WO2020052436A1 - 车辆超载报警方法、装置、电子设备及存储介质 - Google Patents

车辆超载报警方法、装置、电子设备及存储介质 Download PDF

Info

Publication number
WO2020052436A1
WO2020052436A1 PCT/CN2019/103030 CN2019103030W WO2020052436A1 WO 2020052436 A1 WO2020052436 A1 WO 2020052436A1 CN 2019103030 W CN2019103030 W CN 2019103030W WO 2020052436 A1 WO2020052436 A1 WO 2020052436A1
Authority
WO
WIPO (PCT)
Prior art keywords
vehicle
boarding
detected
image data
alighting
Prior art date
Application number
PCT/CN2019/103030
Other languages
English (en)
French (fr)
Inventor
余声
Original Assignee
杭州海康威视数字技术股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 杭州海康威视数字技术股份有限公司 filed Critical 杭州海康威视数字技术股份有限公司
Publication of WO2020052436A1 publication Critical patent/WO2020052436A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
    • G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60Q—ARRANGEMENT OF SIGNALLING OR LIGHTING DEVICES, THE MOUNTING OR SUPPORTING THEREOF OR CIRCUITS THEREFOR, FOR VEHICLES IN GENERAL
    • B60Q9/00—Arrangement or adaptation of signal devices not provided for in one of main groups B60Q1/00 - B60Q7/00, e.g. haptic signalling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/08—Detecting or categorising vehicles

Definitions

  • the present application relates to the field of intelligent transportation technology, and in particular, to a method, a device, an electronic device, and a storage medium for a vehicle overload alarm.
  • Overloading the vehicle will cause the vehicle's weight to increase, which will lead to an increase in vehicle braking distance and increased danger. Overloading a vehicle can also increase the load on the tire, causing it to deform or cause a flat tire, which can be dangerous. In addition, overload can also affect the steering performance of the vehicle, which can easily lead to accidents due to out-of-control steering. Drivers driving overloaded and overloaded vehicles often increase psychological burden and mental pressure, and are prone to operational errors, which can cause traffic accidents. And if an overloaded vehicle has a traffic accident, it will bring additional personal injury.
  • the purpose of the embodiments of the present application is to provide a method, a device, an electronic device, and a storage medium for a vehicle overload alarm, so as to implement a vehicle overload alarm.
  • Specific technical solutions are as follows:
  • an embodiment of the present application provides a vehicle overload alarm method, where the method includes:
  • Acquiring target image data including a vehicle to be detected; in the target image data, acquiring identification information of the vehicle to be detected; and determining a passenger carrying threshold and the current number of passengers of the vehicle to be detected according to the identification information; Determine whether there is a boarding or alighting behavior according to the target image data; if there is a boarding or alighting behavior, determine the number of people who board or alight; update the current number of passengers based on the number of people; The magnitude of the passenger carrying threshold is described; if the updated current passenger carrying capacity is greater than the passenger carrying threshold, an overload alarm for the vehicle to be detected is triggered.
  • the identification information includes license plate information and model information; and determining the passenger carrying threshold of the vehicle to be detected according to the identification information.
  • the current number of passengers including: acquiring the current number of passengers of the vehicle to be detected according to the license plate information; and acquiring the passenger number of the vehicle to be detected according to the model information.
  • the identification information includes model information and license plate information
  • the method further includes:
  • the determining whether there is a boarding or getting off behavior based on the image data includes: detecting and analyzing the image data by using a pre-trained deep learning model to determine whether there is a boarding or getting off behavior; in advance;
  • the process of training a deep learning model includes: obtaining multiple image data that includes boarding and alighting behaviors and multiple image data that does not include alighting and alighting behaviors; calibrating the image data including alighting and alighting behaviors as positive samples, and Image data is calibrated as negative samples to train the deep learning model to obtain a pre-trained deep learning model.
  • determining the number of people who get on and off if there is a behavior of getting on and off includes: if there is a behavior of getting on and off, obtaining a video frame set including the behavior of getting on and off; identifying the waiting in the video frame set.
  • Detect the vehicle position of the vehicle and the personnel position of each boarding and alighting person analyze the change of the position of each boarding and alighting person in accordance with the sequence order, and obtain the moving direction of each boarding and alighting person; Personnel, if the moving direction of the boarding person deviates from the vehicle position, it is determined that the boarding person gets off the vehicle; for each of the boarding persons, if the moving direction of the boarding person points to the vehicle position, the boarding person is determined The crew members get in the car.
  • an embodiment of the present application provides a vehicle overload alarm device, where the device includes:
  • An image data acquisition module for acquiring target image data including a vehicle to be detected
  • An identification information acquisition module configured to acquire identification information of the vehicle to be detected in the target image data
  • a first calculation module configured to determine a passenger carrying threshold of the vehicle to be detected and the current number of passengers according to the identification information
  • a first judging module used for an image data acquisition module, for judging whether there is a boarding or getting off behavior according to the target image data
  • a second calculation module configured to determine the number of people who get on and off if there is a boarding behavior
  • a number of people updating module configured to update the current number of passengers according to the number of people getting on and off;
  • a second judgment module configured to judge the updated current number of passengers and the magnitude of the passenger threshold
  • the overload alarm module is configured to trigger an overload alarm for the vehicle to be detected if the updated current number of passengers is greater than the passenger threshold.
  • the identification information includes license plate information and vehicle type information
  • the first calculation module includes:
  • a number of people acquisition submodule configured to obtain the current number of passengers of the vehicle to be detected according to the license plate information
  • a threshold acquisition submodule is configured to acquire a passenger carrying threshold of the vehicle to be detected according to the vehicle model information.
  • the identification information includes model information and license plate information
  • the device further includes:
  • a third determination module configured to determine whether the type of the vehicle to be detected is a decidable type according to the type information
  • the first calculation module is specifically configured to:
  • the vehicle type of the vehicle to be detected is a determinable vehicle type, determining a passenger carrying threshold corresponding to the vehicle type information according to the vehicle type information, and obtaining the passenger carrying threshold of the vehicle to be detected; and acquiring the vehicle based on the license plate information The current number of passengers on the vehicle to be tested.
  • the first judgment module is specifically configured to:
  • the process of pre-training a deep learning model includes:
  • the image data including the boarding and alighting behaviors are labeled as positive samples, and the image data not including the boarding and alighting behaviors are labeled as negative samples.
  • the deep learning model is trained to obtain a pre-trained deep learning model.
  • the second calculation module includes:
  • the video frame acquisition submodule is used to obtain a set of video frames including the boarding and alighting behavior if the boarding and alighting behavior exists;
  • a position determining submodule configured to identify a vehicle position of the vehicle to be detected in the video frame set and a person position of each boarding and alighting person;
  • a direction determining sub-module is configured to analyze a change in the position of each of the boarding and disembarking personnel in accordance with the time sequence order to obtain a moving direction of each of the boarding and disembarking personnel;
  • a get-off determination sub-module is configured to determine, for each of the boarding and alighting persons, that the boarding and alighting person gets off the vehicle if the moving direction of the boarding and alighting person deviates from the vehicle position;
  • a boarding determination sub-module is configured to determine, for each of the boarding and disembarking persons, that the boarding and disembarking person gets on the vehicle if the moving direction of the boarding and disembarking person points to the vehicle position.
  • an embodiment of the present application provides an electronic device including a processor and a memory
  • the memory is used to store a computer program; the processor is used to execute the program stored on the memory to implement the following method steps: obtaining target image data including a vehicle to be detected; and in the target image data Obtaining identification information of the vehicle to be detected; determining the passenger carrying threshold and the current number of passengers of the vehicle to be detected according to the identification information; judging whether there is a boarding behavior based on the target image data; if there is a boarding behavior Vehicle behavior, determine the number of passengers; update the current number of passengers according to the number of passengers; determine the updated number of passengers and the threshold of the number of passengers; if the current number of passengers is greater than the number of passengers Said passenger carrying threshold triggers an overload alarm for the vehicle to be detected.
  • the identification information includes license plate information and vehicle type information; and determining the passenger carrying threshold and the current number of passengers of the vehicle to be detected according to the identifying information includes: according to the license plate Information to obtain the current number of passengers on the vehicle to be detected; and to obtain a passenger threshold on the vehicle to be detected according to the model information.
  • the identification information includes model information and license plate information
  • the processor is further configured to perform the following steps: according to the model information, determine whether the model of the vehicle to be detected is a determineable model ; Determining the passenger carrying threshold and the current number of passengers of the vehicle to be detected according to the identification information includes: if the vehicle type of the vehicle to be detected is a determinable vehicle type, determining the vehicle type according to the vehicle type information The passenger carrying threshold corresponding to the information is used to obtain the passenger carrying threshold of the vehicle to be detected; according to the license plate information, the current number of passengers carrying the vehicle to be detected is obtained.
  • the determining whether there is a boarding or getting off behavior based on the image data includes: detecting and analyzing the image data by using a pre-trained deep learning model to determine whether there is a boarding or getting off behavior; in advance;
  • the process of training a deep learning model includes: obtaining multiple image data that includes boarding and alighting behaviors and multiple image data that does not include alighting and alighting behaviors; calibrating the image data including alighting and alighting behaviors as positive samples, and Image data is calibrated as negative samples to train the deep learning model to obtain a pre-trained deep learning model.
  • determining the number of people who get on and off if there is a behavior of getting on and off includes: if there is a behavior of getting on and off, obtaining a video frame set including the behavior of getting on and off; identifying the waiting in the video frame set.
  • Detect the vehicle position of the vehicle and the personnel position of each boarding and alighting person analyze the change of the position of each boarding and alighting person in accordance with the sequence order, and obtain the moving direction of each boarding and alighting person; for each of the boarding and alighting Personnel, if the moving direction of the boarding person deviates from the vehicle position, it is determined that the boarding person gets off the vehicle; for each of the boarding persons, if the moving direction of the boarding person points to the vehicle position, the boarding person is determined The crew members get in the car.
  • an embodiment of the present application provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, implements any one of the foregoing first aspects.
  • Vehicle overload alarm method
  • the vehicle overload alarm method, device, electronic device and storage medium provided in the embodiments of the present application obtain target image data including the vehicle to be detected; in the target image data, obtain identification information of the vehicle to be detected; and determine the to-be-detected according to the identification information.
  • the passenger carrying threshold of the vehicle and the current number of passengers determine whether there are boarding and alighting behaviors based on the image data; if there are boarding and alighting behaviors, determine the number of passengers; The number of passengers and the passenger carrying threshold; if the updated current passenger carrying is greater than the passenger carrying threshold, an overload alarm for the vehicle to be detected is triggered.
  • the target image data is used to determine the passenger carrying threshold of the vehicle to be detected and the current number of passengers.
  • FIG. 1 is a schematic diagram of a vehicle overload alarm device according to an embodiment of the present application.
  • FIG. 2a is a first schematic diagram of a method for judging the exercise direction of a person according to an embodiment of the present application
  • FIG. 2b is a second schematic diagram of a method for judging an exercise direction of a person according to an embodiment of the present application
  • 2c is a third schematic diagram of a method for judging the exercise direction of a person according to an embodiment of the present application.
  • FIG. 3 is a schematic flowchart of a vehicle overload alarm method according to an embodiment of the present application.
  • FIG. 4 is another schematic flowchart of a vehicle overload alarm method according to an embodiment of the present application.
  • FIG. 5 is another schematic diagram of a vehicle overload alarm device according to an embodiment of the present application.
  • FIG. 6 is a schematic diagram of an electronic device according to an embodiment of the present application.
  • Intelligent transportation It is established by the effective integration and application of information technology, data communication transmission technology, electronic sensing technology, control technology and computer technology in the entire ground traffic management system. Comprehensive transportation management system with real-time, accurate and efficient features. Its prominent feature is the main line of information collection, processing, release, exchange, analysis, and utilization, providing diversified services for traffic participants.
  • Object detection Also called object extraction, is to extract objects of interest from complex background images.
  • Image classification An image processing method that distinguishes different types of objects according to the different characteristics of the objects reflected in the image information. It uses a computer to perform quantitative analysis on the image, and classifies each pixel or area in the image or image into one of several categories to replace human visual interpretation.
  • Target recognition Used to separate the specified target from other targets.
  • CNN Convolutional Neural Networks
  • It is the first learning algorithm that can successfully train multi-layer network structures. It uses spatial relationships to reduce the number of parameters that need to be learned to improve the training performance of the general forward BP (Error Back Propagation) algorithm.
  • Faster RCNN (Faster, Regions, Convolutional, Neural, Network): It is based on deep learning RCNN (Regions With Convolutional Neural Network Features) series of best method for target detection. It is a framework for CNN network to achieve end-to-end target detection. Where R corresponds to Region.
  • YOLO (You Only Look Once): YOLO is a very successful real-time target detection algorithm. It has become an important part of many classifiers, segmentation, human pose and behavior classification, and it is a CNN network's end-to-end target detection framework.
  • the vehicle overload alarm method, device, electronic equipment and storage medium in the embodiments of the present application can be applied to urban road monitoring video analysis. By judging the vehicle model and counting the number of passengers who visit the vehicle within a short period of time, it can be determined whether the vehicle has an overcrowding behavior. And by recording the license plate number, this behavior is used to regulate the behavior of vehicles by the transportation department.
  • a vehicle overload alarm device includes: a license plate recognition module 101, a vehicle model recognition module 102, a boarding and statistics counting module 103, and a discrimination alarm module 104.
  • the license plate recognition module 101 is used to identify the license plate of the vehicle
  • the vehicle model recognition module 102 is used to identify the vehicle model. It is relatively mature in the related technology and can directly reuse the related technology.
  • the judging and alarming module 104 is configured to perform an alarm when the number of persons carried by the vehicle exceeds the threshold of the number of persons carried by the vehicle.
  • the boarding and embarkation statistics module 103 is used to count the number of people boarding and disembarking a vehicle. Statistics of pedestrians on both sides of the vehicle can be used to achieve statistics of boarding and disembarking behavior. Among them, for each pedestrian's boarding and alighting behavior, a pedestrian is detected by a target detection algorithm; a pedestrian trajectory is generated; and a boarding and alighting behavior is discriminated according to the trajectory direction. Calculate the total number of people getting on and off by integrating the behavior of each pedestrian, and then get the number of people carrying the vehicle.
  • the video data is detected by an object detection algorithm, and the position of the pedestrian is detected as shown by a rectangular box in FIG. 2a.
  • the overlap is greater than 0.5, and the two frames are associated Target (pedestrian) in the video frame.
  • the pedestrian detection results for multiple frames of associated video frames are shown in Figure 2c, where # 1, # 2, # 3, and # 4 are the positions of the targets in the four-frame video frames, respectively.
  • # 1, # 2, # 3, and # 4 are the positions of the targets in the four-frame video frames, respectively.
  • the actual number of associated frames can be set during the actual detection process.
  • the pedestrian trajectory direction is calculated as leaving the vehicle, and it is determined as alighting behavior.
  • the process of the vehicle overload alarm method according to the embodiment of the present application can be shown in FIG. 3.
  • the overload behavior can be automatically identified.
  • the overtaking vehicle is located and recorded. Violations will help strengthen traffic management's monitoring of overloading.
  • the monitoring equipment used to collect images in the embodiments of the present application can be off-board, and can directly use public monitoring equipment for urban transportation, which brings the advantage that the regulatory department does not need to install equipment on each vehicle to be monitored, but only needs to In the existing urban intelligent traffic monitoring system, the above functions can be added to complete the capture of overload behavior, and the cost is greatly reduced. And it can reduce the situation of human interference by the supervised person, such as: damage or obstruction of the vehicle camera by smearing, blocking, adjusting the camera direction, etc. And can be combined with models to determine the overload of vehicles of different models. For example, the maximum number of passengers under a taxi is four, and the maximum number of passengers under a regular car is five.
  • an embodiment of the present application provides a vehicle overload alarm method. Referring to FIG. 4, the method includes:
  • the vehicle overload alarm method in the embodiment of the present application may be implemented by an alarm system, and the alarm system is any system capable of implementing the vehicle overload alarm method in the embodiment of the present application.
  • the alarm system is any system capable of implementing the vehicle overload alarm method in the embodiment of the present application.
  • the alarm system may be a kind of equipment, including: a processor, a memory, a communication interface, and a bus; the processor, the memory, and a communication interface are connected through the bus and complete communication with each other; the memory stores executable program code; and the processor reads the memory by The executable program code stored in the program is used to run a program corresponding to the executable program code, and is used to execute the vehicle overload alarm method in the embodiment of the present application.
  • the alarm system may also be an application program for executing the vehicle overload alarm method in the embodiment of the present application during operation.
  • the alarm system may also be a storage medium for storing executable code, and the executable code is used to execute the vehicle overload alarm method in the embodiment of the present application.
  • the alarm system acquires the image data containing the vehicle to be detected through the image acquisition device, that is, the target image data.
  • the image acquisition device may be a device such as a camera.
  • the image acquisition device is a camera in an urban traffic monitoring system.
  • the target image data is analyzed to obtain the identification information of the vehicle to be detected, and the identification information is used to uniquely identify the vehicle to be detected.
  • the target image data is analyzed by RCNN, Faster RCNN, or YOLO to obtain the license plate information of the vehicle to be detected.
  • S403. Determine the passenger carrying threshold of the vehicle to be detected and the current number of passengers according to the identification information.
  • the identification information is the unique identification information of the vehicle.
  • the alarm system will keep a record of the current number of passengers carried by each vehicle.
  • each vehicle will correspond to the corresponding passenger threshold.
  • the passenger threshold of the vehicle is set according to relevant traffic regulations, for example, type
  • the passenger-carrying threshold corresponding to a taxi-based vehicle is 4, and the passenger-carrying threshold corresponding to a type of private car is 5, and so on.
  • the alarm system will keep a record of the current number of passengers in each vehicle.
  • the vehicle at the parking point starts, it will obtain the video data before and after the vehicle starts as the target video data.
  • the initial current number of passengers in the vehicle is 0.
  • the vehicle overload alarm method of the embodiment updates the current number of passengers carried by the vehicle. Subsequently, the vehicle overload alarm method of the embodiment of the present application continues to keep a record of the current number of passengers carried by the vehicle.
  • the alarm system can obtain the current number of passengers of the vehicle indicated by the identification information from the records according to the identification information.
  • S404 Determine whether there is a boarding or getting off behavior based on the target image data.
  • the alarm system analyzes the target image data through a preset algorithm, such as a feature comparison algorithm or a neural network algorithm, to determine whether there is a boarding or getting off behavior.
  • a preset algorithm such as a feature comparison algorithm or a neural network algorithm
  • the above determining whether there is a boarding or getting off behavior based on the image data includes:
  • the process of pre-training a deep learning model includes:
  • Step 1 Obtain a plurality of image data including a boarding behavior and a plurality of image data not including a boarding behavior.
  • Step 2 The image data including the boarding and alighting behaviors are labeled as positive samples, and the image data not including the boarding and alighting behaviors are labeled as negative samples to train the deep learning model to obtain a pre-trained deep learning model.
  • the deep learning model here can be Faster RCNN, YOLO, RCNN, or DMP (Deformable Parts Model).
  • the alarm system determines the number of people boarding or alighting. For example, the alarm system can analyze pre-trained convolutional neural networks to analyze the video frames containing the boarding and getting-off behavior in the target video data to determine the number of people getting on and off.
  • the steps of a pre-trained convolutional neural network include:
  • Step 1 Obtain a plurality of first image data including a boarding behavior and a plurality of second image data not including a boarding behavior.
  • Step 2 Calibrate the boarding behavior and the number of people in the first image data of each first image data, and calibrate the getting off behavior and the number of people in the vehicle behavior in each image data to obtain the calibrated first image data.
  • Step three training the convolutional neural network by using the calibrated first image data as positive samples and using the second image data as negative samples to obtain a pre-trained convolutional neural network.
  • the alarm system Each time the alarm system detects a person getting off the vehicle, the current passenger number of the vehicle to be detected is reduced by 1, and each time a person is detected on the vehicle, the alarm system increases the current passenger number of the vehicle to be detected by 1 to update the current passenger number.
  • S407 Determine the updated current number of passengers and the magnitude of the above-mentioned passenger threshold.
  • the alarm system triggers an overload alarm for the vehicle to be detected.
  • the alarm system notifies the subscriber of the identification information of the vehicle to be detected and the number of passengers currently loaded, and can also alert the location of the vehicle to be detected.
  • the location of the vehicle to be detected can be Obtained through a camera that collects target video data.
  • an automatic alarm for vehicle overload can be realized, and the monitoring equipment for collecting target image data may be off-board, and the public monitoring equipment for urban traffic can be directly used to collect target image data.
  • the supervisory authority does not need to To install equipment on the vehicle to be monitored, it is only necessary to add the above functions to the existing urban intelligent traffic monitoring system to complete the capture of overload behavior, the cost is greatly reduced, and the situation of human interference by the supervised person can be reduced.
  • the identification information includes license plate information and vehicle type information.
  • determining the passenger carrying threshold of the vehicle to be detected and the current number of passengers according to the identification information includes:
  • Step 1 Obtain the current number of passengers of the vehicle to be detected according to the license plate information.
  • the alarm system stores the number of passengers carried by each vehicle in a designated memory. When needed, the alarm system reads the current number of passengers of the vehicle to be detected in the memory according to the license plate information.
  • Step 2 Obtain the passenger carrying threshold of the vehicle to be detected according to the vehicle model information.
  • the passenger carrying threshold corresponding to the vehicle type of the vehicle to be detected that is, the passenger carrying threshold of the vehicle to be detected.
  • the taxi passenger carrying threshold is 4 persons
  • the ordinary passenger car carrying passenger threshold is 5 persons. .
  • the vehicle to be detected can be uniquely determined through the license plate information, and the passenger carrying threshold of the vehicle to be detected can be determined through the vehicle type information, which is convenient and quick.
  • the identification information includes model information and license plate information
  • the method further includes:
  • the vehicle type information it is determined whether the vehicle type of the vehicle to be detected is a determinable vehicle type.
  • determining the passenger carrying threshold of the vehicle to be detected and the current number of passengers according to the identification information includes:
  • the vehicle type of the vehicle to be detected is a determinable vehicle type
  • the alarm system determines that the model of the vehicle to be detected is not a determinable model, and ends the overload determination.
  • the decidable model can be set according to the actual situation.
  • the decidable model can include taxis, private cars, and business cars.
  • the alarm system can set the decidable vehicle type, and only the designated vehicle type is used to get on and off behaviors and subsequent determinations, which can avoid overload determination of vehicles outside the decidable vehicle type and save computing resources.
  • determining the number of people boarding and alighting includes:
  • Step 1 If there is a boarding and alighting behavior, obtain a video frame set including the boarding and alighting behavior.
  • the alarm system determines that the target video data has a boarding and alighting behavior
  • the alarm system extracts a set of video frames containing the characteristics of the boarding and alighting behavior.
  • Step 2 Identify the vehicle position of the vehicle to be detected in the video frame set and the personnel positions of each boarding and alighting person.
  • the alarm system uses a target recognition algorithm to determine the vehicle position of the vehicle to be detected and the personnel position of each boarding and alighting person in each video frame of the video frame set.
  • Step 3 According to the time sequence, the change of the position of each of the boarding and alighting personnel is analyzed to obtain the moving direction of each of the boarding and alighting personnel.
  • the change of the position of the boarding and disembarking personnel is analyzed to obtain the moving direction of the boarding and disembarking personnel.
  • the target tracking algorithm is used to determine the change of the position of the boarding and alighting person to obtain the moving direction of the boarding and alighting person; or by calculating the change of the center position of the boarding and alighting person position, The direction of movement of the vehicle personnel.
  • the center point of the person position of the boarding and disembarking personnel is selected, where the coordinates of the center point are expressed as (X i , Y i ).
  • X i represents the X coordinate of the i-th video frame
  • Y i represents the Y coordinate of the i-th video frame
  • the value of i ranges from [1, N]
  • N is the number of video frames.
  • t is a smooth interval, which can be 1-5.
  • the judge moves in the positive direction of the X axis, otherwise the judge moves in the negative direction of the X axis.
  • Step 4 For each of the boarding and alighting persons, if the moving direction of the boarding and alighting person deviates from the vehicle position, it is determined that the boarding and alighting person gets out of the vehicle.
  • Step 5 For each of the boarding and disembarking persons, if the moving direction of the boarding and disembarking person points to the vehicle position, it is determined that the boarding and disembarking person gets on the vehicle.
  • An embodiment of the present application further provides a vehicle overload alarm device.
  • the device includes:
  • the image data acquisition module 501 is configured to acquire target image data including a vehicle to be detected.
  • the identification information obtaining module 502 is configured to obtain the identification information of the vehicle to be detected in the target image data.
  • the first calculation module 503 is configured to determine the passenger carrying threshold and the current number of passengers of the vehicle to be detected according to the identification information.
  • the first determination module 504 is used for an image data acquisition module, and is configured to determine whether there is a boarding or getting off behavior based on the target image data.
  • the second calculation module 505 is configured to determine the number of people who get on and off if there is a behavior of getting on or off the vehicle.
  • the number of people updating module 506 is configured to update the current number of passengers according to the number of people getting on and off the bus.
  • the second determining module 507 is configured to determine the updated current number of passengers and the magnitude of the above-mentioned passenger threshold.
  • the overload alarm module 508 is configured to trigger an overload alarm for the vehicle to be detected if the updated current number of passengers is greater than the aforementioned passenger threshold.
  • an automatic alarm for vehicle overload can be realized, and the monitoring equipment for collecting target image data may be off-board, and the public monitoring equipment for urban traffic can be directly used to collect target image data.
  • the supervisory authority does not need to To install equipment on the vehicle to be monitored, it is only necessary to add the above functions to the existing urban intelligent traffic monitoring system to complete the capture of overload behavior, the cost is greatly reduced, and the situation of human interference by the supervised person can be reduced.
  • the identification information includes license plate information and vehicle type information
  • the first calculation module 503 includes:
  • the number of people acquisition submodule is configured to obtain the current number of passengers of the vehicle to be detected according to the license plate information.
  • the threshold acquisition submodule is configured to acquire the passenger carrying threshold of the vehicle to be detected according to the vehicle model information.
  • the identification information includes model information and license plate information
  • the device further includes:
  • the third determining module is configured to determine whether the vehicle type of the vehicle to be detected is a determineable vehicle type according to the vehicle type information.
  • the first calculation module 503 is specifically configured to:
  • the vehicle type of the vehicle to be detected is a determinable vehicle type
  • the first determining module 504 is specifically configured to:
  • the process of pre-training a deep learning model includes:
  • the image data including the boarding and alighting behaviors are labeled as positive samples, and the image data not including the boarding and alighting behaviors are labeled as negative samples.
  • the deep learning model is trained to obtain a pre-trained deep learning model.
  • the foregoing second calculation module 505 includes:
  • the video frame acquisition submodule is used to obtain a set of video frames including the boarding and alighting behavior if the boarding and alighting behavior exists;
  • a position determining sub-module is configured to identify a vehicle position of the vehicle to be detected and a person position of each boarding and alighting person in the video frame set.
  • the direction determining sub-module is configured to analyze a change in the position of each of the persons boarding and disembarking in accordance with the time sequence order to obtain the moving direction of each of the persons boarding and disembarking.
  • a get-off determination sub-module is configured to determine, for each of the boarding and disembarking persons, that the boarding and disembarking person gets off the vehicle if the moving direction of the boarding and disembarking person deviates from the vehicle position.
  • the boarding determination sub-module is configured to determine, for each of the boarding and alighting persons, that the boarding and alighting person is on board if the moving direction of the boarding and alighting person points to the vehicle position.
  • An embodiment of the present application provides an electronic device, as shown in FIG. 6, including a processor 601 and a memory 602.
  • the memory 602 is configured to store a computer program, and the processor 601 is configured to execute the following steps when the program stored in the memory 602 is executed:
  • target image data including the vehicle to be detected; in the target image data, obtain the identification information of the vehicle to be detected; determine the passenger carrying threshold and the current number of passengers of the vehicle to be detected according to the identification information; and according to the target image Data to determine whether there is a boarding and alighting behavior; if there is a boarding and alighting behavior, determine the number of people boarding and alighting; update the current number of passengers based on the number of people aboard; The updated current number of passengers is greater than the above-mentioned passenger threshold, which triggers an overload alarm for the vehicle to be detected.
  • an automatic alarm for vehicle overload can be realized, and the monitoring equipment for collecting target image data may be off-board, and the public monitoring equipment for urban traffic can be directly used to collect target image data.
  • the supervisory authority does not need to To install equipment on the vehicle to be monitored, it is only necessary to add the above functions to the existing urban intelligent traffic monitoring system to complete the capture of overload behavior, the cost is greatly reduced, and the situation of human interference by the supervised person can be reduced.
  • the identification information includes license plate information and vehicle type information; and determining the passenger carrying threshold and the current number of passengers of the vehicle to be detected according to the identification information includes: obtaining the waiting information according to the license plate information. Detect the current number of passengers on the vehicle; obtain the passenger threshold for the vehicle to be detected according to the model information.
  • the identification information includes model information and license plate information
  • the processor is further configured to perform the following steps: determine whether the model of the vehicle to be detected is a determinable model according to the model information; Identification information to determine the passenger carrying threshold of the vehicle to be detected and the current number of passengers, including: if the vehicle type of the vehicle to be detected is a determinable vehicle type, determining the passenger carrying threshold corresponding to the vehicle type information according to the vehicle type information, and obtaining the passenger waiting value Detect the passenger carrying threshold of the vehicle; obtain the current number of passengers of the vehicle to be detected according to the license plate information.
  • the above-mentioned determining whether to get on or off based on the image data includes: detecting and analyzing the above-mentioned image data by using a pre-trained deep learning model to determine whether there is a getting-on or off behavior; pre-training deep learning
  • the process of the model includes: obtaining multiple image data including the behavior of getting on and off the vehicle and multiple image data not including the behavior of getting on and off the vehicle; calibrating the image data including the behavior of getting on and off the vehicle as a positive sample, and calibrating the image data not including the behavior of getting on and off the vehicle
  • the deep learning model is trained for negative samples to obtain a pre-trained deep learning model.
  • determining the number of people boarding and alighting includes: obtaining a video frame set including the boarding and alighting behavior if there is a boarding and alighting behavior; and identifying the vehicle to be detected in the video frame set.
  • Vehicle position and personnel position of each boarding and alighting person According to the time sequence, analyze the change of the position of each of the boarding and alighting person to obtain the moving direction of each boarding and alighting person; The moving direction of the person deviates from the vehicle position, and the boarding person is determined to get off; for each of the boarding persons, if the moving direction of the boarding person points to the vehicle position, the boarding person is determined to get on.
  • the processor 601 when the processor 601 is configured to execute a program stored in the memory 602, the processor 601 can also implement any of the foregoing methods for warning an overload of a vehicle.
  • the electronic device in the embodiment of the present application further includes a communication interface and a communication bus, wherein the processor 601, the communication interface, and the memory 602 complete communication with each other through the communication bus.
  • the communication bus mentioned in the above electronic device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc.
  • PCI Peripheral Component Interconnect
  • EISA Extended Industry Standard Architecture
  • the communication bus can be divided into an address bus, a data bus, a control bus, and the like.
  • the figure only uses a thick line to represent, but it does not mean that there is only one bus or one type of bus.
  • the communication interface is used for communication between the aforementioned electronic device and other devices.
  • the memory may include Random Access Memory (RAM), and may also include Non-Volatile Memory (NVM), such as at least one disk memory.
  • RAM Random Access Memory
  • NVM Non-Volatile Memory
  • the memory may also be at least one storage device located far from the foregoing processor.
  • the aforementioned processor 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 (Digital Signal Processing, DSP), special integration Circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
  • CPU central processing unit
  • NP network processor
  • DSP Digital Signal Processing
  • ASIC Application Specific Integrated Circuit
  • FPGA Field-Programmable Gate Array
  • An embodiment of the present application further provides a computer-readable storage medium.
  • a computer program is stored in the computer-readable storage medium. When the computer program is executed by a processor, the following steps are implemented:
  • target image data including the vehicle to be detected; in the target image data, obtain the identification information of the vehicle to be detected; determine the passenger carrying threshold and the current number of passengers of the vehicle to be detected according to the identification information; and according to the target image Data to determine whether there is a boarding and alighting behavior; if there is a boarding and alighting behavior, determine the number of people boarding and alighting; update the current number of passengers based on the number of people boarding and alighting; determine the size of the updated current number of passengers and the threshold of the number of passengers; The updated current number of passengers is greater than the above-mentioned passenger threshold, which triggers an overload alarm for the vehicle to be detected.
  • an automatic alarm for vehicle overload can be realized, and the monitoring equipment for collecting target image data may be off-board, and the public monitoring equipment for urban traffic can be directly used to collect target image data.
  • the supervisory authority does not need to To install equipment on the vehicle to be monitored, it is only necessary to add the above functions to the existing urban intelligent traffic monitoring system to complete the capture of overload behavior, the cost is greatly reduced, and the situation of human interference by the supervised person can be reduced.
  • any one of the foregoing vehicle overload alarm methods can also be implemented.
  • An embodiment of the present application further provides a computer program product.
  • the computer program product is executed by a computer, the following steps are implemented:
  • target image data including the vehicle to be detected; in the target image data, obtain the identification information of the vehicle to be detected; determine the passenger carrying threshold and the current number of passengers of the vehicle to be detected according to the identification information; and according to the target image Data to determine whether there is a boarding and alighting behavior; if there is a boarding and alighting behavior, determine the number of people boarding and alighting; update the current number of passengers based on the number of people boarding and alighting; determine the size of the updated current number of passengers and the threshold of the number of passengers; The updated current number of passengers is greater than the above-mentioned passenger threshold, which triggers an overload alarm for the vehicle to be detected.
  • an automatic alarm for vehicle overload can be realized, and the monitoring equipment for collecting target image data may be off-board, and the public monitoring equipment for urban traffic can be directly used to collect target image data.
  • the supervisory authority does not need to To install equipment on the vehicle to be monitored, it is only necessary to add the above functions to the existing urban intelligent traffic monitoring system to complete the capture of overload behavior, the cost is greatly reduced, and the situation of human interference by the supervised person can be reduced.
  • any one of the foregoing vehicle overload alarm methods can also be implemented.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Mechanical Engineering (AREA)
  • Data Mining & Analysis (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Traffic Control Systems (AREA)
  • Image Analysis (AREA)

Abstract

一种车辆超载报警方法、装置、电子设备及存储介质，该方法包括获取包含待检测车辆的目标图像数据；在目标图像数据中，获取待检测车辆的识别信息；按照识别信息，确定待检测车辆的载客阈值及当前载客数；根据图像数据，判断是否存在上下车行为；若存在上下车行为，确定上下车人数；按照上下车人数，更新当前载客数；判断更新后的当前载客数与载客阈值的大小；若更新后的当前载客数大于载客阈值，触发针对待检测车辆的超载报警。通过目标图像数据，确定待检测车辆的载客阈值及当前载客数，在当前载客数大于载客阈值时，触发针对待检测车辆的超载报警。本申请实施例的车辆超载报警方法可以实现对车辆的超载进行报警。

Description

车辆超载报警方法、装置、电子设备及存储介质
本申请要求于2018年09月12日提交中国专利局、申请号为201811061118.X发明名称为“车辆超载报警方法、装置、电子设备及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及智能交通技术领域，特别是涉及车辆超载报警方法、装置、电子设备及存储介质。
背景技术
车辆超载会造成车辆的重量增加，从而导致车辆制动距离增加，危险性增大。车辆超载还会增加轮胎的负荷，造成轮胎变形或引起发爆胎，从而导致危险。另外，超载还会影响车辆的转向性能，易因转向失控而导致事故。驾驶人驾驶超限超载的车辆，往往会增加心理负担和思想压力，容易出现操作错误，从而引发交通事故。并且超载车辆一旦发生交通事故，将带来额外的人员伤害。
在城市交通中，出租车、私家车等车辆的超载行为较为普遍，因此希望对车辆超载进行报警。
发明内容
本申请实施例的目的在于提供一种车辆超载报警方法、装置、电子设备及存储介质，以实现对车辆的超载进行报警。具体技术方案如下：
第一方面，本申请实施例提供了一种车辆超载报警方法，所述方法包括：
获取包含待检测车辆的目标图像数据；在所述目标图像数据中，获取所述待检测车辆的识别信息；按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数；根据所述目标图像数据，判断是否存在上下车行为；若存在上下车行为，确定上下车人数；按照所述上下车人数，更新所述当前载客数；判断更新后的当前载客数与所述载客阈值的大小；若更新后的当前载客数大于所述载客阈值，触发针对所述待检测车辆的超载报警。
在一种可能的实施方式中，在本申请实施例的车辆超载报警方法中，所 述识别信息包括车牌信息及车型信息；所述按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数，包括：按照所述车牌信息，获取所述待检测车辆的当前载客数；按照所述车型信息，获取所述待检测车辆的载客阈值。
在一种可能的实施方式中，在本申请实施例的车辆超载报警方法中，所述识别信息包括车型信息及车牌信息，所述方法还包括：
按照所述车型信息，判断所述待检测车辆的车型是否为可判定车型；所述按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数，包括：若所述待检测车辆的车型为可判定车型，按照所述车型信息，确定所述车型信息对应的载客阈值，得到所述待检测车辆的载客阈值；按照所述车牌信息，获取所述待检测车辆的当前载客数。
在一种可能的实施方式中，所述根据所述图像数据，判断是否存在上下车行为，包括：通过预先训练的深度学习模型对所述图像数据进行检测分析，判断是否存在上下车行为；预先训练深度学习模型的过程包括：获取多个包含上下车行为的图像数据及多个不包含上下车行为的图像数据；将包含上下车行为的图像数据标定为正样本，将不包含上下车行为的图像数据标定为负样本对深度学习模型进行训练，得到预先训练的深度学习模型。
在一种可能的实施方式中，所述若存在上下车行为，确定上下车人数，包括：若存在上下车行为，获取包含上下车行为的视频帧集合；识别所述视频帧集合中所述待检测车辆的车辆位置及各上下车人员的人员位置；按照时序顺序，分析各所述上下车人员的人员位置的变化情况，得到各所述上下车人员的移动方向；针对每个所述上下车人员，若该上下车人员的移动方向背离所述车辆位置，判定该上下车人员下车；针对每个所述上下车人员，若该上下车人员的移动方向指向所述车辆位置，判定该上下车人员上车。
第二方面，本申请实施例提供了一种车辆超载报警装置，所述装置包括：
图像数据获取模块，用于获取包含待检测车辆的目标图像数据；
识别信息获取模块，用于在所述目标图像数据中，获取所述待检测车辆的识别信息；
第一计算模块，用于按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数；
第一判断模块，用于图像数据获取模块，用于根据所述目标图像数据，判断是否存在上下车行为；
第二计算模块，用于若存在上下车行为，确定上下车人数；
人数更新模块，用于按照所述上下车人数，更新所述当前载客数；
第二判断模块，用于判断更新后的当前载客数与所述载客阈值的大小；
超载报警模块，用于若更新后的当前载客数大于所述载客阈值，触发针对所述待检测车辆的超载报警。
在一种可能的实施方式中，在本申请实施例的车辆超载报警装置中，所述识别信息包括车牌信息及车型信息；
所述第一计算模块，包括：
人数获取子模块，用于按照所述车牌信息，获取所述待检测车辆的当前载客数；
阈值获取子模块，用于按照所述车型信息，获取所述待检测车辆的载客阈值。
在一种可能的实施方式中，在本申请实施例的车辆超载报警装置中，所述识别信息包括车型信息及车牌信息，所述装置还包括：
第三判定模块，用于按照所述车型信息，判断所述待检测车辆的车型是否为可判定车型；
所述第一计算模块，具体用于：
若所述待检测车辆的车型为可判定车型，按照所述车型信息，确定所述车型信息对应的载客阈值，得到所述待检测车辆的载客阈值；按照所述车牌信息，获取所述待检测车辆的当前载客数。
在一种可能的实施方式中，所述第一判断模块，具体用于：
通过预先训练的深度学习模型对所述图像数据进行检测分析，判断是否存在上下车行为；
其中，预先训练深度学习模型的过程包括：
获取多个包含上下车行为的图像数据及多个不包含上下车行为的图像数据；
将包含上下车行为的图像数据标定为正样本，将不包含上下车行为的图像数据标定为负样本对深度学习模型进行训练，得到预先训练的深度学习模型。
在一种可能的实施方式中，所述第二计算模块，包括：
视频帧获取子模块，用于若存在上下车行为，获取包含上下车行为的视频帧集合；
位置确定子模块，用于识别所述视频帧集合中所述待检测车辆的车辆位置及各上下车人员的人员位置；
方向确定子模块，用于按照时序顺序，分析各所述上下车人员的人员位置的变化情况，得到各所述上下车人员的移动方向；
下车判定子模块，用于针对每个所述上下车人员，若该上下车人员的移动方向背离所述车辆位置，判定该上下车人员下车；
上车判定子模块，用于针对每个所述上下车人员，若该上下车人员的移动方向指向所述车辆位置，判定该上下车人员上车。
第三方面，本申请实施例提供了一种电子设备，包括处理器及存储器；
所述存储器，用于存放计算机程序；所述处理器，用于执行所述存储器上所存放的程序时，实现如下方法步骤：获取包含待检测车辆的目标图像数据；在所述目标图像数据中，获取所述待检测车辆的识别信息；按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数；根据所述目标图像数据，判断是否存在上下车行为；若存在上下车行为，确定上下车人数；按照所述上下车人数，更新所述当前载客数；判断更新后的当前载客数与所述载客阈值的大小；若更新后的当前载客数大于所述载客阈值，触发针对所述 待检测车辆的超载报警。
在一种可能的实施方式中，所述识别信息包括车牌信息及车型信息；所述按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数，包括：按照所述车牌信息，获取所述待检测车辆的当前载客数；按照所述车型信息，获取所述待检测车辆的载客阈值。
在一种可能的实施方式中，所述识别信息包括车型信息及车牌信息，所述处理器还用于执行如下步骤：按照所述车型信息，判断所述待检测车辆的车型是否为可判定车型；所述按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数，包括：若所述待检测车辆的车型为可判定车型，按照所述车型信息，确定所述车型信息对应的载客阈值，得到所述待检测车辆的载客阈值；按照所述车牌信息，获取所述待检测车辆的当前载客数。
在一种可能的实施方式中，所述根据所述图像数据，判断是否存在上下车行为，包括：通过预先训练的深度学习模型对所述图像数据进行检测分析，判断是否存在上下车行为；预先训练深度学习模型的过程包括：获取多个包含上下车行为的图像数据及多个不包含上下车行为的图像数据；将包含上下车行为的图像数据标定为正样本，将不包含上下车行为的图像数据标定为负样本对深度学习模型进行训练，得到预先训练的深度学习模型。
在一种可能的实施方式中，所述若存在上下车行为，确定上下车人数，包括：若存在上下车行为，获取包含上下车行为的视频帧集合；识别所述视频帧集合中所述待检测车辆的车辆位置及各上下车人员的人员位置；按照时序顺序，分析各所述上下车人员的人员位置的变化情况，得到各所述上下车人员的移动方向；针对每个所述上下车人员，若该上下车人员的移动方向背离所述车辆位置，判定该上下车人员下车；针对每个所述上下车人员，若该上下车人员的移动方向指向所述车辆位置，判定该上下车人员上车。
第四方面，本申请实施例提供了一种计算机可读存储介质，所述计算机可读存储介质内存储有计算机程序，所述计算机程序被处理器执行时实现上述第一方面任一所述的车辆超载报警方法。
本申请实施例提供的车辆超载报警方法、装置、电子设备及存储介质， 获取包含待检测车辆的目标图像数据；在目标图像数据中，获取待检测车辆的识别信息；按照识别信息，确定待检测车辆的载客阈值及当前载客数；根据图像数据，判断是否存在上下车行为；若存在上下车行为，确定上下车人数；按照上下车人数，更新当前载客数；判断更新后的当前载客数与载客阈值的大小；若更新后的当前载客数大于载客阈值，触发针对待检测车辆的超载报警。通过目标图像数据，确定待检测车辆的载客阈值及当前载客数，在当前载客数大于载客阈值时，触发针对待检测车辆的超载报警，可以实现对车辆的超载进行报警。当然，实施本申请的任一产品或方法并不一定需要同时达到以上所述的所有优点。
附图说明
为了更清楚地说明本申请实施例和现有技术的技术方案，下面对实施例和现有技术中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1为本申请实施例的车辆超载报警装置的一种示意图；
图2a为本申请实施例的人员行使方向的判断方法的第一种示意图；
图2b为本申请实施例的人员行使方向的判断方法的第二种示意图；
图2c为本申请实施例的人员行使方向的判断方法的第三种示意图；
图3为本申请实施例的车辆超载报警方法的一种流程示意图；
图4为本申请实施例的车辆超载报警方法的另一种流程示意图；
图5为本申请实施例的车辆超载报警装置的另一种示意图；
图6为本申请实施例的电子设备的一种示意图。
具体实施方式
为使本申请的目的、技术方案、及优点更加清楚明白，以下参照附图并举实施例，对本申请进一步详细说明。显然，所描述的实施例仅仅是本申请 一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
术语解释：
智能交通：是将信息技术、数据通讯传输技术、电子传感技术、控制技术及计算机技术等有效地集成运用于整个地面交通管理系统而建立的，一种在大范围内、全方位发挥作用的综合交通运输管理系统，具备实时、准确、高效的特征。它的突出特点是以信息的收集、处理、发布、交换、分析、利用为主线，为交通参与者提供多样性的服务。
目标检测：也叫目标提取，是将人们感兴趣的物体，从复杂的背景图像中提取出来。
图像分类：根据目标在图像信息中所反映的不同特征，把不同类别的目标区分开来的图像处理方法。它利用计算机对图像进行定量分析，把图像或图像中的每个像元或区域划归为若干个类别中的某一种，以代替人的视觉判读。
目标识别：用于将指定目标，从其他目标中分离出来。
CNN(Convolutional Neural Networks，卷积神经网络)，是一种深度的监督学习下的机器学习模型，是第一个能够成功训练多层网络结构的学习算法。它利用空间关系减少需要学习的参数数目以提高一般前向BP(Error Back Propagation)算法的训练性能。
Faster RCNN(Faster Regions Convolutional Neural Network)：是基于深度学习RCNN(Regions With Convolutional Neural Network Features)系列目标检测最好的方法，是一个CNN网络实现端到端目标检测的框架。其中的R对应于Region，即区域。
YOLO(You Only Look Once)：YOLO是一种非常成功的实时目标检测算法，目前已成为众多分类器、分割、人体姿态和行为分类的重要部分，是一个CNN网络实现端到端的目标检测框架。
在城市交通中，出租车、私有车辆的超载行为非常普遍。超载行为存在 显然的交通安全隐患，一旦发生交通事故，将带来额外的人员、经济损失。
本申请实施例的车辆超载报警方法、装置、电子设备及存储介质可以应用于城市道路监控视频分析，通过判断车型、统计车辆短时间内的下客人数，判断车辆是否存在超员行为。并通过记录车牌号，记录该行为，用于交通部门规范车辆行为。
本申请实施例的主要的监管对象可以为城市内的出租车及私家车等小型车辆。参见图1，本申请实施例的车辆超载报警装置包括：车牌识别模块101、车型识别模块102、上下车人数统计模块103及判别报警模块104。车牌识别模块101用于识别车辆的车牌，车型识别模块102用于识别车辆的车型，在相关技术中已经较为成熟，可以直接复用相关技术。判别报警模块104用于在车辆承载的人数超过该车辆的承载人数阈值时进行报警。
上下车人数统计模块103用于统计车辆的上下车人数，可以通过跟踪车辆两边的行人来实现上下车行为的统计。其中，针对每个行人的上下车行为，通过目标检测算法以行人作为目标进行检测；生成行人移动轨迹；根据轨迹方向判别上下客行为。综合每个行人的上下车行为统计总的上/下车人数，进而得到车辆的承载的人数。
例如，通过目标检测算法对视频数据进行检测，检测出行人的位置如图2a中矩形框所示。继续对视频数据进行检测，检测出该行人新的位置如图2b中的矩形框所示，与上一帧行人位置(如图2b中虚线框所示)比较，重叠度大于0.5，关联两帧视频帧中的目标(行人)。继续对视频数据进行分析，关联视频帧多帧的行人检测结果如图2c所示，其中，#1、#2、#3、#4分别为四帧视频帧中目标的位置。此处仅以四帧视频帧的关联为例，实际检测过程中关联帧数可以自行设定。通过多帧的结果计算行人轨迹方向为离开车辆，判定为下车行为。
具体的，本申请实施例的车辆超载报警方法的流程可以如图3所示，通过本申请实施例，能够实现自动识别超载行为，通过对城市道路监控视频进行分析，定位超员车辆，并记录其违规行为，有助于加强交通管理部门对超载行为的监控。
本申请实施例中采集图像的监控器材可以是非车载的，可以直接利用城市交通的公共监控器材，带来的优势是：监管部门不需要在每一辆待监管车辆上安装器材，而只是需要在现有的城市智能交通监控系统中，增加上述功能，即可完成超载行为的抓捕，成本大幅度降低。并且能够减少被监管者人为干扰的情况，例如：破坏或通过涂抹、遮挡，调整摄像头方向等阻碍车载相机。并且可以结合车型，判断不同车型车辆的超载。例如，出租车下客人数上限为4人，普通小轿车下客人数上限为5人等。
为了实现对超载车辆进行报警，本申请实施例提供了一种车辆超载报警方法，参见图4，该方法包括：
S401，获取包含待检测车辆的目标图像数据。
本申请实施例中的车辆超载报警方法可以通过报警系统实现，报警系统为任意能够实现本申请实施例的车辆超载报警方法的系统。例如：
报警系统可以为一种设备，包括：处理器、存储器、通信接口和总线；处理器、存储器和通信接口通过总线连接并完成相互间的通信；存储器存储可执行程序代码；处理器通过读取存储器中存储的可执行程序代码来运行与可执行程序代码对应的程序，以用于执行本申请实施例的车辆超载报警方法。
报警系统还可以为一种应用程序，用于在运行时执行本申请实施例的车辆超载报警方法。
报警系统还可以为一种存储介质，用于存储可执行代码，可执行代码用于执行本申请实施例的车辆超载报警方法。
报警系统通过图像采集设备获取包含待检测车辆的图像数据，即目标图像数据。图像采集设备可以为摄像头等设备，在一种可能的实施方式中，图像采集设备为城市交通监控系统中的摄像头。
S402，在上述目标图像数据中，获取上述待检测车辆的识别信息。
通过目标检测算法，对目标图像数据进行分析，得到待检测车辆的识别信息，识别信息用于唯一识别待检测车辆。例如，通过RCNN、Faster RCNN或YOLO等对目标图像数据进行分析，得到待检测车辆的车牌信息。
S403，按照上述识别信息，确定上述待检测车辆的载客阈值及当前载客数。
识别信息为车辆的唯一标识信息，报警系统会保持记录各车辆的当前载客数，同时每个车辆会对应相应的载客阈值，车辆的载客阈值根据相关交通法规进行设定，例如，类型为出租车的车辆对应的载客阈值为4，类型为私家车的车辆对应的载客阈值为5等。
报警系统会保持记录各车辆的当前载客数，在停车点的车辆启动时，获取车辆启动前后的视频数据，作为目标视频数据，并默认该车辆初始的当前载客数为0，通过本申请实施例的车辆超载报警方法，更新该车辆当前载客数。后续继续通过本申请实施例的车辆超载报警方法保持记录该车辆的当前载客数。报警系统可以按照识别信息，从记录中获取识别信息表示的车辆的当前载客数。
S404，根据上述目标图像数据，判断是否存在上下车行为。
报警系统通过预设算法，例如，特征对比算法或神经网络算法等，对目标图像数据进行分析，判断是否存在上下车行为。
在一种可能的实施方式中，上述根据上述图像数据，判断是否存在上下车行为，包括：
通过预先训练的深度学习模型对上述图像数据进行检测分析，判断是否存在上下车行为；
预先训练深度学习模型的过程包括：
步骤一，获取多个包含上下车行为的图像数据及多个不包含上下车行为的图像数据。
步骤二，将包含上下车行为的图像数据标定为正样本，将不包含上下车行为的图像数据标定为负样本对深度学习模型进行训练，得到预先训练的深度学习模型。
此处的深度学习模型可以为Faster RCNN、YOLO、RCNN或DMP(Deformable Parts Model，可变形部件模型)等。
S405，若存在上下车行为，确定上下车人数。
在存在上下车行为时，报警系统确定上下车人数。例如，报警系统可以通过预先训练的卷积神经网络，对目标视频数据中包含上下车行为的视频帧进行分析，确定上下车人数。
预先训练的卷积神经网络的步骤包括：
步骤一，获取多个包含上下车行为的第一图像数据及多个不包含上下车行为的第二图像数据。
步骤二，标定各第一图像数据中的上车行为及上车行为的人数，标定各图像数据中的下车行为及下车行为的人数，得到标定后的第一图像数据。
步骤三，将标定后的第一图像数据作为正样本，将第二图像数据作为负样本对卷积神经网络进行训练，得到预先训练的卷积神经网络。
S406，按照上述上下车人数，更新上述当前载客数。
报警系统每检测到一人下车，将待检测车辆的当前载客数减1，报警系统每检测到一人上车，将待检测车辆的当前载客数加1，以实现更新当前载客数。
S407，判断更新后的当前载客数与上述载客阈值的大小。
S408，若更新后的当前载客数大于上述载客阈值，触发针对上述待检测车辆的超载报警。
报警系统触发针对上述待检测车辆的超载报警，例如，报警系统向订阅端报警待检测车辆的识别信息及当前载客人数，同时还可以报警待检测车辆的位置，其中，待检测车辆的位置可以通过采集目标视频数据的摄像头获取。
在本申请实施例中，可以实现车辆超载的自动报警，并且采集目标图像数据的监控器材可以是非车载的，能够直接利用城市交通的公共监控器材采集目标图像数据，监管部门不需要在每一辆待监测车辆上安装器材，而只是需要在现有的城市智能交通监控系统中，增加上述功能，即可完成超载行为的抓捕，成本大幅度降低，并且能够减少被监管者人为干扰的情况。
在一种可能的实施方式中，上述识别信息包括车牌信息及车型信息。
相应的，上述按照上述识别信息，确定上述待检测车辆的载客阈值及当前载客数，包括：
步骤一，按照上述车牌信息，获取上述待检测车辆的当前载客数。
报警系统将各车辆的载客数存储在指定的存储器中。在需要使用时，报警系统按照车牌信息，在存储器中读取该待检测车辆的当前载客数。
步骤二，按照上述车型信息，获取上述待检测车辆的载客阈值。
按照待检测车辆的车型信息，确定待检测车辆的车型对应的载客阈值，即待检测车辆的载客阈值，例如，出租车载客阈值为4人，普通小轿车载客阈值为5人等。
在本申请实施例中，可以通过车牌信息唯一确定待检测车辆，可以通过车型信息，确定待检测车辆的载客阈值，方便快捷。
在一种可能的实施方式中，上述识别信息包括车型信息及车牌信息，上述方法还包括：
按照上述车型信息，判断上述待检测车辆的车型是否为可判定车型。
相应的，上述按照上述识别信息，确定上述待检测车辆的载客阈值及当前载客数，包括：
若上述待检测车辆的车型为可判定车型，按照上述车型信息，确定上述车型信息对应的载客阈值，得到上述待检测车辆的载客阈值；按照上述车牌信息，获取上述待检测车辆的当前载客数。
在报警系统判定上述待检测车辆的车型不为可判定车型，结束本次超载判定。
可判定车型可以根据实际情况自行设定，例如，可判定车型可以包括出租车、私家车及商务车等。
在本申请实施例中，报警系统可以对可判定车型进行设定，仅针对指定的车型进行上下车行为及后续判断，可以避免对可判定车型外的车辆的超载判定，节约计算资源。
在一种可能的实施方式中，上述若存在上下车行为，确定上下车人数，包括：
步骤一，若存在上下车行为，获取包含上下车行为的视频帧集合。
在报警系统判定目标视频数据中存在上下车行为时，报警系统提取包含上下车行为的特征的视频帧集合。
步骤二，识别上述视频帧集合中上述待检测车辆的车辆位置及各上下车人员的人员位置。
报警系统通过目标识别算法，确定在视频帧集合的各视频帧中，待检测车辆的车辆位置及各上下车人员的人员位置。
步骤三，按照时序顺序，分析各上述上下车人员的人员位置的变化情况，得到各上述上下车人员的移动方向。
按照视频帧的时序顺序，分析上下车人员的人员位置的变化情况，得到上下车人员的移动方向。例如，针对每个上下车人员，通过目标跟踪算法确定上下车人员的人员位置的变化情况，从而得到上下车人员的移动方向；或通过计算上下车人员的人员位置中心点的变化情况，得到上下车人员的移动方向。
按照视频帧的时序顺序，选取上下车人员的人员位置的中心点，其中，中心点的坐标表示为(X i，Y i)。其中，X i表示第i帧视频帧的X坐标，Y i表示第i帧视频帧的Y坐标，i的取值范围为[1，N]，N为视频帧数量。通过
Figure PCTCN2019103030-appb-000001
其中，t为平滑间隔，可以取1-5。
若D1～D5中大于0的数量，多于D1～D5中小于0的数量，则判断人员延X轴正方向移动，反之则判断人员延X轴负方向移动。
类似的，可以得到人员延Y轴的移动方向。
步骤四，针对每个上述上下车人员，若该上下车人员的移动方向背离上述车辆位置，判定该上下车人员下车。
步骤五，针对每个上述上下车人员，若该上下车人员的移动方向指向上述车辆位置，判定该上下车人员上车。
在本申请实施例中，给出了统计上下车人数的具体方法，计算方便。
本申请实施例还提供了一种车辆超载报警装置，参见图5，该装置包括：
图像数据获取模块501，用于获取包含待检测车辆的目标图像数据。
识别信息获取模块502，用于在上述目标图像数据中，获取上述待检测车辆的识别信息。
第一计算模块503，用于按照上述识别信息，确定上述待检测车辆的载客阈值及当前载客数。
第一判断模块504，用于图像数据获取模块，用于根据上述目标图像数据，判断是否存在上下车行为。
第二计算模块505，用于若存在上下车行为，确定上下车人数。
人数更新模块506，用于按照上述上下车人数，更新上述当前载客数。
第二判断模块507，用于判断更新后的当前载客数与上述载客阈值的大小。
超载报警模块508，用于若更新后的当前载客数大于上述载客阈值，触发针对上述待检测车辆的超载报警。
在本申请实施例中，可以实现车辆超载的自动报警，并且采集目标图像数据的监控器材可以是非车载的，能够直接利用城市交通的公共监控器材采集目标图像数据，监管部门不需要在每一辆待监测车辆上安装器材，而只是需要在现有的城市智能交通监控系统中，增加上述功能，即可完成超载行为的抓捕，成本大幅度降低，并且能够减少被监管者人为干扰的情况。
在一种可能的实施方式中，在本申请实施例的车辆超载报警装置中，上述识别信息包括车牌信息及车型信息；
上述第一计算模块503，包括：
人数获取子模块，用于按照上述车牌信息，获取上述待检测车辆的当前载客数。
阈值获取子模块，用于按照上述车型信息，获取上述待检测车辆的载客阈值。
在一种可能的实施方式中，在本申请实施例的车辆超载报警装置中，上述识别信息包括车型信息及车牌信息，上述装置还包括：
第三判定模块，用于按照上述车型信息，判断上述待检测车辆的车型是否为可判定车型。
上述第一计算模块503，具体用于：
若上述待检测车辆的车型为可判定车型，按照上述车型信息，确定上述车型信息对应的载客阈值，得到上述待检测车辆的载客阈值；按照上述车牌信息，获取上述待检测车辆的当前载客数。
在一种可能的实施方式中，上述第一判断模块504，具体用于：
通过预先训练的深度学习模型对上述图像数据进行检测分析，判断是否存在上下车行为。
其中，预先训练深度学习模型的过程包括：
获取多个包含上下车行为的图像数据及多个不包含上下车行为的图像数据；
将包含上下车行为的图像数据标定为正样本，将不包含上下车行为的图像数据标定为负样本对深度学习模型进行训练，得到预先训练的深度学习模型。
在一种可能的实施方式中，上述第二计算模块505，包括：
视频帧获取子模块，用于若存在上下车行为，获取包含上下车行为的视频帧集合；
位置确定子模块，用于识别上述视频帧集合中上述待检测车辆的车辆位置及各上下车人员的人员位置。
方向确定子模块，用于按照时序顺序，分析各上述上下车人员的人员位置的变化情况，得到各上述上下车人员的移动方向。
下车判定子模块，用于针对每个上述上下车人员，若该上下车人员的移动方向背离上述车辆位置，判定该上下车人员下车。
上车判定子模块，用于针对每个上述上下车人员，若该上下车人员的移动方向指向上述车辆位置，判定该上下车人员上车。
本申请实施例提供了一种电子设备，参见图6，包括处理器601及存储器602；
上述存储器602，用于存放计算机程序；上述处理器601，用于执行上述存储器602上所存放的程序时，实现如下步骤：
获取包含待检测车辆的目标图像数据；在上述目标图像数据中，获取上述待检测车辆的识别信息；按照上述识别信息，确定上述待检测车辆的载客阈值及当前载客数；根据上述目标图像数据，判断是否存在上下车行为；若存在上下车行为，确定上下车人数；按照上述上下车人数，更新上述当前载客数；判断更新后的当前载客数与上述载客阈值的大小；若更新后的当前载客数大于上述载客阈值，触发针对上述待检测车辆的超载报警。
在本申请实施例中，可以实现车辆超载的自动报警，并且采集目标图像数据的监控器材可以是非车载的，能够直接利用城市交通的公共监控器材采集目标图像数据，监管部门不需要在每一辆待监测车辆上安装器材，而只是需要在现有的城市智能交通监控系统中，增加上述功能，即可完成超载行为的抓捕，成本大幅度降低，并且能够减少被监管者人为干扰的情况。
在一种可能的实施方式中，上述识别信息包括车牌信息及车型信息；上述按照上述识别信息，确定上述待检测车辆的载客阈值及当前载客数，包括：按照上述车牌信息，获取上述待检测车辆的当前载客数；按照上述车型信息，获取上述待检测车辆的载客阈值。
在一种可能的实施方式中，上述识别信息包括车型信息及车牌信息，上述处理器还用于执行如下步骤：按照上述车型信息，判断上述待检测车辆的车型是否为可判定车型；上述按照上述识别信息，确定上述待检测车辆的载 客阈值及当前载客数，包括：若上述待检测车辆的车型为可判定车型，按照上述车型信息，确定上述车型信息对应的载客阈值，得到上述待检测车辆的载客阈值；按照上述车牌信息，获取上述待检测车辆的当前载客数。
在一种可能的实施方式中，上述根据上述图像数据，判断是否存在上下车行为，包括：通过预先训练的深度学习模型对上述图像数据进行检测分析，判断是否存在上下车行为；预先训练深度学习模型的过程包括：获取多个包含上下车行为的图像数据及多个不包含上下车行为的图像数据；将包含上下车行为的图像数据标定为正样本，将不包含上下车行为的图像数据标定为负样本对深度学习模型进行训练，得到预先训练的深度学习模型。
在一种可能的实施方式中，上述若存在上下车行为，确定上下车人数，包括：若存在上下车行为，获取包含上下车行为的视频帧集合；识别上述视频帧集合中上述待检测车辆的车辆位置及各上下车人员的人员位置；按照时序顺序，分析各上述上下车人员的人员位置的变化情况，得到各上述上下车人员的移动方向；针对每个上述上下车人员，若该上下车人员的移动方向背离上述车辆位置，判定该上下车人员下车；针对每个上述上下车人员，若该上下车人员的移动方向指向上述车辆位置，判定该上下车人员上车。
在一种可能的实施方式中，上述处理器601，用于执行上述存储器602上所存放的程序时，还能够实现上述任一车辆超载报警方法。
在一种可能的实施方式中，本申请实施例的电子设备还包括，通信接口和通信总线，其中，处理器601，通信接口，存储器602通过通信总线完成相互间的通信。
上述电子设备提到的通信总线可以是外设部件互连标准(Peripheral Component Interconnect，PCI)总线或扩展工业标准结构(Extended Industry Standard Architecture，EISA)总线等。该通信总线可以分为地址总线、数据总线、控制总线等。为便于表示，图中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。
通信接口用于上述电子设备与其他设备之间的通信。
存储器可以包括随机存取存储器(Random Access Memory，RAM)，也可 以包括非易失性存储器(Non-Volatile Memory，NVM)，例如至少一个磁盘存储器。在一种可能的实施方式中，存储器还可以是至少一个位于远离前述处理器的存储装置。
上述的处理器可以是通用处理器，包括中央处理器(Central Processing Unit，CPU)、网络处理器(Network Processor，NP)等；还可以是数字信号处理器(Digital Signal Processing，DSP)、专用集成电路(Application Specific Integrated Circuit，ASIC)、现场可编程门阵列(Field-Programmable Gate Array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
本申请实施例还提供了一种计算机可读存储介质，上述计算机可读存储介质内存储有计算机程序，上述计算机程序被处理器执行时实现如下步骤：
获取包含待检测车辆的目标图像数据；在上述目标图像数据中，获取上述待检测车辆的识别信息；按照上述识别信息，确定上述待检测车辆的载客阈值及当前载客数；根据上述目标图像数据，判断是否存在上下车行为；若存在上下车行为，确定上下车人数；按照上述上下车人数，更新上述当前载客数；判断更新后的当前载客数与上述载客阈值的大小；若更新后的当前载客数大于上述载客阈值，触发针对上述待检测车辆的超载报警。
在本申请实施例中，可以实现车辆超载的自动报警，并且采集目标图像数据的监控器材可以是非车载的，能够直接利用城市交通的公共监控器材采集目标图像数据，监管部门不需要在每一辆待监测车辆上安装器材，而只是需要在现有的城市智能交通监控系统中，增加上述功能，即可完成超载行为的抓捕，成本大幅度降低，并且能够减少被监管者人为干扰的情况。
在一种可能的实施方式中，上述计算机程序被处理器执行时，还能够实现上述任一车辆超载报警方法。
本申请实施例还提供了一种计算机程序产品，上述计算机程序产品被计算机执行时，实现如下步骤：
获取包含待检测车辆的目标图像数据；在上述目标图像数据中，获取上述待检测车辆的识别信息；按照上述识别信息，确定上述待检测车辆的载客 阈值及当前载客数；根据上述目标图像数据，判断是否存在上下车行为；若存在上下车行为，确定上下车人数；按照上述上下车人数，更新上述当前载客数；判断更新后的当前载客数与上述载客阈值的大小；若更新后的当前载客数大于上述载客阈值，触发针对上述待检测车辆的超载报警。
在本申请实施例中，可以实现车辆超载的自动报警，并且采集目标图像数据的监控器材可以是非车载的，能够直接利用城市交通的公共监控器材采集目标图像数据，监管部门不需要在每一辆待监测车辆上安装器材，而只是需要在现有的城市智能交通监控系统中，增加上述功能，即可完成超载行为的抓捕，成本大幅度降低，并且能够减少被监管者人为干扰的情况。
在一种可能的实施方式中，上述计算机程序产品被计算机执行时，还能够实现上述任一车辆超载报警方法。
需要说明的是，在本文中，诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来，而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
本说明书中的各个实施例均采用相关的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于装置、电子设备及存储介质的实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
以上所述仅为本申请的较佳实施例而已，并不用以限制本申请，凡在本申请的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本申请保护的范围之内。

Claims (16)

  1. 一种车辆超载报警方法，其特征在于，所述方法包括：
    获取包含待检测车辆的目标图像数据；
    在所述目标图像数据中，获取所述待检测车辆的识别信息；
    按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数；
    根据所述目标图像数据，判断是否存在上下车行为；
    若存在上下车行为，确定上下车人数；
    按照所述上下车人数，更新所述当前载客数；
    判断更新后的当前载客数与所述载客阈值的大小；
    若更新后的当前载客数大于所述载客阈值，触发针对所述待检测车辆的超载报警。
  2. 根据权利要求1所述的方法，其特征在于，所述识别信息包括车牌信息及车型信息；
    所述按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数，包括：
    按照所述车牌信息，获取所述待检测车辆的当前载客数；
    按照所述车型信息，获取所述待检测车辆的载客阈值。
  3. 根据权利要求1所述的方法，其特征在于，所述识别信息包括车型信息及车牌信息，所述方法还包括：
    按照所述车型信息，判断所述待检测车辆的车型是否为可判定车型；
    所述按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数，包括：
    若所述待检测车辆的车型为可判定车型，按照所述车型信息，确定所述车型信息对应的载客阈值，得到所述待检测车辆的载客阈值；按照所述车牌信息，获取所述待检测车辆的当前载客数。
  4. 根据权利要求1所述的方法，其特征在于，所述根据所述图像数据，判断是否存在上下车行为，包括：
    通过预先训练的深度学习模型对所述图像数据进行检测分析，判断是否存在上下车行为；
    预先训练深度学习模型的过程包括：
    获取多个包含上下车行为的图像数据及多个不包含上下车行为的图像数据；
    将包含上下车行为的图像数据标定为正样本，将不包含上下车行为的图像数据标定为负样本对深度学习模型进行训练，得到预先训练的深度学习模型。
  5. 根据权利要求1所述的方法，其特征在于，所述若存在上下车行为，确定上下车人数，包括：
    若存在上下车行为，获取包含上下车行为的视频帧集合；
    识别所述视频帧集合中所述待检测车辆的车辆位置及各上下车人员的人员位置；
    按照时序顺序，分析各所述上下车人员的人员位置的变化情况，得到各所述上下车人员的移动方向；
    针对每个所述上下车人员，若该上下车人员的移动方向背离所述车辆位置，判定该上下车人员下车；
    针对每个所述上下车人员，若该上下车人员的移动方向指向所述车辆位置，判定该上下车人员上车。
  6. 一种车辆超载报警装置，其特征在于，所述装置包括：
    图像数据获取模块，用于获取包含待检测车辆的目标图像数据；
    识别信息获取模块，用于在所述目标图像数据中，获取所述待检测车辆的识别信息；
    第一计算模块，用于按照所述识别信息，确定所述待检测车辆的载客阈 值及当前载客数；
    第一判断模块，用于图像数据获取模块，用于根据所述目标图像数据，判断是否存在上下车行为；
    第二计算模块，用于若存在上下车行为，确定上下车人数；
    人数更新模块，用于按照所述上下车人数，更新所述当前载客数；
    第二判断模块，用于判断更新后的当前载客数与所述载客阈值的大小；
    超载报警模块，用于若更新后的当前载客数大于所述载客阈值，触发针对所述待检测车辆的超载报警。
  7. 根据权利要求6所述的装置，其特征在于，所述识别信息包括车牌信息及车型信息；
    所述第一计算模块，包括：
    人数获取子模块，用于按照所述车牌信息，获取所述待检测车辆的当前载客数；
    阈值获取子模块，用于按照所述车型信息，获取所述待检测车辆的载客阈值。
  8. 根据权利要求6所述的装置，其特征在于，所述识别信息包括车型信息及车牌信息，所述装置还包括：
    第三判定模块，用于按照所述车型信息，判断所述待检测车辆的车型是否为可判定车型；
    所述第一计算模块，具体用于：
    若所述待检测车辆的车型为可判定车型，按照所述车型信息，确定所述车型信息对应的载客阈值，得到所述待检测车辆的载客阈值；按照所述车牌信息，获取所述待检测车辆的当前载客数。
  9. 根据权利要求6所述的装置，其特征在于，所述第一判断模块，具体用于：
    通过预先训练的深度学习模型对所述图像数据进行检测分析，判断是否存在上下车行为；
    其中，预先训练深度学习模型的过程包括：
    获取多个包含上下车行为的图像数据及多个不包含上下车行为的图像数据；
    将包含上下车行为的图像数据标定为正样本，将不包含上下车行为的图像数据标定为负样本对深度学习模型进行训练，得到预先训练的深度学习模型。
  10. 根据权利要求6所述的装置，其特征在于，所述第二计算模块，包括：
    视频帧获取子模块，用于若存在上下车行为，获取包含上下车行为的视频帧集合；
    位置确定子模块，用于识别所述视频帧集合中所述待检测车辆的车辆位置及各上下车人员的人员位置；
    方向确定子模块，用于按照时序顺序，分析各所述上下车人员的人员位置的变化情况，得到各所述上下车人员的移动方向；
    下车判定子模块，用于针对每个所述上下车人员，若该上下车人员的移动方向背离所述车辆位置，判定该上下车人员下车；
    上车判定子模块，用于针对每个所述上下车人员，若该上下车人员的移动方向指向所述车辆位置，判定该上下车人员上车。
  11. 一种电子设备，其特征在于，包括处理器及存储器；
    所述存储器，用于存放计算机程序；
    所述处理器，用于执行所述存储器上所存放的程序时，实现如下方法步骤：
    获取包含待检测车辆的目标图像数据；
    在所述目标图像数据中，获取所述待检测车辆的识别信息；
    按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数；
    根据所述目标图像数据，判断是否存在上下车行为；
    若存在上下车行为，确定上下车人数；
    按照所述上下车人数，更新所述当前载客数；
    判断更新后的当前载客数与所述载客阈值的大小；
    若更新后的当前载客数大于所述载客阈值，触发针对所述待检测车辆的超载报警。
  12. 根据权利要求11所述的电子设备，其特征在于，所述识别信息包括车牌信息及车型信息；
    所述按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数，包括：
    按照所述车牌信息，获取所述待检测车辆的当前载客数；
    按照所述车型信息，获取所述待检测车辆的载客阈值。
  13. 根据权利要求12所述的电子设备，其特征在于，所述识别信息包括车型信息及车牌信息，所述处理器还用于执行如下步骤：
    按照所述车型信息，判断所述待检测车辆的车型是否为可判定车型；
    所述按照所述识别信息，确定所述待检测车辆的载客阈值及当前载客数，包括：
    若所述待检测车辆的车型为可判定车型，按照所述车型信息，确定所述车型信息对应的载客阈值，得到所述待检测车辆的载客阈值；按照所述车牌信息，获取所述待检测车辆的当前载客数。
  14. 根据权利要求11所述的电子设备，其特征在于，所述根据所述图像数据，判断是否存在上下车行为，包括：
    通过预先训练的深度学习模型对所述图像数据进行检测分析，判断是否存在上下车行为；
    预先训练深度学习模型的过程包括：
    获取多个包含上下车行为的图像数据及多个不包含上下车行为的图像数据；
    将包含上下车行为的图像数据标定为正样本，将不包含上下车行为的图像数据标定为负样本对深度学习模型进行训练，得到预先训练的深度学习模型。
  15. 根据权利要求11所述的电子设备，其特征在于，所述若存在上下车行为，确定上下车人数，包括：
    若存在上下车行为，获取包含上下车行为的视频帧集合；
    识别所述视频帧集合中所述待检测车辆的车辆位置及各上下车人员的人员位置；
    按照时序顺序，分析各所述上下车人员的人员位置的变化情况，得到各所述上下车人员的移动方向；
    针对每个所述上下车人员，若该上下车人员的移动方向背离所述车辆位置，判定该上下车人员下车；
    针对每个所述上下车人员，若该上下车人员的移动方向指向所述车辆位置，判定该上下车人员上车。
  16. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质内存储有计算机程序，所述计算机程序被处理器执行时实现权利要求1-5任一所述的方法步骤。
PCT/CN2019/103030 2018-09-12 2019-08-28 车辆超载报警方法、装置、电子设备及存储介质 WO2020052436A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811061118.X 2018-09-12
CN201811061118.XA CN110895662A (zh) 2018-09-12 2018-09-12 车辆超载报警方法、装置、电子设备及存储介质

Publications (1)

Publication Number Publication Date
WO2020052436A1 true WO2020052436A1 (zh) 2020-03-19

Family

ID=69777334

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/103030 WO2020052436A1 (zh) 2018-09-12 2019-08-28 车辆超载报警方法、装置、电子设备及存储介质

Country Status (2)

Country Link
CN (1) CN110895662A (zh)
WO (1) WO2020052436A1 (zh)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111444828A (zh) * 2020-03-25 2020-07-24 腾讯科技（深圳）有限公司 一种模型训练的方法、目标检测的方法、装置及存储介质
CN112333155A (zh) * 2020-10-16 2021-02-05 济南浪潮数据技术有限公司 一种异常流量的检测方法、系统、电子设备及存储介质
CN112614102A (zh) * 2020-12-18 2021-04-06 浙江大华技术股份有限公司 一种车辆检测方法、终端及其计算机可读存储介质
CN113688662A (zh) * 2021-07-05 2021-11-23 浙江大华技术股份有限公司 机动车过车警示方法、装置、电子装置和计算机设备
CN114120250A (zh) * 2021-11-30 2022-03-01 北京文安智能技术股份有限公司 一种基于视频的机动车辆违法载人检测方法
CN114131631A (zh) * 2021-12-16 2022-03-04 山东新一代信息产业技术研究院有限公司 一种巡检机器人报警阈值设置方法、装置及介质
CN114169851A (zh) * 2021-11-29 2022-03-11 苏州斯普锐智能系统股份有限公司 一种高速路口快速检测通行系统
CN114758267A (zh) * 2022-03-14 2022-07-15 北京明略软件系统有限公司 一种确定装卸作业效率的方法和装置
CN117315943A (zh) * 2023-11-28 2023-12-29 湖南省交通科学研究院有限公司 一种超限运输违法监测分析与预警方法及系统
CN117830970A (zh) * 2023-12-30 2024-04-05 北京华录高诚科技有限公司 一种基于gps数据的班车违规上下客检测方法及智能设备

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111688716B (zh) * 2020-04-29 2021-11-05 北汽福田汽车股份有限公司 防止车辆超载的方法、装置、存储介质、电子设备及车辆
CN111931644A (zh) * 2020-08-10 2020-11-13 济南博观智能科技有限公司 一种车载人数检测的方法、系统、设备及可读存储介质
CN112874463A (zh) * 2021-03-01 2021-06-01 长安大学 一种儿童被困高温车内的保护与报警系统及方法
CN113077627B (zh) * 2021-03-30 2022-08-05 杭州海康威视系统技术有限公司 检测车辆的超限源头的方法、装置及计算机存储介质
CN117576919B (zh) * 2024-01-19 2024-04-02 北京工业大学 车辆超载识别系统、方法及电子设备

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101197050A (zh) * 2006-12-05 2008-06-11 王德文 公路客车安全辅助监控管理系统——客车黑匣子
CN101593410A (zh) * 2008-05-29 2009-12-02 上海理工大学 长途客车安全监视系统
CN103021059A (zh) * 2012-12-12 2013-04-03 天津大学 一种基于视频监控的公交客流计数方法
US20170286780A1 (en) * 2016-03-18 2017-10-05 Shenzhen University Method and system for calculating passenger crowdedness degree

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103854501A (zh) * 2012-11-28 2014-06-11 西安中科麦特电子技术设备有限公司 一种客车运行信息远程交换系统
CN103280108B (zh) * 2013-05-20 2015-04-22 中国人民解放军国防科学技术大学 基于视觉感知和车联网的客车安全预警系统
DE102014218849A1 (de) * 2014-09-19 2016-03-24 Bayerische Motoren Werke Aktiengesellschaft Verfahren und System zur fahrtbezogenen Fahrzeug-Inventur
CN104282154B (zh) * 2014-10-29 2017-07-11 合肥指南针电子科技有限责任公司 一种车辆超载监控系统及方法
CN105702050A (zh) * 2016-04-22 2016-06-22 安徽皖通科技股份有限公司 一种高速公路超限超载管理控制方法
WO2017206152A1 (zh) * 2016-06-02 2017-12-07 深圳市锐明技术股份有限公司 车辆人员的统计方法及装置
CN106652464A (zh) * 2016-11-10 2017-05-10 深圳市元征软件开发有限公司 车辆超载监管装置、系统及方法
CN108122414A (zh) * 2016-11-30 2018-06-05 杭州海康威视数字技术股份有限公司 公路上客车上下客的检测方法及装置
CN107481351A (zh) * 2017-07-28 2017-12-15 广东兴达顺科技有限公司 一种车辆状态的确定方法、检测设备及车辆设备
CN107657211A (zh) * 2017-08-11 2018-02-02 广州烽火众智数字技术有限公司 一种hov车道的车辆乘客人数检测方法和装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101197050A (zh) * 2006-12-05 2008-06-11 王德文 公路客车安全辅助监控管理系统——客车黑匣子
CN101593410A (zh) * 2008-05-29 2009-12-02 上海理工大学 长途客车安全监视系统
CN103021059A (zh) * 2012-12-12 2013-04-03 天津大学 一种基于视频监控的公交客流计数方法
US20170286780A1 (en) * 2016-03-18 2017-10-05 Shenzhen University Method and system for calculating passenger crowdedness degree

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111444828A (zh) * 2020-03-25 2020-07-24 腾讯科技（深圳）有限公司 一种模型训练的方法、目标检测的方法、装置及存储介质
CN112333155A (zh) * 2020-10-16 2021-02-05 济南浪潮数据技术有限公司 一种异常流量的检测方法、系统、电子设备及存储介质
CN112333155B (zh) * 2020-10-16 2022-07-22 济南浪潮数据技术有限公司 一种异常流量的检测方法、系统、电子设备及存储介质
CN112614102A (zh) * 2020-12-18 2021-04-06 浙江大华技术股份有限公司 一种车辆检测方法、终端及其计算机可读存储介质
CN113688662A (zh) * 2021-07-05 2021-11-23 浙江大华技术股份有限公司 机动车过车警示方法、装置、电子装置和计算机设备
CN114169851A (zh) * 2021-11-29 2022-03-11 苏州斯普锐智能系统股份有限公司 一种高速路口快速检测通行系统
CN114120250A (zh) * 2021-11-30 2022-03-01 北京文安智能技术股份有限公司 一种基于视频的机动车辆违法载人检测方法
CN114120250B (zh) * 2021-11-30 2024-04-05 北京文安智能技术股份有限公司 一种基于视频的机动车辆违法载人检测方法
CN114131631A (zh) * 2021-12-16 2022-03-04 山东新一代信息产业技术研究院有限公司 一种巡检机器人报警阈值设置方法、装置及介质
CN114131631B (zh) * 2021-12-16 2024-02-02 山东新一代信息产业技术研究院有限公司 一种巡检机器人报警阈值设置方法、装置及介质
CN114758267A (zh) * 2022-03-14 2022-07-15 北京明略软件系统有限公司 一种确定装卸作业效率的方法和装置
CN117315943A (zh) * 2023-11-28 2023-12-29 湖南省交通科学研究院有限公司 一种超限运输违法监测分析与预警方法及系统
CN117315943B (zh) * 2023-11-28 2024-02-06 湖南省交通科学研究院有限公司 一种超限运输违法监测分析与预警方法及系统
CN117830970A (zh) * 2023-12-30 2024-04-05 北京华录高诚科技有限公司 一种基于gps数据的班车违规上下客检测方法及智能设备

Also Published As

Publication number Publication date
CN110895662A (zh) 2020-03-20

Similar Documents

Publication Publication Date Title
WO2020052436A1 (zh) 车辆超载报警方法、装置、电子设备及存储介质
US11840239B2 (en) Multiple exposure event determination
US11024165B2 (en) Driver behavior monitoring
US20240290201A1 (en) Driver behavior monitoring
WO2018103313A1 (zh) 发生交通事故的风险预测方法、装置及系统
WO2019223655A1 (zh) 非机动车载人的检测
US20020072847A1 (en) Vision-based method and apparatus for monitoring vehicular traffic events
CN110866427A (zh) 一种车辆行为检测方法及装置
CN111629181B (zh) 消防生命通道监控系统及方法
CN105788269A (zh) 基于无人机的异常交通识别方法
US11862022B2 (en) Methods for characterizing a vehicle collision
US11884285B2 (en) Systems for characterizing a vehicle collision
CN110901385B (zh) 一种基于驾驶员疲劳状态的主动限速方法
US11941986B2 (en) Methods for characterizing a low-impact vehicle collision using high-rate acceleration data
Ravish et al. Intelligent traffic violation detection
CN110909641A (zh) 一种检测摩托车超载的方法、装置及系统
Wang et al. Vision-based highway traffic accident detection
CN112800841B (zh) 一种行人计数方法、装置、系统及计算机可读存储介质
CN113850995A (zh) 一种基于隧道雷视数据融合的事件检测方法、装置及系统
CN114141022A (zh) 应急车道占用行为检测方法、装置、电子设备及存储介质
CN110634120B (zh) 一种车辆损伤判别方法及装置
Al Jarouf et al. A hybrid method to detect and verify vehicle crash with haar-like features and svm over the web
CN114282623A (zh) 一种基于客运车辆违规上下客的分析方法
CN108932453B (zh) 一种车辆备胎检测方法及装置
Maaloul Video-based algorithms for accident detections

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19860507

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19860507

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 19860507

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 15/10/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 19860507

Country of ref document: EP

Kind code of ref document: A1