CN109784254A - A kind of method, apparatus and electronic equipment of rule-breaking vehicle event detection - Google Patents
A kind of method, apparatus and electronic equipment of rule-breaking vehicle event detection Download PDFInfo
- Publication number
- CN109784254A CN109784254A CN201910011455.6A CN201910011455A CN109784254A CN 109784254 A CN109784254 A CN 109784254A CN 201910011455 A CN201910011455 A CN 201910011455A CN 109784254 A CN109784254 A CN 109784254A
- Authority
- CN
- China
- Prior art keywords
- car
- traffic events
- vehicle
- motion profile
- rule
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Landscapes
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The present embodiments relate to network technique fields, disclose the method, apparatus and electronic equipment of a kind of rule-breaking vehicle event detection.The method of rule-breaking vehicle event detection in the present invention, comprising: acquire the image data of all vehicles in predeterminable area in real time, described image data include the image at least 2 frame regions;According to image data, the motion profile of each car in predeterminable area is determined;Model is determined according to the motion profile of each car and traffic events, determines the current affiliated traffic events of each car, wherein traffic events determine corresponding relationship of the model between vehicle movement track and the affiliated traffic events of vehicle;The traffic events according to belonging to each car, output instruction vehicle belong to the warning information of violation traffic events.Present embodiment allows to accurately carry out the vehicle in various scenes the detection of violation event, reduces the frequency of rule-breaking vehicle event.
Description
Technical field
The present embodiments relate to network technique field, in particular to a kind of method, apparatus of rule-breaking vehicle event detection
And electronic equipment.
Background technique
With social economy, the continuous development of science and technology, living standard is continuously improved, and urban automobile quantity increases rapidly
Long, urban traffic environment runs down, and road safety issues have been to be concerned by more and more people.In road safety, vehicle road
The behaviors such as road is stopped, vehicle drives in the wrong direction belong to high frequency traffic violation.Once the behaviors pole such as road vehicle parking, retrograde occurs
Easily cause congestion in road even traffic accident.Therefore, stopped using the realization of existing road video monitoring system to road vehicle,
The real time automatic detection of retrograde equal behaviors determines, has a very important significance in actual production and life.
At least there are the following problems in the prior art for inventor's discovery: the different violation events occurred at present for vehicle need
To be judged using different detection modes, for example, can be true using light stream detection mode to the detection of road vehicle parking
Movement velocity and the direction for determining vehicle are compared according to Velicle motion velocity with the movement velocity of normally travel vehicle, according to
Comparison result judges whether vehicle occurs the violation event of road parking;It and to the detection of vehicle driving in reverse is transported according to vehicle
The whether consistent testing result in direction in dynamic direction and proper motion vehicle, determines the violation testing result of vehicle.As it can be seen that being directed to
The detection of the different violation events of vehicle needs in different ways, using inconvenience, be unfavorable for violation event to vehicle into
Row detection;Simultaneously;Above-mentioned detection method, accuracy is low, cannot play the role of detection to the vehicle travelled in road, lose inspection
The meaning of survey.
Summary of the invention
The method, apparatus for being designed to provide a kind of rule-breaking vehicle event detection and electronics of embodiment of the present invention are set
It is standby, allow to accurately carry out the vehicle in various scenes the detection of violation event, reduces the generation of rule-breaking vehicle event
Number.
In order to solve the above technical problems, embodiments of the present invention provide a kind of method of rule-breaking vehicle event detection,
It include: the image data of all vehicles in real-time acquisition predeterminable area, image data includes the figure in region described at least 2 frames
Picture;According to image data, the motion profile of each car in predeterminable area is determined;According to the motion profile of each car and traffic
Event determines model, determines the current affiliated traffic events of each car, wherein traffic events determine that model is vehicle movement track
With the corresponding relationship between the affiliated traffic events of vehicle;The traffic events according to belonging to each car, output instruction vehicle belong to separated
Advise the warning information of traffic events.
Embodiments of the present invention additionally provide a kind of device of rule-breaking vehicle event detection, comprising: acquisition module, first
Determining module, the second determining module and message output module;Acquisition module for acquiring all vehicles in predeterminable area in real time
Image data, image data include the image at least 2 frame regions;First determining module is used to be determined according to image data
The motion profile of each car in predeterminable area;Second determining module is used for true according to the motion profile and traffic events of each car
Cover half type determines the current affiliated traffic events of each car, wherein traffic events determine that model is vehicle movement track and vehicle
Corresponding relationship between affiliated traffic events;Message output module is used for the traffic events according to belonging to each car, output instruction
Vehicle belongs to the warning information of violation traffic events.
Embodiments of the present invention additionally provide a kind of electronic equipment, comprising: at least one processor;And at least
The memory of one processor communication connection;Wherein, memory is stored with the instruction that can be executed by least one described processor,
Instruction is executed by least one processor, so that at least one processor is able to carry out the side of above-mentioned rule-breaking vehicle event detection
Method.
Embodiment of the present invention in terms of existing technologies, passes through the picture number of all vehicles in acquisition predeterminable area
According to determining the motion profile of each car in the predeterminable area according to the frame image in image data;According to the fortune of each car
Dynamic rail mark and traffic events determine model, that is, can determine that traffic events belonging to current vehicle, due to without according to different
Detection demand vehicle is arranged different detection methods, improves the model for being applicable in scene that violation event detection is carried out to vehicle
It encloses, and only needs detection that can once accurately determine out traffic events belonging to current vehicle, accelerate the detection speed to vehicle
Degree;Since the traffic events of vehicle are related to state of motion of vehicle, pass through the motion profile and traffic events of each car of acquisition
It determines that model determines the affiliated traffic events of vehicle jointly, the accuracy to traffic incidents detection can be improved, effectively subtracted
The generation of few road violation event.
In addition, before determining the current affiliated traffic events of each car, the method for rule-breaking vehicle event detection further include:
Obtain predeterminable area vehicle in the first preset time period history image data, and obtain the first preset time period in each
Traffic events belonging to vehicle;According to history image data, each car going through in predeterminable area in the first preset time period is determined
History motion profile;According to traffic events belonging to each car in historical movement track and the first preset time period, traffic is constructed
Event determines model.The basis for determining model as building traffic events by a large amount of historical datas, so that the traffic thing determined
Part determines that model is more accurate.
In addition, traffic events determine that model is constructed using deep neural network mode, wherein traffic events determine model
Input layer includes n node, and each node corresponds to a sampled point of each car motion profile, and output layer includes m output section
Point, each output node correspond to a kind of traffic events, and n and m are the integer greater than 1.Friendship is constructed using deep neural network mode
Interpreter's part determines model, so that the traffic events determine that model more combines reality, accuracy is high;The sampling number of motion profile
It is corresponding with number of nodes, so as to accurately construct the deep neural network model.
In addition, determining model according to the motion profile of each car and traffic events, the current affiliated friendship of each car is determined
Interpreter's part, specifically includes: obtaining the interstitial content that traffic events determine mode input layer;According to the interstitial content of input layer, divide
The other motion profile to each car pre-processes;The motion profile of pretreated each car is determined into mould as traffic events
The input of type determines the current traffic events of each car.Fortune of the input layer number to each car of model is determined according to traffic
Dynamic rail mark is pre-processed, so that the motion profile of each car meets the input that traffic events determine model.
In addition, pre-processing, specifically including to the motion profile of each car respectively according to the interstitial content of input layer:
Following pretreatment is carried out to the motion profile of each car: judging whether the number for the sampled point for including in motion profile is equal to node
Number, however, it is determined that the sampled point number for including in motion profile is not equal to interstitial content, then adjusts the sampled point in motion profile
Number, so that sampled point number is equal to interstitial content in motion profile adjusted.By adjust sampling number, with ensure no matter
Actual sampled point is that mostly less, can determine corresponding traffic events according to the sampled point in motion profile.
In addition, after obtaining traffic events and determining the interstitial content of mode input layer, the side of rule-breaking vehicle event detection
Method further include: the motion profile of each car is normalized.By normalization operation, convenient for subsequent quickly to movement rail
Mark is handled, simultaneously because motion profile is mapped between 0 to 1, is decreased because of two sampled point distances in motion profile
It differs larger and leads to the big error occurred.
In addition, determining the motion profile of each car in predeterminable area according to image data, specifically including: detecting every frame
Position of each car in predeterminable area in image;It is handled as follows for each car in image: in the image of successive frame
Identify vehicle, and using the position of vehicle in every frame image as the sampled point in vehicle movement track;According to vehicle in region
The sampled point of motion profile determines the motion profile of vehicle.It can relatively accurately identify the vehicle in every frame image, and due to
The continuity of image, can relatively accurately track the position of each car in the picture, and then can be very fast and accurately determine
Each car corresponds to motion profile in predeterminable area image out, shortens the time for determining motion profile.
In addition, determining the input of model using the motion profile of pretreated each car as traffic events, each is determined
Traffic events belonging to vehicle is current, specifically include: each car are handled as follows: obtaining the corresponding every kind of traffic of current vehicle
The confidence level of event;According to confidence level, vehicle currently affiliated traffic events are determined.It more can objectively be determined according to confidence level
The affiliated traffic events of vehicle out.
Detailed description of the invention
One or more embodiments are illustrated by the picture in corresponding attached drawing, these exemplary theorys
The bright restriction not constituted to embodiment, the element in attached drawing with same reference numbers label are expressed as similar element, remove
Non- to have special statement, composition does not limit the figure in attached drawing.
Fig. 1 is a kind of specific stream of the method for the rule-breaking vehicle event detection provided in first embodiment according to the present invention
Journey schematic diagram;
Fig. 2 is sampled point schematic diagram in the motion profile of vehicle in first embodiment according to the present invention;
Fig. 3 is the specific of the traffic events belonging to the determination each car that provides in second embodiment according to the present invention is current
Flow diagram;
Fig. 4 is a kind of specific knot of the device of the rule-breaking vehicle event detection provided in third embodiment according to the present invention
Structure is intended to;
Fig. 5 is that the specific structure of a kind of electronic equipment provided in the 4th embodiment according to the present invention is intended to.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with attached drawing to the present invention
Each embodiment be explained in detail.However, it will be understood by those skilled in the art that in each embodiment party of the present invention
In formula, in order to make the reader understand this application better, many technical details are proposed.But even if without these technical details
And various changes and modifications based on the following respective embodiments, the application technical solution claimed also may be implemented.
The first embodiment of the present invention is related to a kind of methods of rule-breaking vehicle event detection.The rule-breaking vehicle event detection
Method be applied on terminal or electronic equipment, and the terminal or electronic equipment are mounted on and need to detect rule-breaking vehicle event
The positions such as road, parking lot, to be measured in real time to the vehicle in predeterminable area.The method of the rule-breaking vehicle event detection
Specific process is as shown in Figure 1.
Step 101: acquiring the image data of all vehicles in predeterminable area in real time, image data includes at least 2 frame regions
Interior image.
Specifically, the equipment that the acquisition image data such as camera, infrared monitoring equipment can be used, acquisition is default in real time
The image data of all vehicles in region.Predeterminable area can be determined according to the scene of rule-breaking vehicle event detection, detection
Scene can be parking lot, high speed crossing or road etc..Wherein, predeterminable area can be set to the acquisition equipment to current
The acquisition maximum acquisition range of scene, for example, detection scene is crossing position if acquisition equipment is camera, then can incite somebody to action
The range for the image that the camera is shot at current crossing is as predeterminable area.Since acquisition equipment is usually fixed at detection
Scene, so that the predeterminable area immobilizes.
Image data includes at least the image in the 2 frames predeterminable area, for example, acquiring preset areas in real time according to camera
Image in domain, that is, acquiring is continuous frame image.
It should be noted that needing to construct current preset region in advance before determining the current traffic events of each car
Traffic events determine model.Determine that the traffic events determine that the mode of model can be in the following way.
In one concrete implementation, the history image data of predeterminable area vehicle in the first preset time period are obtained, with
And obtain traffic events belonging to each car in the first preset time period;According to history image data, the first preset time is determined
Historical movement track of each car in predeterminable area in section;According in historical movement track and the first preset time period each
Traffic events belonging to vehicle, building traffic events determine model.
Specifically, need to detect vehicle whether there are many scenes of violation, predeterminable area is different, thus can be according to every
A predeterminable area determines that corresponding traffic events determine model.The traffic events determine that model can use neural network
The mode of study is constructed.It can be learnt by a large amount of historical data, so that constructing the traffic events determines mould
Type.First preset time period can be chosen from historical time section, for the accuracy of study, can choose and current time phase
The adjacent period.For example, if currently be on January 1st, 2018, can with the first preset time period can on January 1st, 2017 extremely
Period between on December 31st, 2017.The history image data of predeterminable area vehicle in the first preset time period are obtained,
It identifies the vehicle in every frame image, and according to continuous frame image, determines fortune of each car in the predeterminable area image
Dynamic rail mark, using the historical movement track of each car as the input data of traffic events cover half type, each car that will acquire
Affiliated traffic events as the traffic events cover half type output as a result, using machine learning algorithm or neural network algorithm,
The traffic events that can be constructed in the predeterminable area determine model.It is handed over it is understood that present embodiment does not limit building
Interpreter's part determines the algorithm of model.
Step 102: according to image data, determining the motion profile of each car in predeterminable area.
In one concrete implementation, position of each car in predeterminable area in every frame image is detected;For every in image
Vehicle is handled as follows: identifying vehicle in the image of successive frame, and transports the position of vehicle in every frame image as vehicle
Sampled point in dynamic rail mark;According to the sampled point of the motion profile of vehicle in region, the motion profile of vehicle is determined.
Specifically, it identifies the vehicle in the every frame image for including in image data, and detects each car in the image
In position, since predeterminable area is fixed, and it is also fixed for acquiring equipment, so that the background in every frame image is (fixed
Motionless building) it is all the same, by the location information of each car in different frame image, that is, it can determine that the figure in the predeterminable area
The motion profile of each car as in, wherein the position of vehicle can be used as a sampled point in motion profile in every frame image.
Vehicle in identification image can carry out the identification of vehicle by the way of feature extraction, and determine that the vehicle is being worked as
Region in preceding image can choose the center of each car region for the ease of obtaining the motion profile of the vehicle
Position of the point as the vehicle in the images can determine that the motion profile of the vehicle by multiple image, for example, such as
Shown in Fig. 2, frame A, frame B and frame C respectively represent region of the vehicle y in 3 frame consecutive images, then obtain the central point of frame A
The line of the midpoint s3 of the central point s2 and frame C of s1, frame B, 3 points of s1, s2 and the s3 are the movement rail that can be identified as vehicle y
Mark.
The identification of vehicle can also pass through yolo (You Only Look Once, referred to as " yolo ") real-time target detection method
Or object detection algorithms ssd (single shot multibox detector, referred to as " ssd ") is realized, can be passed through later
Hungary Algorithm tracking belongs to position of the identical vehicle in predeterminable area image, so as to get the movement of each car
Track, wherein the specific calculating process of yolo or ssd and Hungary Algorithm will be repeated no longer in present embodiment.
Step 103: model is determined according to the motion profile of each car and traffic events, determine each car it is current belonging to
Traffic events, wherein traffic events determine corresponding relationship of the model between vehicle movement track and the affiliated traffic events of vehicle.
In one concrete implementation, the motion profile of each car is handled as follows: using the motion profile of vehicle as
Traffic events determine the input of model, obtain the confidence level that vehicle corresponds to every kind of traffic events;According to confidence level, determine that vehicle is worked as
Traffic events belonging to preceding.
Specifically, the input parameter request that model is determined according to traffic events, should by the motion profile input of each car
Traffic events determine in model that the traffic events model calculates the corresponding every kind of friendship of this vehicle according to the motion profile of input
The confidence level of interpreter's part can determine the affiliated traffic events of current vehicle with the confidence level size of every kind of traffic events.Such as: if
The traffic events determine that corresponding output result includes in model, and illegal parking, vehicle drives in the wrong direction and the normal vehicle operation three
Kind traffic events obtain above-mentioned three kinds of traffic events pair after the fortune track of vehicle A to be inputted to the traffic events and determines model
The confidence level answered, respectively parking confidence level is 0.1, the confidence level that vehicle is retrograde is 0.8, and the confidence of the normal vehicle operation
Degree is 0.1, therefore, can determine that the affiliated traffic events of vehicle A are retrograde according to the size of the vehicle confidence level.When
So, it is to be understood that traffic events determine in model it is corresponding belonging to traffic events type can according to actual needs into
Row setting, is not limited to three kinds of traffic events cited in present embodiment.
Step 104: the traffic events according to belonging to each car, output instruction vehicle belong to the alarm letter of violation traffic events
Breath.
Specifically, it is determining that the affiliated traffic events of vehicle belong to the traffic events of violation, then can export and indicate the vehicle
Belong to the warning information of violation traffic events, the mode of output can be voice notification, can also be and is uploaded to network of relation
Platform (such as network of communication lines platform).
Embodiment of the present invention in terms of existing technologies, passes through the picture number of all vehicles in acquisition predeterminable area
According to determining the motion profile of each car in the predeterminable area according to the frame image in image data;According to the fortune of each car
Dynamic rail mark and traffic events determine model, that is, can determine that traffic events belonging to current vehicle, due to without according to different
Detection demand vehicle is arranged different detection methods, improves the model for being applicable in scene that violation event detection is carried out to vehicle
It encloses, and only needs detection that can once accurately determine out traffic events belonging to current vehicle, accelerate the detection speed to vehicle
Degree;Since the traffic events of vehicle are related to state of motion of vehicle, pass through the motion profile and traffic events of each car of acquisition
It determines that model determines the affiliated traffic events of vehicle jointly, the accuracy to traffic incidents detection can be improved, effectively subtracted
The generation of few road violation event.
Second embodiment of the present invention is related to a kind of method of rule-breaking vehicle event detection.Second embodiment is to
Step 103 in one embodiment: determining model according to the motion profile of each car and traffic events, determines the current institute of each car
The specific refinement of the traffic events of category.Wherein it is determined that the detailed process of the traffic events belonging to each car is current is as shown in Figure 3.
Step 2031: obtaining the interstitial content that traffic events determine mode input layer.
In one concrete implementation, traffic events determine that model is constructed using deep neural network mode, wherein traffic thing
Part determines that the input layer of model includes n node, and each node corresponds to a sampled point of each car motion profile, output layer packet
Containing m output node, each output node corresponds to a kind of traffic events, and n and m are the integer greater than 1.
Specifically, deep neural network model (Deep Neural Network, referred to as " DNN " model) includes input
Three layer, hidden layer and output layer parts.Each node of input layer corresponds to a sampled point of a vehicle motion profile, example
Such as, input layer includes 4 nodes, then 4 sampled points of 4 nodes corresponding to the motion profile of vehicle A.And hidden layer can be set
Single-layer or multi-layer is set, and every layer also can be set multiple nodes, the level of hidden layer is more, and calculating is more complicated, can be according to reality
The number of plies for needing to be arranged hidden layer and every layer of number of nodes.Output layer includes m output node, and each output node is corresponding a kind of
Traffic events, such as normally travel, curb parking and reverse driving traffic events.
Since traffic events determine that model is built in advance, the traffic events can be directly obtained and determine that model is defeated
Enter the interstitial content of layer.
Step 2032: according to the interstitial content of input layer, the motion profile of each car being pre-processed respectively.
In one concrete implementation, following pretreatment is carried out to the motion profile of each car: judge include in motion profile
The number of sampled point whether be equal to interstitial content, however, it is determined that the sampled point number for including in motion profile is not equal to number of nodes
Mesh then adjusts the sampled point number in motion profile, so that sampled point number is equal to interstitial content in motion profile adjusted.
It is illustrated by taking the wherein motion profile of a vehicle as an example below.
Specifically, sampled point number included in the motion profile of this vehicle is obtained first, judges the sampling number
Whether mesh is equal to the number of nodes in input layer, if equal, then directly carries out each sampling in this vehicle in motion profile
Point determines that the input of model then adjusts the sampled point number of motion profile in this vehicle if unequal as traffic events, with
Keep the sampled point number equal with the interstitial content of input layer.When the sampled point number in motion profile is greater than the node of input layer
Number, then between the maximum value and minimum value of the random erasure motion profile sample point coordinate data in any coordinate data,
Until the quantity of the motion profile coordinate data of the vehicle is equal to n.When the sampled point number in motion profile is less than input layer
Interstitial content, then between the maximum value and minimum value of the vehicle movement track sample point coordinate data stochastic simulation generate it is new
Sampled point, until the quantity of the motion profile coordinate data of the vehicle is equal to n.
It should be noted that after obtaining traffic events and determining the interstitial content of mode input layer, and respectively to every
Before the motion profile of vehicle is pre-processed, the motion profile of each car can also be normalized.Pass through normalizing
Change operation, quickly motion profile is handled convenient for subsequent, simultaneously because motion profile is mapped between 0 to 1, is decreased
Lead to the big error occurred due to two sampled point distances differ larger in motion profile.
Step 2033: determining the input of model using the motion profile of pretreated each car as traffic events, determine
The current traffic events of each car.
In one specific embodiment, the confidence level of the corresponding every kind of traffic events of current vehicle is obtained;According to confidence
Degree determines vehicle currently affiliated traffic events.Traffic events are determined according to confidence level in the process and first embodiment
Process is roughly the same, will not be described in great detail herein.
The method of the rule-breaking vehicle event detection provided in present embodiment constructs traffic using deep neural network mode
Event determines model, so that the traffic events determine that model more combines reality, accuracy is high;The sampling number of motion profile with
Number of nodes is corresponding, so as to construct deep neural network model building;It is no matter real to ensure and by adjusting sampling number
The sampled point on border is that mostly less, can determine corresponding traffic events according to the sampled point in motion profile.
The step of various methods divide above, be intended merely to describe it is clear, when realization can be merged into a step or
Certain steps are split, multiple steps are decomposed into, as long as including identical logical relation, all in the protection scope of this patent
It is interior;To adding inessential modification in algorithm or in process or introducing inessential design, but its algorithm is not changed
Core design with process is all in the protection scope of the patent.
Third embodiment of the invention is related to a kind of device of rule-breaking vehicle event detection, the rule-breaking vehicle event detection
Device 30 includes: acquisition module 301, the first determining module 302, the second determining module 303 and message output module 304, specifically
Structure is as shown in Figure 4.
Acquisition module 301 includes at least 2 for acquiring the image data of all vehicles in predeterminable area, image data in real time
Image in frame region;First determining module 302 is used to determine the movement rail of each car in predeterminable area according to image data
Mark;Second determining module 303 determines that each car is current for determining model according to the motion profile and traffic events of each car
Affiliated traffic events, wherein traffic events determine pair of the model between vehicle movement track and the affiliated traffic events of vehicle
It should be related to;Message output module 304 is used for the traffic events according to belonging to each car, and output instruction vehicle belongs to violation traffic thing
The warning information of part.
It is not difficult to find that present embodiment is Installation practice corresponding with first embodiment, present embodiment can be with
First embodiment is worked in coordination implementation.The relevant technical details mentioned in first embodiment still have in the present embodiment
Effect, in order to reduce repetition, which is not described herein again.Correspondingly, the relevant technical details mentioned in present embodiment are also applicable in
In first embodiment.
It is noted that each module involved in present embodiment is logic module, and in practical applications, one
A logic unit can be a physical unit, be also possible to a part of a physical unit, can also be with multiple physics lists
The combination of member is realized.In addition, in order to protrude innovative part of the invention, it will not be with solution institute of the present invention in present embodiment
The technical issues of proposition, the less close unit of relationship introduced, but this does not indicate that there is no other single in present embodiment
Member.
Four embodiment of the invention is related to a kind of electronic equipment, which includes: at least one processor
401;And the memory 402 with the communication connection of at least one processor 401;Wherein, be stored with can be by least for memory 402
The instruction that one processor 401 executes, instruction is executed by least one processor 401, so that at least one processor 401 can
The method for executing first embodiment or the rule-breaking vehicle event detection in second embodiment.The specific structure of the electronic equipment
As shown in Figure 5.
Wherein, memory 402 is connected with processor 401 using bus mode, and bus may include any number of interconnection
Bus and bridge, bus the various circuits of one or more processors 401 and memory 402 are linked together.Bus may be used also
To link together various other circuits of such as peripheral equipment, voltage-stablizer and management circuit or the like, these are all
It is known in the art, therefore, it will not be further described herein.Bus interface provides between bus and transceiver
Interface.Transceiver can be an element, be also possible to multiple element, such as multiple receivers and transmitter, provide for
The unit communicated on transmission medium with various other devices.The data handled through processor are carried out on the radio medium by antenna
Transmission, further, antenna also receives data and transfers data to processor.
Processor 401 is responsible for management bus and common processing, can also provide various functions, including timing, periphery connects
Mouthful, voltage adjusting, power management and other control functions.And memory can be used for storage processor when executing operation
Used data.
It will be appreciated by those skilled in the art that implementing the method for the above embodiments is that can pass through
Program is completed to instruct relevant hardware, which is stored in a storage medium, including some instructions are used so that one
A equipment (can be single-chip microcontroller, chip etc.) or processor (processor) execute each embodiment the method for the application
All or part of the steps.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only
Memory), random access memory (RAM, Random Access Memory), magnetic or disk etc. are various can store journey
The medium of sequence code.
It will be understood by those skilled in the art that the respective embodiments described above are to realize specific embodiments of the present invention,
And in practical applications, can to it, various changes can be made in the form and details, without departing from the spirit and scope of the present invention.
Claims (10)
1. a kind of method of rule-breaking vehicle event detection characterized by comprising
The image data of all vehicles in predeterminable area is acquired in real time, and described image data include in region described at least 2 frames
Image;
According to described image data, the motion profile of each car in the predeterminable area is determined;
Model is determined according to the motion profile of each car and traffic events, determines the current affiliated traffic events of each car,
In, the traffic events determine corresponding relationship of the model between vehicle movement track and the affiliated traffic events of vehicle;
The traffic events according to belonging to each car, output instruction vehicle belong to the warning information of violation traffic events.
2. the method for rule-breaking vehicle event detection according to claim 1, which is characterized in that determining the current institute of each car
Before the traffic events of category, the method for the rule-breaking vehicle event detection further include:
The history image data of predeterminable area vehicle in the first preset time period are obtained, and obtains described first and presets
Traffic events belonging to each car in period;
According to the history image data, historical movement rail of each car in predeterminable area in the first preset time period is determined
Mark;
According to traffic events belonging to each car in the historical movement track and first preset time period, described in building
Traffic events determine model.
3. the method for rule-breaking vehicle event detection according to claim 1 or 2, which is characterized in that the traffic events are true
Cover half type is constructed using deep neural network mode, wherein and the traffic events determine that the input layer of model includes n node,
Each node corresponds to a sampled point of each car motion profile, and the output layer includes m output node, each output
Node corresponds to a kind of traffic events, and n and m are the integer greater than 1.
4. the method for rule-breaking vehicle event detection according to claim 3, which is characterized in that according to the movement rail of each car
Mark and traffic events determine model, determine the current affiliated traffic events of each car, specifically include:
Obtain the interstitial content that the traffic events determine mode input layer;
According to the interstitial content of the input layer, the motion profile of each car is pre-processed respectively;
The input that model is determined using the motion profile of pretreated each car as the traffic events determines that each car is current
The traffic events.
5. the method for rule-breaking vehicle event detection according to claim 4, which is characterized in that according to the section of the input layer
Point number, respectively pre-processes the motion profile of each car, specifically includes:
Following pretreatment is carried out to the motion profile of each car:
Judge whether the number for the sampled point for including in the motion profile is equal to the interstitial content, however, it is determined that the movement rail
The sampled point number for including in mark is not equal to the interstitial content, then adjusts the sampled point number in the motion profile, so that
Sampled point number is equal to the interstitial content in the motion profile adjusted.
6. the method for rule-breaking vehicle event detection according to claim 5, which is characterized in that obtaining the traffic events
After the interstitial content for determining mode input layer, and before being pre-processed respectively to the motion profile of each car, the vehicle
The method of violation event detection further include:
The motion profile of each car is normalized.
7. the method for rule-breaking vehicle event detection according to claim 1, which is characterized in that according to described image data,
The motion profile for determining each car in the predeterminable area, specifically includes:
Detect position of each car in the predeterminable area in every frame image;
It is handled as follows for each car in image:
The vehicle is identified in the image of successive frame, and using the position of vehicle described in every frame image as the vehicle movement
Sampled point in track;
According to the sampled point of the motion profile of the vehicle in the region, the motion profile of the vehicle is determined.
8. the method for rule-breaking vehicle event detection according to any one of claim 4 to 6, which is characterized in that will locate in advance
The motion profile of each car after reason determines the input of model as the traffic events, determines the current affiliated traffic of each car
Event specifically includes:
Each car is handled as follows:
Obtain the confidence level of the corresponding every kind of traffic events of current vehicle;
According to the confidence level, the vehicle currently affiliated traffic events are determined.
9. a kind of device of rule-breaking vehicle event detection characterized by comprising acquisition module, the first determining module, second are really
Cover half block and message output module;
The acquisition module includes at least for acquiring the image data of all vehicles in predeterminable area, described image data in real time
Image in region described in 2 frames;
First determining module is used to determine the movement rail of each car in the predeterminable area according to described image data
Mark;
Second determining module determines that each car is worked as determining model according to the motion profile and traffic events of each car
Traffic events belonging to preceding, wherein the traffic events determine model be vehicle movement track and the affiliated traffic events of vehicle it
Between corresponding relationship;
The message output module is used for the traffic events according to belonging to each car, and output instruction vehicle belongs to violation traffic events
Warning information.
10. a kind of electronic equipment characterized by comprising
At least one processor;And
The memory being connect at least one described processor communication;Wherein,
The memory is stored with the instruction that can be executed by least one described processor, and described instruction is by described at least one
It manages device to execute, so that at least one described processor is able to carry out rule-breaking vehicle event inspection a method as claimed in any one of claims 1-8
The method of survey.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910011455.6A CN109784254B (en) | 2019-01-07 | 2019-01-07 | Vehicle violation event detection method and device and electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910011455.6A CN109784254B (en) | 2019-01-07 | 2019-01-07 | Vehicle violation event detection method and device and electronic equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109784254A true CN109784254A (en) | 2019-05-21 |
CN109784254B CN109784254B (en) | 2021-06-25 |
Family
ID=66500060
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910011455.6A Active CN109784254B (en) | 2019-01-07 | 2019-01-07 | Vehicle violation event detection method and device and electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109784254B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110634302A (en) * | 2019-11-14 | 2019-12-31 | 广东优世联合控股集团股份有限公司 | Parking lot retrograde vehicle detection system based on BIM |
CN111091098A (en) * | 2019-12-20 | 2020-05-01 | 浙江大华技术股份有限公司 | Training method and detection method of detection model and related device |
CN111209880A (en) * | 2020-01-13 | 2020-05-29 | 南京新一代人工智能研究院有限公司 | Vehicle behavior identification method and device |
CN112419722A (en) * | 2020-11-18 | 2021-02-26 | 百度(中国)有限公司 | Traffic abnormal event detection method, traffic control method, device and medium |
CN112435470A (en) * | 2020-11-11 | 2021-03-02 | 宁波职业技术学院 | Traffic incident video detection system |
CN112712696A (en) * | 2020-12-30 | 2021-04-27 | 北京嘀嘀无限科技发展有限公司 | Method and device for determining road section with illegal parking |
CN114019947A (en) * | 2020-07-15 | 2022-02-08 | 广州汽车集团股份有限公司 | Driving control method and system for vehicle at intersection and computer readable storage medium |
WO2022105243A1 (en) * | 2020-11-23 | 2022-05-27 | 北京旷视科技有限公司 | Event detection method, apparatus, electronic device, and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102289660A (en) * | 2011-07-26 | 2011-12-21 | 华南理工大学 | Method for detecting illegal driving behavior based on hand gesture tracking |
US20120170802A1 (en) * | 2010-12-30 | 2012-07-05 | Pelco Inc. (Clovis, CA) | Scene activity analysis using statistical and semantic features learnt from object trajectory data |
CN103593430A (en) * | 2013-11-11 | 2014-02-19 | 胡宝清 | Clustering method based on mobile object spatiotemporal information trajectory subsections |
CN104809470A (en) * | 2015-04-23 | 2015-07-29 | 杭州中威电子股份有限公司 | Vehicle converse running detection device and method based on SVM |
CN106846803A (en) * | 2017-02-08 | 2017-06-13 | 广西交通科学研究院有限公司 | Traffic incidents detection device and method based on audio |
CN108345894A (en) * | 2017-01-22 | 2018-07-31 | 北京同方软件股份有限公司 | A kind of traffic incidents detection method based on deep learning and entropy model |
-
2019
- 2019-01-07 CN CN201910011455.6A patent/CN109784254B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20120170802A1 (en) * | 2010-12-30 | 2012-07-05 | Pelco Inc. (Clovis, CA) | Scene activity analysis using statistical and semantic features learnt from object trajectory data |
CN102289660A (en) * | 2011-07-26 | 2011-12-21 | 华南理工大学 | Method for detecting illegal driving behavior based on hand gesture tracking |
CN103593430A (en) * | 2013-11-11 | 2014-02-19 | 胡宝清 | Clustering method based on mobile object spatiotemporal information trajectory subsections |
CN104809470A (en) * | 2015-04-23 | 2015-07-29 | 杭州中威电子股份有限公司 | Vehicle converse running detection device and method based on SVM |
CN108345894A (en) * | 2017-01-22 | 2018-07-31 | 北京同方软件股份有限公司 | A kind of traffic incidents detection method based on deep learning and entropy model |
CN106846803A (en) * | 2017-02-08 | 2017-06-13 | 广西交通科学研究院有限公司 | Traffic incidents detection device and method based on audio |
Non-Patent Citations (2)
Title |
---|
DONGDONG ZHANG ET AL: "traffic peccancy processing system based on image recognition", 《PROCEEDINGS OF THE 2003 IEEE INTERNATIONAL CONFERENCE ON INTELLIGENT TRANSPORTATION SYSTEMS》 * |
王文 等: "基于视频大数据的隧道安全监测***", 《中国交通信息》 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110634302A (en) * | 2019-11-14 | 2019-12-31 | 广东优世联合控股集团股份有限公司 | Parking lot retrograde vehicle detection system based on BIM |
CN111091098A (en) * | 2019-12-20 | 2020-05-01 | 浙江大华技术股份有限公司 | Training method and detection method of detection model and related device |
CN111091098B (en) * | 2019-12-20 | 2023-08-15 | 浙江大华技术股份有限公司 | Training method of detection model, detection method and related device |
CN111209880A (en) * | 2020-01-13 | 2020-05-29 | 南京新一代人工智能研究院有限公司 | Vehicle behavior identification method and device |
CN114019947A (en) * | 2020-07-15 | 2022-02-08 | 广州汽车集团股份有限公司 | Driving control method and system for vehicle at intersection and computer readable storage medium |
CN114019947B (en) * | 2020-07-15 | 2024-03-12 | 广州汽车集团股份有限公司 | Method and system for controlling vehicle to travel at intersection and computer readable storage medium |
CN112435470A (en) * | 2020-11-11 | 2021-03-02 | 宁波职业技术学院 | Traffic incident video detection system |
CN112419722A (en) * | 2020-11-18 | 2021-02-26 | 百度(中国)有限公司 | Traffic abnormal event detection method, traffic control method, device and medium |
WO2022105243A1 (en) * | 2020-11-23 | 2022-05-27 | 北京旷视科技有限公司 | Event detection method, apparatus, electronic device, and storage medium |
CN112712696A (en) * | 2020-12-30 | 2021-04-27 | 北京嘀嘀无限科技发展有限公司 | Method and device for determining road section with illegal parking |
Also Published As
Publication number | Publication date |
---|---|
CN109784254B (en) | 2021-06-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109784254A (en) | A kind of method, apparatus and electronic equipment of rule-breaking vehicle event detection | |
CN110390262B (en) | Video analysis method, device, server and storage medium | |
Chen et al. | A short-term traffic prediction model in the vehicular cyber–physical systems | |
CN103456175B (en) | Accompanying vehicle real-time detection method based on vehicle registration plate recognition and meshing monitoring | |
CN112700470B (en) | Target detection and track extraction method based on traffic video stream | |
US11836985B2 (en) | Identifying suspicious entities using autonomous vehicles | |
CN102254429B (en) | Video identification-based detection method of detection device of violation vehicles | |
CN106971544B (en) | A kind of direct method that vehicle congestion is detected using still image | |
CN107845259A (en) | Public transport operation situation real-time feedback system and public transport real-time running data processing method | |
EP4089659A1 (en) | Map updating method, apparatus and device | |
CN109446926A (en) | A kind of traffic monitoring method and device, electronic equipment and storage medium | |
CN111507235A (en) | Video-based railway perimeter foreign matter intrusion detection method | |
CN112233428B (en) | Traffic flow prediction method, device, storage medium and equipment | |
Parada et al. | Machine learning-based trajectory prediction for vru collision avoidance in v2x environments | |
Minnikhanov et al. | Detection of traffic anomalies for a safety system of smart city | |
Deng | [Retracted] Anomaly Detection of Highway Vehicle Trajectory under the Internet of Things Converged with 5G Technology | |
CN111739293A (en) | Data fusion method and device | |
Hardjono et al. | Virtual Detection Zone in smart phone, with CCTV, and Twitter as part of an Integrated ITS | |
CN109800685A (en) | The determination method and device of object in a kind of video | |
Yao et al. | Embedded technology and algorithm for video-based vehicle queue length detection | |
CN104408942A (en) | Intelligent vehicle speed measuring device and method | |
Mostafi et al. | Real-Time Jaywalking Detection and Notification System using Deep Learning and Multi-Object Tracking | |
ElHakim et al. | Traffisense: A smart integrated visual sensing system for traffic monitoring | |
CN114662792B (en) | Traffic flow prediction method of recurrent neural network based on dynamic diffusion graph convolution | |
Xu et al. | Wi‐CL: Low‐Cost WiFi‐Based Detection System for Nonmotorized Traffic Travel Mode Classification |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |