:

[0103] A first determining unit, configured to determine facial features and behavioral features of the target person based on the captured image sequence;

[0104] The second determination unit is used to determine the drinking confidence of the target person according to the facial features and behavioral features.

[0105] Optionally, the first determining module 202 includes:

[0106] A second acquisition submodule is used to acquire the identity of the driver;

[0107] A first comparison submodule, used for comparing the driver's identity with the marked person's identity to obtain a comparison result;

[0108] The third determination submodule is used to determine whether the driver is a marked person according to the comparison result.

[0109] Optionally, the second acquisition module 203 includes:

[0110] The first acquisition submodule is used to acquire road data according to the time information and spatial information of the traveling vehicle to obtain a road acquisition result;

[0111] A fourth determination submodule, used to determine the behavior data of the driver and the driving data of the driving vehicle according to the road collection result;

[0112] The fifth determination submodule is used to determine the driving data based on the behavior data of the driver and the driving data of the driving vehicle.

[0113] Optionally, the second determining module 204 includes:

[0114] A sixth determination submodule, configured to determine a drunk driving confidence level corresponding to the driver based on the driving data;

[0115] a seventh determination submodule, configured to determine that the driver is driving under the influence of alcohol when the confidence level of driving under the influence of alcohol is greater than or equal to a second confidence threshold;

[0116] The eighth determination submodule is used to determine that the driver is not driving under the influence of alcohol when the confidence level of driving under the influence of alcohol is less than a second confidence threshold.

[0117] Optionally, the sixth determining submodule includes:

[0118] a third determining unit, configured to determine a behavior characteristic of the driver based on the behavior data of the driver, and to determine a driving characteristic of the driving vehicle based on the driving data of the driving vehicle;

[0119] The fourth determination unit is used to determine the drunk driving confidence level corresponding to the driver according to the behavior characteristics and the driving characteristics.

[0120] It should be noted that the drunk driving identification device provided in the embodiment of the present invention can be applied to devices such as photographing equipment, smart phones, computers, servers, etc. that can perform drunk driving identification.

[0121] The drunk driving identification device provided in the embodiment of the present invention can implement each process implemented by the drunk driving identification method in the above method embodiment, and can achieve the same beneficial effects. To avoid repetition, it will not be described here.

[0122] See also Figure 3 , Figure 3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 3 As shown, it includes: a memory 302, a processor 301, and a computer program of a drunk driving identification method stored in the memory 302 and executable on the processor 301, wherein:

[0123] The processor 301 is used to call the computer program stored in the memory 302 and execute the following steps:

[0124] Acquire a road shot image of a moving vehicle, and determine the driver of the moving vehicle based on the road shot image;

[0125] Determine whether the driver is a marked person according to preset person marking information, wherein the person marking information is obtained by marking the person involved in alcohol in the alcohol-related place;

[0126] If the driver is a marked person, obtaining the driving data of the driver within a preset time and space;

[0127] Determine whether the driver is driving under the influence of alcohol based on the driving data.

[0128] Optionally, before acquiring a road shot image of a moving vehicle and determining a driver of the moving vehicle according to the road shot image, the method executed by the processor 301 further includes:

[0129] Acquire a sequence of photographed images of target persons in the alcohol-related venue;

[0130] Determining the drinking confidence level corresponding to the target person based on the captured image sequence;

[0131] When the drinking confidence is greater than or equal to a first confidence threshold, determining the target person as a marked person;

[0132] Obtain the personnel marking information corresponding to the marked personnel.

[0133] Optionally, the processor 301 determines the drinking confidence level corresponding to the target person based on the captured image sequence, including:

[0134] Based on the captured image sequence, determining facial features and behavioral features of the target person;

[0135] The drinking confidence of the target person is determined based on the facial features and behavioral features.

[0136] Optionally, the personnel marking information executed by the processor 301 includes an identity of the marked person, and determining whether the driver is a marked person according to the preset personnel marking information includes:

[0137] Obtaining the identity of the driver;

[0138] Comparing the driver's identity with the marked person's identity to obtain a comparison result;

[0139] According to the comparison result, it is determined whether the driver is a marked person.

[0140] Optionally, if the driver is a marked person, the acquiring of driving data of the driver in a preset time and space performed by the processor 301 includes:

[0141] Performing road collection according to the time information and spatial information of the traveling vehicle to obtain a road collection result;

[0142] Determining the driver's behavior data and the driving data of the driving vehicle according to the road collection result;

[0143] The driving data is determined based on the behavior data of the driver and the driving data of the driving vehicle.

[0144] Optionally, the determining whether the driver is driving under the influence of alcohol according to the driving data performed by the processor 301 includes:

[0145] Determining a drunk driving confidence level corresponding to the driver based on the driving data;

[0146] When the drunk driving confidence level is greater than or equal to a second confidence level threshold, determining that the driver is drunk driving;

[0147] When the drunk driving confidence level is less than a second confidence level threshold, it is determined that the driver is not drunk driving.

[0148] Optionally, the driving data executed by the processor 301 includes behavior data of the driver and driving data of the driving vehicle, and determining the drunk driving confidence level corresponding to the driver based on the driving data includes:

[0149] Determining the behavior characteristics of the driver based on the behavior data of the driver, and determining the driving characteristics of the driving vehicle based on the driving data of the driving vehicle;

[0150] The drunk driving confidence level corresponding to the driver is determined according to the behavior characteristics and the driving characteristics.

[0151] It should be noted that the electronic device provided in the embodiment of the present invention can be applied to smart phones, computers, servers and other devices that can perform the drunk driving identification method.

[0152] The electronic device provided in the embodiment of the present invention can implement each process implemented by the drunk driving identification method in the above method embodiment, and can achieve the same beneficial effects. To avoid repetition, it will not be described here.

[0153] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the drunk driving identification method or the application-side drunk driving identification method provided in the embodiment of the present invention are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0154] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0155] The above disclosure is only the preferred embodiment of the present invention, which certainly cannot be used to limit the scope of the present invention. Therefore, equivalent changes made according to the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for identifying drunk driving, characterized in that: The method comprises the following steps: Acquire a road shot image of a moving vehicle, and determine the driver of the moving vehicle based on the road shot image; Determine whether the driver is a marked person according to preset person marking information, wherein the person marking information is obtained by marking the person involved in alcohol in the alcohol-related place; If the driver is a marked person, obtaining the driving data of the driver within a preset time and space; Determine whether the driver is driving under the influence of alcohol based on the driving data.

2. The drunk driving identification method according to claim 1, characterized in that: Before acquiring the road photographed image of the moving vehicle and determining the driver of the moving vehicle according to the road photographed image, the method further includes: Acquire a sequence of photographed images of target persons in the alcohol-related venue; Determining the drinking confidence level corresponding to the target person based on the captured image sequence; When the drinking confidence is greater than or equal to a first confidence threshold, determining the target person as a marked person; Obtain the personnel marking information corresponding to the marked personnel.

3. The drunk driving identification method according to claim 2, characterized in that: The step of determining the drinking confidence level corresponding to the target person based on the captured image sequence includes: Based on the captured image sequence, determining facial features and behavioral features of the target person; The drinking confidence of the target person is determined based on the facial features and behavioral features.

4. The drunk driving identification method according to any one of claims 1 to 3, characterized in that: The person marking information includes an identity of the marked person, and determining whether the driver is a marked person according to the preset person marking information includes: Obtaining the identity of the driver; Comparing the driver's identity with the marked person's identity to obtain a comparison result; According to the comparison result, it is determined whether the driver is a marked person.

5. The drunk driving identification method according to any one of claims 1 to 4, characterized in that: If the driver is a marked person, obtaining the driving data of the driver in a preset time and space includes: Performing road collection according to the time information and spatial information of the traveling vehicle to obtain a road collection result; Determining the driver's behavior data and the driving data of the driving vehicle according to the road collection result; The driving data is determined based on the behavior data of the driver and the driving data of the driving vehicle.

6. The drunk driving identification method according to any one of claims 1 to 5, characterized in that: The determining whether the driver is driving under the influence of alcohol according to the driving data comprises: Based on the driving data, determining a drunk driving confidence level corresponding to the driver; When the drunk driving confidence level is greater than or equal to a second confidence level threshold, determining that the driver is drunk driving; When the drunk driving confidence level is less than a second confidence level threshold, it is determined that the driver is not drunk driving.

7. The drunk driving identification method according to claim 6, characterized in that: The driving data includes the behavior data of the driver and the driving data of the driving vehicle, and determining the drunk driving confidence level corresponding to the driver based on the driving data includes: Determining the behavior characteristics of the driver based on the behavior data of the driver, and determining the driving characteristics of the driving vehicle based on the driving data of the driving vehicle; The drunk driving confidence level corresponding to the driver is determined according to the behavior characteristics and the driving characteristics.

8. A drunk driving identification device, characterized in that: The drunk driving identification device comprises: A first acquisition module is used to acquire a road shot image of a moving vehicle and determine the driver of the moving vehicle according to the road shot image; A first determination module is used to determine whether the driver is a marked person according to preset person marking information, wherein the person marking information is obtained by marking alcohol-related persons in alcohol-related places; A second acquisition module is used to acquire driving data of the driver within a preset time and space if the driver is a marked person; The second determination module is used to determine whether the driver is driving under the influence of alcohol based on the driving data.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the drunk driving identification method as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the drunk driving identification method according to any one of claims 1 to 7 are implemented.

Images