A bus epidemic monitoring system and method

Abstract

The present invention relates to a bus epidemic monitoring system and method, the system includes a roadbed end, a vehicle-mounted end and a cloud end, wherein: the vehicle-mounted end is used to input the acquired face image into a mask detection network to realize the mask recognition function, and upload the face image to the roadbed end at the same time; the roadbed end is used to extract the facial vector information of the face image and upload it to the cloud end; the cloud end is used to screen the facial vector information through a stored graph database model, generate alarm information, and send it to the vehicle-mounted end and / or the command center. The present invention realizes functions such as epidemic monitoring and regular risk prevention and control, ensures the safety of people on the bus under the epidemic environment, and has low cost.

Description

Technical Field

[0001] The present invention relates to the field of transportation technology, and in particular to a bus epidemic monitoring system and method. Background Art

[0002] The development of urban public transportation indicates the degree of modernization of the city. It not only provides large-capacity transportation, but also improves the travel conditions of residents. Its advantages such as energy saving and congestion relief make public transportation play an indispensable role in modern life.

[0003] Public transportation has wide coverage, many intersections, and weak personnel controllability. If it becomes a route for the spread of the epidemic, it will cause serious consequences with high intensity, wide range, and wide groups. Epidemic infections and diseases may transmit pathogens to susceptible people through direct transmission, contact transmission, aerosol transmission, and other methods. After passengers enter public transportation, their riding space is relatively free, and the boarding time and place are relatively disordered. The "keeping distance" that can effectively control the infection depends more on the passengers' consciousness, and the driver's method of measuring the temperature of passengers is cumbersome and easily affected by the environment. Therefore, if a passenger carries infectious pathogens, it may be transmitted to all corners of the city through the well-connected public transportation routes, bringing great health and safety risks. Therefore, how to effectively eliminate the hidden dangers caused by pathogen carriers to public transportation is an urgent problem to be solved. Summary of the invention

[0004] In view of this, it is necessary to provide a bus epidemic monitoring system and method to overcome the problem in the prior art of lacking a method for efficiently identifying pathogen carriers riding public transportation.

[0005] In order to solve the above technical problems, the present invention provides a bus epidemic monitoring system, including a roadbed end, a vehicle-mounted end and a cloud end, wherein:

[0006] The vehicle-mounted end is used to input the acquired face image into the mask detection network to realize the mask recognition function, and upload the face image to the roadbed end at the same time;

[0007] The roadbed end is used to extract facial vector information of the facial image and upload it to the cloud;

[0008] The cloud is used to screen the facial vector information through a stored graph database model, generate alarm information, and send it to the vehicle-mounted terminal and / or the command center.

[0009] The present invention also provides a bus epidemic monitoring method, which is applied to the roadbed end of the bus epidemic monitoring system as described above, and the method comprises:

[0010] Input the face image uploaded by the vehicle into the trained facial recognition model and output the center point;

[0011] Constructing a three-dimensional face model according to the center point;

[0012] The 3D face model is symmetrically detected to determine corrected facial vector information, which is then uploaded to the cloud.

[0013] Furthermore, the step of inputting the facial image uploaded from the vehicle-mounted end into the fully trained facial recognition model specifically includes: converting the facial image into an HSV color space, and inputting the converted facial color image into the fully trained facial recognition model.

[0014] Furthermore, the fully trained facial recognition model includes a normalization layer, a downsampling layer, a mapping layer, a classification layer, a shift layer, and a regression layer, wherein:

[0015] The normalization layer is used to calculate the normalized confidence of the face color image and determine a normalized image;

[0016] The downsampling layer is used to downsample the normalized image to determine a downsampled image;

[0017] The mapping layer is used to map the similarity of the target point in the downsampled image to the heat map through a Gaussian kernel function, wherein if the channel output confidence of the target point is equal to a preset value, it is determined to be consistent with the feature point in the heat map; if the channel output confidence of the target point is not equal to the preset value, it is determined to exist in other channels or the background;

[0018] The classification layer is used to classify feature points according to the possibility of feature points when the heat maps overlap, and to perform pixel segmentation, and to calculate the corresponding focal loss function according to the channel output confidence;

[0019] The offset layer is used to add an offset to the feature point and calculate a corresponding offset loss function according to the offset and the classification probability;

[0020] The regression layer is used to regress and calculate the target size, estimate all the center points according to the feature points, and calculate the size loss function;

[0021] Among them, the total loss function of the fully trained facial recognition model is the sum of the focus loss function, the offset loss function and the size loss function.

[0022] Furthermore, constructing a three-dimensional face model according to the center point includes:

[0023] Adding corresponding depth information, three-dimensional scale information, and direction information to the center point;

[0024] According to the depth information of the center point, the three-dimensional scale information, and the direction information, width estimation, height estimation, and depth estimation are performed on each target to construct the three-dimensional face model.

[0025] Furthermore, the symmetry detection of the three-dimensional face model is performed to determine the corrected facial vector information and upload it to the cloud, including:

[0026] Performing symmetry detection on the three-dimensional face model to determine the rotation angle of the face;

[0027] According to the rotation angle, the corrected facial vector information is determined, and the facial vector information is uploaded to the cloud.

[0028] The present invention also provides a bus epidemic monitoring method, which is applied to the vehicle-mounted terminal in the bus epidemic monitoring system as described above, and the method comprises:

[0029] Get surveillance video stream;

[0030] The face image in the monitoring video stream is uploaded to the roadbed end, and the face image is input into the mask detection network to realize the mask recognition function, and the face image is uploaded to the roadbed end at the same time.

[0031] The present invention also provides a bus epidemic monitoring method, which is applied to the cloud in the bus epidemic monitoring system as described above, and the method comprises:

[0032] Get facial vector information;

[0033] Through the stored dual database model, a corresponding index is established, the facial vector information is screened, the pathogen carrier is determined, an alarm message is generated, and the alarm message is sent to the vehicle terminal and / or the command center.

[0034] Furthermore, the dual database model includes a record database and a query database, and the index establishment process includes:

[0035] Upload the facial vector information uploaded by the roadbed end to the record database, write it into the record file, and execute the update instruction;

[0036] The data in the record database is uploaded to the query database, the record file is executed, and the update instruction is executed again to establish a corresponding index.

[0037] Furthermore, the screening of the facial vector information, determining the pathogen carrier, and generating alarm information includes:

[0038] All feature point information in the pre-collected facial vector information is saved as a feature vector;

[0039] Performing a rotation operation on the eigenvector until maximum symmetry is obtained;

[0040] For the input facial vector information, calculate the distances of the corresponding feature points and determine the sum of the Euclidean distances;

[0041] When the sum of the Euclidean distances is less than a set threshold, they are determined to be the same person. If they are pathogen carriers, an alarm message is issued.

[0042] Compared with the prior art, the beneficial effects of the present invention include: by setting up the vehicle-mounted terminal, the detection of the mask function is realized, and the facial information is uploaded to the roadbed terminal at the same time; by setting up the roadbed terminal, the facial information of the vehicle-mounted terminal is received, and the facial information is processed, the digital facial feature point information is extracted, and data is interacted with the cloud; by setting up the cloud, the pre-stored dual database model is used to perform efficient indexing, and the facial vector information uploaded by the roadbed terminal is retrieved and screened to achieve efficient and rapid identification. In summary, the present invention realizes functions such as epidemic monitoring and constant risk prevention and control, ensures the safety of people on the bus under the epidemic environment, has low cost, uniformly stores facial feature point information, and provides efficient deep search, which can quickly complete the screening of people who have contact risks with designated personnel and return to the edge terminal. In addition, the cloud platform is linked with the command center. When a pathogen carrier is found, while realizing that the vehicle-mounted terminal is alarmed when the risk person gets on board, the cloud will also send detailed alarm information to the command center. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of the structure of an embodiment of a bus epidemic monitoring system provided by the present invention;

[0044] Figure 2 A schematic diagram of a process flow of an embodiment of a bus epidemic monitoring method provided by the present invention;

[0045] Figure 3 A schematic diagram of a flow chart of another embodiment of the bus epidemic monitoring method provided by the present invention;

[0046] Figure 4 A schematic diagram of a flow chart of another embodiment of the bus epidemic monitoring method provided by the present invention;

[0047] Figure 5 The present invention provides Figure 4 A flowchart of an embodiment of the index establishment process in step S402;

[0048] Figure 6 The present invention provides Figure 4A schematic diagram of a flow chart of an embodiment of generating alarm information in step S402;

[0049] Figure 7 A schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0051] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0052] In the description of the present invention, reference to an "embodiment" means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive of other embodiments. It is explicitly and implicitly understood by those skilled in the art that the described embodiments may be combined with other embodiments.

[0053] The present invention provides a bus epidemic monitoring system and method, which utilizes the mutually communicating vehicle-mounted terminal, roadbed terminal, and cloud terminal to realize the effective identification of pathogen carriers, and provides a new idea for further improving the safety of public transportation.

[0054] Before describing the embodiments, the following definitions are given for the relevant terms:

[0055] In-vehicle terminal: The full name is in-vehicle terminal equipment. In the IT field, the terminal generally refers to the processing equipment outside the network. It is the carrier of user input data and output of calculation results. In a broad sense, it can be understood as common smart devices such as PCs, smart phones, and platform computers.

[0056] Roadbed end: data processing equipment installed on the track or road surface;

[0057] Cloud: It is a software platform that uses application virtualization technology "Application Virtualization" and integrates multiple functions such as software search, download, use, management, and backup;

[0058] Based on the description of the above technical terms, in the prior art, when pathogen carriers take public transportation, it will cause huge hidden dangers to public safety and cause the spread of the epidemic. Therefore, the present invention aims to propose a bus epidemic monitoring system and method.

[0059] The specific embodiments are described in detail below:

[0060] The embodiment of the present invention provides a bus epidemic monitoring system, combined with Figure 1 Come and see, Figure 1 The schematic diagram of the structure of an embodiment of the bus epidemic monitoring system provided by the present invention includes a roadbed end 102, a vehicle-mounted end 101 and a cloud end 103, wherein:

[0061] The vehicle-mounted terminal 101 is used to input the acquired face image into the mask detection network to realize the mask recognition function, and upload the face image to the roadbed terminal at the same time;

[0062] The roadbed end 102 is used to extract facial vector information of the facial image and upload it to the cloud;

[0063] The cloud 103 is used to screen the facial vector information through the stored graph database model, generate alarm information, and send it to the vehicle-mounted terminal and / or the command center.

[0064] In an embodiment of the present invention, by setting up a vehicle-mounted terminal, the function of the mask is detected, and the facial information is uploaded to the roadbed terminal at the same time; by setting up the roadbed terminal, the facial information of the vehicle-mounted terminal is received, and the facial information is processed, the digital facial feature point information is extracted, and data is interacted with the cloud; by setting up the cloud, the pre-stored dual database model is used to perform efficient indexing, and the facial vector information uploaded by the roadbed terminal is retrieved and screened to achieve efficient and rapid identification.

[0065] In a specific embodiment of the present invention, the vehicle-mounted terminal is based on the K210 platform, runs the YOLOv2 network to realize face recognition and mask detection, and has a hardware onboard camera, infrared fill light (to assist camera recognition in dim environments), buttons, display screens, etc. Based on stm32, a temperature sensor is used to realize temperature measurement, and the face information is uploaded to the roadbed terminal. Among them, stm32 is also connected to the CAN network to control the vehicle to perform disinfection, ventilation and other operations;

[0066] The roadbed uses an edge computing board, which uses a lightweight neural network model based on the CenterNet method and converts the color space for the special lighting environment of the bus, thereby further improving the recognition accuracy of the neural network, extracting digital facial feature point information, and interacting with the cloud for data.

[0067] The cloud uses a dual database model to index the graph database, improve the efficiency of data retrieval, and quickly complete the screening of personnel who are at risk of contact with designated personnel and return them to the edge. The cloud platform is linked with the command center. When a pathogen carrier is found, the vehicle-mounted terminal will alarm when the risk person gets on board, and the cloud will also send detailed alarm information to the command center.

[0068] The embodiment of the present invention also provides a bus epidemic monitoring method, which is applied to the roadbed end of the above-mentioned bus epidemic monitoring system, combined with Figure 2 Come and see, Figure 2 The flowchart of an embodiment of the bus epidemic monitoring method provided by the present invention includes steps S201 to S203, wherein:

[0069] In step S201, the face image uploaded by the vehicle-mounted terminal is input into a well-trained facial recognition model, and the center point is output;

[0070] In step S202, a three-dimensional face model is constructed according to the center point;

[0071] In step S203, a symmetry check is performed on the three-dimensional face model to determine the corrected facial vector information and upload it to the cloud.

[0072] In an embodiment of the present invention, the roadbed end can receive facial information from the vehicle-mounted end, process the facial information, extract digital facial feature point information, form facial vector information, facilitate subsequent recognition, and interact with the cloud for data.

[0073] As a preferred embodiment, the above step S201 specifically includes:

[0074] The face image is converted into the HSV color space, and the converted face color image is input into a well-trained facial recognition model.

[0075] In the embodiment of the present invention, the picture sent by the vehicle-mounted terminal is effectively converted into the HSV color space to form rich image information.

[0076] As a preferred embodiment, the fully trained facial recognition model includes a normalization layer, a downsampling layer, a mapping layer, a classification layer, a shift layer, and a regression layer, wherein:

[0077] The normalization layer is used to calculate the normalized confidence of the face color image and determine a normalized image;

[0078] The downsampling layer is used to downsample the normalized image to determine a downsampled image;

[0079] The mapping layer is used to map the similarity of the target point in the downsampled image to the heat map through a Gaussian kernel function, wherein if the channel output confidence of the target point is equal to a preset value, it is determined to be consistent with the feature point in the heat map; if the channel output confidence of the target point is not equal to the preset value, it is determined to exist in other channels or the background;

[0080] The classification layer is used to classify feature points according to the possibility of feature points when the heat maps overlap, and to perform pixel segmentation, and to calculate the corresponding focal loss function according to the channel output confidence;

[0081] The offset layer is used to add an offset to the feature point and calculate a corresponding offset loss function according to the offset and the classification probability;

[0082] The regression layer is used to regress and calculate the target size, estimate all the center points according to the feature points, and calculate the size loss function;

[0083] Among them, the total loss function of the fully trained facial recognition model is the sum of the focus loss function, the offset loss function and the size loss function.

[0084] In an embodiment of the present invention, the CenterNet method is used and the MobileNetV3 network model is adopted to effectively extract facial feature points from images in the HSV color space.

[0085] As a preferred embodiment, the above step S202 specifically includes:

[0086] Adding corresponding depth information, three-dimensional scale information, and direction information to the center point;

[0087] According to the depth information of the center point, the three-dimensional scale information, and the direction information, width estimation, height estimation, and depth estimation are performed on each target to construct the three-dimensional face model.

[0088] In the embodiment of the present invention, efficient modeling is performed using the center point extracted by the network and its corresponding depth information, three-dimensional scale information, and direction information.

[0089] As a preferred embodiment, the above step S203 specifically includes:

[0090] Performing symmetry detection on the three-dimensional face model to determine the rotation angle of the face;

[0091] According to the rotation angle, the corrected facial vector information is determined, and the facial vector information is uploaded to the cloud.

[0092] In the embodiment of the present invention, effective correction is performed through symmetry detection to form accurate facial vector information.

[0093] In a specific embodiment of the present invention, the roadbed uses the CenterNet method and the MobileNet V3 network model to extract facial feature points from an image in the HSV color space as follows:

[0094] The first step is to calculate the normalized confidence of the pixel. The calculation formula is: Where R is the desired scaling ratio of the output image, C is the number of output channels, that is, the number of feature points, and the image width and height are W and H respectively;

[0095] In the second step, the target is downsampled, and then the similarity of the target point can be mapped to the heat map through the Gaussian kernel function. When it is found that for the image at the point (x, y), the confidence is close to 1 when the c channel is output, that is, Y xyc =1, the point is considered to be consistent with the expected feature point. Otherwise, it is considered that the point is more likely to exist in other channels or background. The formula is as follows:

[0096]

[0097] The third step is to select the most likely point for classification when the feature point heat map overlaps, and also perform pixel segmentation on the image. The focal loss is as follows:

[0098]

[0099] Among them, α and β are the hyperparameters in the formula, and N is the number of feature points;

[0100] The fourth step is to add an offset to the feature point to ensure the accuracy of positioning and calculate the average loss. The calculation formula is as follows:

[0101]

[0102]

[0103] The fifth step is to regress the size of the target and use feature point estimation to get all the center points. The loss function is as follows:

[0104]

[0105] The sixth step is to calculate the total loss. The total loss function is as follows:

[0106] L det =L k +λ size L size +λoffset L offset

[0107] The embodiment of the present invention also provides a bus epidemic monitoring method, which is applied to the vehicle-mounted terminal in the above-mentioned bus epidemic monitoring system, combined with Figure 3 Come and see, Figure 3 A flow chart of another embodiment of the bus epidemic monitoring method provided by the present invention includes steps S301 to S302, wherein:

[0108] In step S301, a surveillance video stream is obtained;

[0109] In step S302, the facial image in the monitoring video stream is uploaded to the roadbed end, and the facial image is input into the mask detection network to realize the mask recognition function, and the facial image is uploaded to the roadbed end at the same time.

[0110] In an embodiment of the present invention, a mask recognition function is added to the vehicle-mounted end, which can perform mask detection while performing facial contour recognition and upload the facial image to the roadbed end.

[0111] As a more specific embodiment, a YOLOv2 network is used for mask recognition. In the embodiment of the present invention, an efficient mask recognition function is achieved by introducing the YOLO-MASK algorithm.

[0112] The embodiment of the present invention also provides a bus epidemic monitoring method, which is applied to the cloud in the above-mentioned bus epidemic monitoring system, combined with Figure 4 Come and see, Figure 4 A flow chart of another embodiment of the bus epidemic monitoring method provided by the present invention includes steps S401 to S402, wherein:

[0113] In step S401, facial vector information is obtained;

[0114] In step S402, a corresponding index is established through the stored dual database model, the facial vector information is screened, the pathogen carrier is determined, an alarm message is generated, and the alarm message is sent to the vehicle terminal and / or the command center.

[0115] In the embodiment of the present invention, the cloud uses a dual database model to index the graph database, improve the efficiency of data retrieval, and quickly complete the screening of personnel who are at risk of contact with designated personnel and return them to the edge. The cloud platform is linked with the command center. When a pathogen carrier is found, the vehicle-mounted terminal will send an alarm when the risk person gets on board, and the cloud will also send detailed alarm information to the command center.

[0116] As a preferred embodiment, Figure 5 Come and see, Figure 5 The present invention provides Figure 4 The flowchart of an embodiment of the index establishment process in step S402 includes steps S501 to S502, wherein:

[0117] In step S501, the facial vector information uploaded by the roadbed end is uploaded to the record database, written into the record file, and the update instruction is executed;

[0118] In step S502, the data in the record database is uploaded to the query database, the record file is executed, and the update instruction is executed again to establish a corresponding index.

[0119] In the embodiment of the present invention, a corresponding index is effectively established to facilitate retrieval of facial vector information.

[0120] In a specific embodiment of the present invention, the dual database model includes a record database and a query database, and its operating principle is as follows:

[0121] Step 1: The edge computing network obtains the data upload record database;

[0122] Step 2, write the data into the record file and execute the update command;

[0123] Step 3, the record database uploads the data to the query database;

[0124] Step 4: Execute the record file, then execute the update command, and then create the index.

[0125] As a preferred embodiment, Figure 6 Come and see, Figure 6 The present invention provides Figure 4 The flowchart of an embodiment of generating alarm information in step S402 includes steps S601 to S604, wherein:

[0126] In step S601, all feature point information in the pre-collected facial vector information is saved as a feature vector;

[0127] In step S602, the eigenvector is rotated until maximum symmetry is obtained;

[0128] In step S603, for the input facial vector information, the distances of the corresponding feature points are calculated to determine the sum of the Euclidean distances;

[0129] In step S604, when the sum of the Euclidean distances is less than a set threshold, they are determined to be the same person, and if they are pathogen carriers, an alarm message is issued.

[0130] In an embodiment of the present invention, K-degree search is adopted, and whether it is the same person is determined by comparing feature distances, so as to achieve the purpose of efficient recognition.

[0131] In a specific embodiment of the present invention, the data retrieval adopts K-degree search, and whether it is the same person is determined by comparing feature distances. The specific method is as follows:

[0132] Step 01: Save all the feature point information in the collected facial information as a vector.

[0133] Step 02: Perform a rotation operation on the feature vector of the face to make the face as symmetric as possible for subsequent comparison.

[0134] Step 03: For the re-input feature vector, calculate the distances of its feature points and take the sum of its Euclidean distances. The calculation formula is:

[0135] Step 04: When the distance is less than the set threshold, that is, d sum <threshold, it is considered to be the same person.

[0136] The embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the above-mentioned bus epidemic monitoring method is implemented.

[0137] Generally speaking, the computer instructions for implementing the method of the present invention can be carried by any combination of one or more computer-readable storage media. A non-transitory computer-readable storage medium can include any computer-readable medium except the signal propagating temporarily itself.

[0138] The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the computer-readable storage medium can be any tangible medium that contains or stores a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0139] Computer program code for performing the operation of the present invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" language or similar programming languages, in particular, Python language suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. can be used. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0140] The embodiment of the present invention further provides an electronic device, Figure 7 Come and see, Figure 7 This is a structural diagram of an embodiment of an electronic device provided by the present invention. The electronic device 700 includes a processor 701, a memory 702, and a computer program stored in the memory 702 and executable on the processor 701. When the processor 701 executes the program, the bus epidemic monitoring method as described above is implemented.

[0141] As a preferred embodiment, the electronic device 700 further includes a display 703 for displaying that the processor 701 executes the bus epidemic monitoring method as described above.

[0142] Exemplarily, the computer program may be divided into one or more modules / units, one or more modules / units are stored in the memory 702, and executed by the processor 701 to complete the present invention. One or more modules / units may be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device 700. For example, the computer program may be divided into multiple units, and the specific functions of each unit are as described above, which will not be repeated here.

[0143] The electronic device 700 may be a desktop computer, a notebook, a PDA or a smart phone with an adjustable camera module.

[0144] Among them, the processor 701 may be an integrated circuit chip with signal processing capabilities. The above-mentioned processor 701 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The disclosed methods, steps and logic block diagrams in the embodiments of the present invention can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0145] The memory 702 may be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), etc. The memory 702 is used to store programs, and the processor 701 executes the program after receiving an execution instruction. The process definition method disclosed in any of the embodiments of the present invention may be applied to the processor 701 or implemented by the processor 701.

[0146] The display 703 may be an LCD display screen or an LED display screen, for example, a display screen on a mobile phone.

[0147] Understandably, Figure 7 The structure shown is only a schematic diagram of the structure of the electronic device 700. The electronic device 700 may also include Figure 7 More or fewer components as shown. Figure 7 Each component shown in the figure can be implemented by hardware, software or a combination thereof.

[0148] The computer-readable storage medium and electronic device provided according to the above embodiments of the present invention can be implemented with reference to the specific description of the bus epidemic monitoring method as described above according to the present invention, and have similar beneficial effects as the bus epidemic monitoring method, which will not be repeated here.

[0149] The present invention discloses a bus epidemic monitoring system and method. By setting a vehicle-mounted terminal, the function of a mask can be detected, and the facial information can be uploaded to the roadbed terminal at the same time. By setting the roadbed terminal, the facial information of the vehicle-mounted terminal is received, and the facial information is processed, the digital facial feature point information is extracted, and data is interacted with the cloud. By setting the cloud, a pre-stored dual database model is used to perform efficient indexing, and the facial vector information uploaded by the roadbed terminal is retrieved and screened to achieve efficient and rapid identification.

[0150] The technical solution of the present invention realizes functions such as epidemic monitoring and constant risk prevention and control, ensures the safety of people on the bus under the epidemic environment, has low cost, uniformly stores facial feature point information, and provides efficient deep search, which can quickly complete the screening of people who are at risk of contact with designated personnel and return to the edge end. And the cloud platform is linked with the command center. When a pathogen carrier is found, the vehicle-mounted end will send an alarm while realizing that when the risk person gets on board, the cloud will also send detailed alarm information to the command center. The present invention uses the YOLOV2 network based on the K210 platform to perform face contour recognition and mask monitoring, uses the algorithm in the intersection edge computing module to detect face feature points, establishes a dual database model, uses K-degree search, accurately identifies and records pathogen carriers, efficiently searches for potential contacts, and provides strong support for minimizing the risk of infection. At the same time, temperature measurement is achieved through the sensor base, and the vehicle is controlled through CAN communication for disinfection and ventilation. This system can effectively monitor and ensure the safety of people in the bus under the epidemic environment.

[0151] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A bus epidemic monitoring method, characterized in that: Applied to the roadbed end of a bus epidemic monitoring system, the bus epidemic monitoring system includes a roadbed end, a vehicle-mounted end and a cloud end, wherein: The vehicle-mounted end is used to input the acquired face image into the mask detection network to realize the mask recognition function, and upload the face image to the roadbed end at the same time; The roadbed end is used to extract facial vector information of the facial image and upload it to the cloud; The cloud is used to screen the facial vector information through a stored graph database model, generate alarm information, and send it to the vehicle terminal and / or the command center; The method comprises: The face image uploaded by the vehicle is converted into HSV color space, and the converted face color image is input into the trained facial recognition model to output the center point; Constructing a three-dimensional face model according to the center point; Performing symmetry detection on the three-dimensional face model, determining corrected facial vector information, and uploading it to the cloud; The fully trained facial recognition model includes a normalization layer, a downsampling layer, a mapping layer, a classification layer, a bias layer, and a regression layer, wherein: The normalization layer is used to calculate the normalized confidence of the face color image and determine a normalized image; The downsampling layer is used to downsample the normalized image to determine a downsampled image; The mapping layer is used to map the similarity of the target point in the downsampled image to the heat map through a Gaussian kernel function, wherein if the channel output confidence of the target point is equal to a preset value, it is determined to be consistent with the feature point in the heat map, and if the channel output confidence of the target point is not equal to the preset value, it is determined to exist in other channels or the background; The classification layer is used to classify feature points according to the possibility of feature points when the heat maps overlap, and to perform pixel segmentation, and to calculate the corresponding focal loss function according to the channel output confidence; The offset layer is used to add an offset to the feature point and calculate a corresponding offset loss function according to the offset and the classification probability; The regression layer is used to regress and calculate the target size, estimate all the center points according to the feature points, and calculate the size loss function; Among them, the total loss function of the fully trained facial recognition model is the sum of the focus loss function, the offset loss function and the size loss function.

2. The bus epidemic monitoring method according to claim 1, characterized in that: The step of constructing a three-dimensional face model according to the center point includes: Adding corresponding depth information, three-dimensional scale information, and direction information to the center point; According to the depth information of the center point, the three-dimensional scale information, and the direction information, width estimation, height estimation, and depth estimation are performed on each target to construct the three-dimensional face model.

3. The bus epidemic monitoring method according to claim 1, characterized in that: The symmetry detection of the three-dimensional face model, determining the corrected facial vector information, and uploading it to the cloud includes: Performing symmetry detection on the three-dimensional face model to determine the rotation angle of the face; According to the rotation angle, the corrected facial vector information is determined, and the facial vector information is uploaded to the cloud.

4. A bus epidemic monitoring method, characterized in that: Applied to the on-board terminal in the bus epidemic monitoring system according to claim 1, the method comprises: Get surveillance video stream; The face image in the monitoring video stream is uploaded to the roadbed end, and the face image is input into the mask detection network to realize the mask recognition function, and the face image is uploaded to the roadbed end at the same time.

5. A bus epidemic monitoring method, characterized in that: Applied to the cloud in the bus epidemic monitoring system of claim 1, the method comprises: Get facial vector information; Through the stored dual database model, a corresponding index is established, the facial vector information is screened, the pathogen carrier is determined, an alarm message is generated, and the alarm message is sent to the vehicle terminal and / or the command center.

6. The bus epidemic monitoring method according to claim 5, characterized in that: The dual database model includes a record database and a query database, and the index establishment process includes: Upload the facial vector information uploaded by the roadbed end to the record database, write it into the record file, and execute the update instruction; The data in the record database is uploaded to the query database, the record file is executed, and the update instruction is executed again to establish a corresponding index.

7. The bus epidemic monitoring method according to claim 6, characterized in that: The step of screening the facial vector information, determining the pathogen carrier, and generating alarm information includes: All feature point information in the pre-collected facial vector information is saved as a feature vector; Performing a rotation operation on the eigenvector until maximum symmetry is obtained; For the input facial vector information, calculate the distances of the corresponding feature points and determine the sum of the Euclidean distances; When the sum of the Euclidean distances is less than a set threshold, they are determined to be the same person. If they are pathogen carriers, an alarm message is issued.

Images