CN114220074A - Method and system for identifying abnormal behavior state of vehicle - Google Patents
Method and system for identifying abnormal behavior state of vehicle Download PDFInfo
- Publication number
- CN114220074A CN114220074A CN202111515694.9A CN202111515694A CN114220074A CN 114220074 A CN114220074 A CN 114220074A CN 202111515694 A CN202111515694 A CN 202111515694A CN 114220074 A CN114220074 A CN 114220074A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- state
- preset
- new
- area image
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method and a system for identifying abnormal behavior states of a vehicle, which relate to the field of intelligent parking management and comprise the following steps: the method comprises the steps of matching a trainable vehicle state identification model with an existing vehicle detection identification model and fusing identification result information to realize identification and monitoring of abnormal states of the vehicle, and feeding back abnormal vehicle information when the abnormal vehicle is confirmed through comparing a door opening probability and/or a trunk opening probability in a vehicle state identification vector with a preset state judgment threshold value. The method can reduce the difficulty in realizing the algorithm, and can independently deploy the vehicle state recognition model according to the actual demand condition independently of the vehicle detection recognition model, thereby ensuring the high efficiency, stability and controllability of the recognition algorithm of the abnormal behavior state of the whole vehicle; meanwhile, the multi-classification vehicle state identification model adopted in the invention does not need positioning information, so that the marking requirement of a target detailed region is reduced, and the difficulty of data marking is greatly reduced.
Description
Technical Field
The invention relates to the field of intelligent parking management, in particular to a method and a system for identifying abnormal behavior states of a vehicle.
Background
Under a high-order video-based dynamic and static traffic monitoring scene, multi-dimensional information of vehicle states in a visual area is captured, correlation integration of various information of vehicles and surrounding related targets is carried out, and necessary data support can be provided for abnormal behavior states of the vehicles under a complex scene with open monitoring. The main vehicle behavior states include, in addition to driving and parking, the opening and closing of the doors and trunk: the opening of the trunk is also highly related to the monitoring action of the driver for avoiding the camera. Therefore, the method has an important role in monitoring the states of the vehicle door and the trunk and screening illegal behaviors such as getting off passengers on illegal road sections and intentionally avoiding electronic information acquisition by a camera.
At present, when the states of a vehicle door and a trunk are identified, a deep learning network based on target detection is generally adopted, model training data corresponding to the deep learning network not only needs to judge whether the vehicle door and the trunk are opened, but also needs to mark position information of the vehicle door and the trunk in advance, so that excessive additionally-increased marking information greatly increases application cost; and the open state of the car door trunk is characterized by a continuous process and has morphological similarity, and under the actual multi-view sampling state, the state confusion can be caused to a great extent by utilizing the information of too local areas for classification and identification, so that the identification accuracy is low.
Disclosure of Invention
In order to solve the technical problems, the invention provides a method and a system for identifying an abnormal behavior state of a vehicle, which can solve the problems of high cost and low accuracy of illegal parking management of the existing vehicle based on a directional bounding box of the vehicle.
To achieve the above object, in one aspect, the present invention provides a method for identifying an abnormal behavior state of a vehicle, the method including:
acquiring a monitoring area image acquired in real time, and acquiring a position coordinate of a vehicle detection frame from the monitoring area image according to a preset vehicle detection identification model;
acquiring a cutting frame coordinate corresponding to a vehicle state identification frame according to the position coordinate of the vehicle detection frame, and processing the monitored area image according to the cutting frame coordinate to obtain a monitored area image sub-picture set;
processing the monitoring area image sub-map set according to a preset vehicle state identification model to obtain a vehicle state identification vector corresponding to the monitoring area image, wherein the preset vehicle state identification model is used for identifying the vehicle door opening probability, the trunk opening probability and the probability of people around the vehicle, and the vehicle state identification vector comprises the vehicle door opening probability and/or the trunk opening probability;
and comparing the door opening probability and/or trunk opening probability in the vehicle state identification vector with a preset state judgment threshold value to confirm whether the vehicle has abnormal state behaviors.
Further, before the step of acquiring the monitoring area image acquired in real time, the method further includes:
acquiring a vehicle detection frame from a monitored area image stored in a database through the preset vehicle detection and identification model, and cutting and storing a local target area of the monitored area image data according to the preset size of the length and width extension of the vehicle detection frame to be used as a training image data set;
marking the training picture data set according to the ternary vectors corresponding to the classification type sequence to obtain model training data, wherein the categories of the ternary vectors corresponding to the classification type sequence comprise vehicle door state information, trunk state information and state information of whether people exist around the vehicle;
and inputting the model training data into a preset classification model to obtain a three-component vector Pv ═ P _ do, P _ to and P _ per as the preset vehicle state identification model, wherein P _ do is the vehicle door opening probability, P _ to is the trunk opening probability, and P _ per is the probability of existence of people near the vehicle.
Further, the step of obtaining the coordinates of the cropping frame corresponding to the vehicle state identification frame according to the coordinates of the position of the vehicle detection frame includes:
and expanding the length and width dimensions of the vehicle detection frame by preset dimensions to obtain the coordinates of the cutting frame corresponding to the vehicle state identification frame.
Further, the step of processing the monitoring area image sub-image set according to a preset vehicle state identification model to obtain a vehicle state identification vector corresponding to the monitoring area image includes:
acquiring vehicle door state information according to a formula P _ do _ new ═ a1 ═ P _ do + a2 × (P _ per), wherein P _ do _ new is a door opening probability considering human presence, a1 ═ exp (P _ do)/(exp (P _ do) + exp (P _ per)), and a2 ═ exp (P _ per)/(exp (P _ do) + exp (P _ per));
vehicle trunk status information is obtained according to the formula P _ to _ new (1-P _ per), wherein P _ to _ new takes into account the trunk opening probability after human presence.
Further, the step of comparing the door opening probability and/or trunk opening probability in the vehicle state identification vector with a preset state judgment threshold value to determine whether the vehicle has abnormal state behavior includes:
obtaining a vehicle state indication vector state [ S _ d, S _ t ] according to a preset state determination threshold thr [ [ thr _ do, thr _ to ] and a vehicle state identification vector Pv _ new [ [ P _ do _ new, P _ to _ new ], wherein S _ d ═ (P _ do _ new > thr _ do), S _ t ═ P _ to _ new > thr _ to);
if the value of the vehicle state indication vector is 1, confirming that the vehicle has abnormal state behaviors;
and outputting alarm information carrying the vehicle state information and the vehicle target detection identification result.
In another aspect, the present invention provides a system for identifying an abnormal behavior state of a vehicle, the system comprising: the system comprises an acquisition unit, a processing unit and a display unit, wherein the acquisition unit is used for acquiring a monitoring area image acquired in real time and acquiring a vehicle detection frame position coordinate from the monitoring area image according to a preset vehicle detection identification model; (ii) a
The acquisition unit is further used for acquiring a cutting frame coordinate corresponding to the vehicle state identification frame according to the vehicle detection frame position coordinate, and processing the monitored area image according to the cutting frame coordinate to obtain a monitored area image sub-atlas;
the processing unit is used for processing the monitoring area image sub-atlas according to a preset vehicle state identification model to obtain a vehicle state identification vector corresponding to the monitoring area image, the preset vehicle state identification model is used for identifying the vehicle door opening probability, the trunk opening probability and the probability of people around the vehicle, and the vehicle state identification vector comprises the vehicle door opening probability and/or the trunk opening probability;
and the determining unit is used for comparing the door opening probability and/or trunk opening probability in the vehicle state identification vector with a preset state judgment threshold value to confirm whether the vehicle has abnormal state behaviors.
Further, the system further comprises:
the cutting unit is used for acquiring a vehicle detection frame from a monitoring area image stored in a database through the preset vehicle detection recognition model, and cutting and storing a local target area of the monitoring area image data according to the preset size of the length and width extension of the vehicle detection frame to serve as a training image data set;
the marking unit is used for acquiring a vehicle detection frame from a monitored area image stored in a database through the preset vehicle detection recognition model, and cutting and storing a local target area of the monitored area picture data according to the preset size of the length and width extension of the vehicle detection frame to serve as a training picture data set;
and the generating unit is used for inputting the model training data into a preset classification model to obtain a three-component vector Pv [ P _ do, P _ to, P _ per ] as the preset vehicle state identification model, wherein P _ do is the door opening probability, P _ to is the trunk opening probability, and P _ per is the existence probability of people near the vehicle.
Further, the obtaining unit is specifically configured to expand the length and width of the vehicle detection frame by a preset size to obtain a cropping frame coordinate corresponding to the vehicle state identification frame.
Further, the processing unit is specifically configured to obtain the vehicle door state information according to a formula P _ do _ new ═ a1 × P _ do + a2 × P _ per, where P _ do _ new is a door opening probability in consideration of human presence factors, a1 ═ exp (P _ do)/(exp (P _ do) + exp (P _ per)), and a2 ═ exp (P _ per)/(exp (P _ do) + exp (P _ per)); vehicle trunk status information is obtained according to the formula P _ to _ new (1-P _ per), wherein P _ to _ new takes into account the trunk opening probability after human presence.
Further, the confirming unit is specifically configured to obtain a vehicle state indication vector state [ S _ d, S _ t ], based on the preset state determination threshold thr [ [ thr _ do, thr _ to ] and the vehicle state identification vector Pv _ new [ [ P _ do _ new, P _ to _ new ], where S _ d ═ (P _ do _ new > thr _ do), S _ t ═ P _ to _ new > thr _ to); if the value of the vehicle state indication vector is 1, confirming that the vehicle has abnormal state behaviors; and outputting alarm information carrying the vehicle state information and the vehicle target detection identification result.
The method and the system for identifying the abnormal behavior state of the vehicle realize the identification and monitoring of the abnormal state of the vehicle by utilizing the existing vehicle detection and identification model to match with the trainable vehicle state identification model and fuse the identification result information, and feed back the information of the vehicle with the abnormality when the abnormal vehicle is confirmed by comparing the door opening probability and/or the trunk opening probability in the vehicle state identification vector with the preset state judgment threshold value. The invention is carried out by the existing vehicle detection and identification model, so the realization difficulty is reduced, and the invention can independently deploy the vehicle state identification model according to the actual demand condition independently of the vehicle detection and identification model, thereby ensuring the high efficiency, stability and controllability of the identification algorithm of the whole vehicle abnormal behavior state; meanwhile, the multi-classification vehicle state identification model adopted in the invention does not need positioning information, so that the marking requirement of a target detailed region is reduced, and the difficulty of data marking is greatly reduced.
Drawings
FIG. 1 is a flow chart of a method for identifying an abnormal behavior state of a vehicle according to the present invention;
fig. 2 is a schematic structural diagram of a vehicle abnormal behavior state recognition system provided by the invention.
Detailed Description
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
As shown in fig. 1, an embodiment of the present invention provides a method for identifying an abnormal behavior state of a vehicle, including the following steps:
101. and acquiring a monitoring area image acquired in real time, and acquiring the position coordinates of a vehicle detection frame from the monitoring area image according to a preset vehicle detection identification model.
For the embodiment of the present invention, step 101 may further include: acquiring a vehicle detection frame from a monitored area image stored in a database through the preset vehicle detection and identification model, and cutting and storing a local target area of the monitored area image data according to the preset size of the length and width extension of the vehicle detection frame to be used as a training image data set; marking the training picture data set according to the ternary vectors corresponding to the classification type sequence to obtain model training data, wherein the categories of the ternary vectors corresponding to the classification type sequence comprise vehicle door state information, trunk state information and state information of whether people exist around the vehicle; and inputting the model training data into a preset classification model to obtain a three-component vector Pv ═ P _ do, P _ to and P _ per as the preset vehicle state identification model, wherein P _ do is the vehicle door opening probability, P _ to is the trunk opening probability, and P _ per is the probability of existence of people near the vehicle.
The preset size may be 125%, 130%, 135% of the original size, and the like, and the embodiment of the present invention is not limited. The step of labeling the training picture data set according to the ternary vectors corresponding to the classification type order to obtain model training data may specifically include: carrying out data annotation on the collected training picture data set, and setting annotation data as a ternary vector corresponding to the classification type sequence: according to the vehicle state identification requirement, the classification types are respectively set as door opening, trunk opening and existence of a person, the corresponding label vector label is [ l1, l2 and l3], then the label vector label is marked according to the actual picture content situation, the label data set corresponding to the training picture data set is obtained, and complete model training data D is formed, for example, if the door opening angle in the picture is about >45 degrees, l1 is 1, otherwise l1 is 0; if the opening angle of the trunk in the picture is about >45 degrees, l2 is 1, otherwise l2 is 0; if a person exists in the area near the vehicle in the picture, l3 is equal to 1, otherwise l3 is equal to 0.
102. And acquiring a cutting frame coordinate corresponding to the vehicle state identification frame according to the position coordinate of the vehicle detection frame, and processing the monitored area image according to the cutting frame coordinate to obtain a monitored area image sub-picture set.
For the embodiment of the present invention, the step of obtaining coordinates of the cropping frame corresponding to the vehicle state identification frame according to the position of the vehicle detection frame includes: and expanding the length and width dimensions of the vehicle detection frame by preset dimensions to obtain the coordinates of the cutting frame corresponding to the vehicle state identification frame. Wherein, the preset size can be 125%, 130%, 135% of the original size, etc.
For the embodiment of the present invention, the step of processing the monitored area image according to the cropping frame coordinates to obtain the monitored area image sub-atlas may specifically include: inputting the camera image data collected in real time into an existing vehicle detection recognition model, obtaining the position coordinates pos0 of the vehicle detection frame of the current image as (x0, y0, w, h), expanding each detection frame area by 30%, obtaining the new cutting frame coordinates pos as (x0, y0, w 1.15, h 1.15), and cutting the target area of the original image to obtain the sub-image set Ip.
103. And processing the monitoring area image sub-image set according to a preset vehicle state identification model to obtain a vehicle state identification vector corresponding to the monitoring area image.
The preset vehicle state identification model is used for identifying the vehicle door opening probability, the trunk opening probability and the existence probability of people around the vehicle, and the vehicle state identification vector comprises the vehicle door opening probability and/or the trunk opening probability.
For the embodiment of the present invention, step 103 may specifically include: acquiring vehicle door state information according to a formula P _ do _ new ═ a1 ═ P _ do + a2 × (P _ per), wherein P _ do _ new is a door opening probability considering human presence, a1 ═ exp (P _ do)/(exp (P _ do) + exp (P _ per)), and a2 ═ exp (P _ per)/(exp (P _ do) + exp (P _ per)); vehicle trunk status information is obtained according to the formula P _ to _ new (1-P _ per), wherein P _ to _ new takes into account the trunk opening probability after human presence.
It should be noted that, the sub-map set Ip is input into the preset vehicle state identification model, and the current picture identification result vector Pv is obtained as [ P _ do, P _ to, P _ per ], and since the interaction relationship between the person and the door opening or trunk opening behavior of the monitoring system is expected to be different, in order to improve the final state identification accuracy, the influence effect of the recognition probability of the person on the other two items is also different, and the following information fusion evaluation mode can be defined according to the situation to obtain the final vehicle state identification vector Pv _ new as [ P _ do _ new, P _ to _ new ]: when the goal of identifying whether the vehicle door is opened is mainly to identify whether a person gets on or off the vehicle illegally, so that the interactive behaviors of the person and the vehicle door are mostly short-time behaviors which are synchronously performed, and therefore, the possibility of indicating an abnormal state is higher when two states exist simultaneously, so that the person is positively correlated with the behavior identification of the vehicle door opening state, and the comprehensive probability of the vehicle door opening state is finally expressed as P _ do _ new ═ a1 ═ P _ do + a2 ═ P _ per, a1 ═ exp (P _ do)/(exp (P _ do) + exp (P _ per)), a2 ═ exp (P _ per)/(exp (P _ do) + (P _ per)); when the trunk opening is identified, the trunk is intentionally opened so as to avoid behaviors such as camera monitoring photographing and payment evasion when a person parks, so that the interaction behavior of the person and the trunk is weak, the trunk is still in an open state after the person leaves in a period of time monitored and photographed at a large probability, the trunk opening state identification is influenced when the person interacts with the trunk, or the interaction behavior is reasonable behavior only in a short time, for example, the trunk object taken during parking belongs to normal behavior, so that the person is in negative correlation with the trunk opening state identification, and the comprehensive probability of the trunk opening state is finally expressed as P _ to _ new as P _ to (1-P _ per).
104. And comparing the door opening probability and/or trunk opening probability in the vehicle state identification vector with a preset state judgment threshold value to confirm whether the vehicle has abnormal state behaviors.
For the embodiment of the present invention, step 104 may specifically include: obtaining a vehicle state indication vector state [ S _ d, S _ t ] according to a preset state determination threshold thr [ [ thr _ do, thr _ to ] and a vehicle state identification vector Pv _ new [ [ P _ do _ new, P _ to _ new ], wherein S _ d ═ (P _ do _ new > thr _ do), S _ t ═ P _ to _ new > thr _ to); if the value of the vehicle state indication vector is 1, confirming that the vehicle has abnormal state behaviors; and outputting alarm information carrying the vehicle state information and the vehicle target detection identification result.
For example, a preset state determination threshold thr is set as [ thr _ do, thr _ to ], a vehicle state identification vector Pv _ new is further determined, and a final vehicle state indication vector state is obtained as [ S _ d, S _ t ], wherein S _ d is (P _ do _ new > thr _ do), S _ t is (P _ to _ new > thr _ to), if the vehicle state indication vector state is [0,0], the identification result is ignored, an abnormal state monitoring signal is not fed back to the system, if (state) is 1, any (), the license plate identification information of the current target vehicle is retrieved, and the information such as the existing vehicle state indication vector state, the vehicle target detection identification result and the like is combined and comprehensively transmitted to a background monitoring system, so as to further perform a subsequent abnormal processing flow; if the license plate identification information which can be obtained by the current frame due to target shielding, poor picture quality and the like is missing or the reliability of the identification information is not high, the current target is tracked by continuously acquiring pictures in front and back time periods, and the license plate information of the corresponding vehicle is captured, so that the acquisition accuracy of the vehicle information in the abnormal state can be improved.
The invention provides a method for identifying abnormal behavior states of a vehicle, which realizes identification and monitoring of abnormal states of the vehicle by matching an existing vehicle detection and identification model with a trainable vehicle state identification model and fusing identification result information, and feeds back the information of the vehicle with the abnormality when the abnormal vehicle is confirmed by comparing the door opening probability and/or trunk opening probability in the vehicle state identification vector with a preset state judgment threshold value. The invention is carried out by the existing vehicle detection and identification model, so the realization difficulty is reduced, and the invention can independently deploy the vehicle state identification model according to the actual demand condition independently of the vehicle detection and identification model, thereby ensuring the high efficiency, stability and controllability of the identification algorithm of the whole vehicle abnormal behavior state; meanwhile, the multi-classification vehicle state identification model adopted in the invention does not need positioning information, so that the marking requirement of a target detailed region is reduced, and the difficulty of data marking is greatly reduced.
In order to implement the method provided by the embodiment of the present invention, an embodiment of the present invention provides a system for identifying an abnormal behavior state of a vehicle, as shown in fig. 2, the system includes: an acquisition unit 21, a processing unit 22, a determination unit 23.
The acquiring unit 21 is configured to acquire a monitoring area image acquired in real time, and acquire a position coordinate of a vehicle detection frame from the monitoring area image according to a preset vehicle detection identification model.
The obtaining unit 21 is further configured to obtain a trimming frame coordinate corresponding to the vehicle state identification frame according to the vehicle detection frame position coordinate, and process the monitored area image according to the trimming frame coordinate to obtain a monitored area image sub-atlas.
And the processing unit 22 is configured to process the monitoring area image sub-image set according to a preset vehicle state identification model to obtain a vehicle state identification vector corresponding to the monitoring area image.
The preset vehicle state identification model is used for identifying the vehicle door opening probability, the trunk opening probability and the existence probability of people around the vehicle, and the vehicle state identification vector comprises the vehicle door opening probability and/or the trunk opening probability.
The determining unit 23 is configured to compare the door opening probability and/or the trunk opening probability in the vehicle state identification vector with a preset state determination threshold, and determine whether the vehicle has an abnormal state behavior.
Further, the system further comprises: and the cutting unit 24 is configured to obtain a vehicle detection frame from a monitored area image stored in the database through the preset vehicle detection recognition model, and to cut and store a local target area of the monitored area image data according to the preset size expanded by the length and width dimensions of the vehicle detection frame, so as to serve as a training image data set.
The preset size may be 125%, 130%, 135% of the original size, and the like, and the embodiment of the present invention is not limited.
Further, the system further comprises: and the labeling unit 25 is configured to obtain a vehicle detection frame from a monitored area image stored in a database through the preset vehicle detection recognition model, and cut and store a local target area of the monitored area image data according to the preset size expanded by the length and width dimensions of the vehicle detection frame, so as to serve as a training image data set.
Further, the system further comprises: the generating unit 26 is configured to input the model training data to a preset classification model, and obtain a three-component vector Pv [ P _ do, P _ to, P _ per ] as the preset vehicle state identification model, where P _ do is a door opening probability, P _ to is a trunk opening probability, and P _ per is a probability that a person exists near the vehicle.
Further, the obtaining unit 21 is specifically configured to expand the length and width of the vehicle detection frame by a preset size to obtain a coordinate of a cropping frame corresponding to the vehicle state identification frame.
Further, the processing unit 22 is specifically configured to obtain the vehicle door state information according to a formula P _ do _ new ═ a1 × P _ do + a2 × P _ per, where P _ do _ new is a door opening probability considering human presence factors, a1 ═ exp (P _ do)/(exp (P _ do) + exp (P _ per)), and a2 ═ exp (P _ per)/(exp (P _ do) + exp (P _ per)); vehicle trunk status information is obtained according to the formula P _ to _ new (1-P _ per), wherein P _ to _ new takes into account the trunk opening probability after human presence.
Further, the confirming unit 23 is specifically configured to obtain a vehicle state indication vector state [ S _ d, S _ t ], based on the preset state determination threshold thr [ [ thr _ do, thr _ to ] and the vehicle state identification vector Pv _ new [ [ P _ do _ new, P _ to _ new ], where S _ d ═ (P _ do _ new > thr _ do), S _ t ═ thr _ to); if the value of the vehicle state indication vector is 1, confirming that the vehicle has abnormal state behaviors; and outputting alarm information carrying the vehicle state information and the vehicle target detection identification result.
The invention provides a vehicle abnormal behavior state identification system, which realizes identification and monitoring of abnormal states of a vehicle by matching an existing vehicle detection identification model with a trainable vehicle state identification model and fusing identification result information, and feeds back abnormal vehicle information when the abnormal vehicle is confirmed by comparing a door opening probability and/or a trunk opening probability in a vehicle state identification vector with a preset state judgment threshold value. The invention is carried out by the existing vehicle detection and identification model, so the realization difficulty is reduced, and the invention can independently deploy the vehicle state identification model according to the actual demand condition independently of the vehicle detection and identification model, thereby ensuring the high efficiency, stability and controllability of the identification algorithm of the whole vehicle abnormal behavior state; meanwhile, the multi-classification vehicle state identification model adopted in the invention does not need positioning information, so that the marking requirement of a target detailed region is reduced, and the difficulty of data marking is greatly reduced.
It should be understood that the specific order or hierarchy of steps in the processes disclosed is an example of exemplary approaches. Based upon design preferences, it is understood that the specific order or hierarchy of steps in the processes may be rearranged without departing from the scope of the present disclosure. The accompanying method claims present elements of the various steps in a sample order, and are not intended to be limited to the specific order or hierarchy presented.
In the foregoing detailed description, various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments of the subject matter require more features than are expressly recited in each claim. Rather, as the following claims reflect, invention lies in less than all features of a single disclosed embodiment. Thus, the following claims are hereby expressly incorporated into the detailed description, with each claim standing on its own as a separate preferred embodiment of the invention.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. To those skilled in the art; various modifications to these embodiments will be readily apparent, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the disclosure. Thus, the present disclosure is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
What has been described above includes examples of one or more embodiments. It is, of course, not possible to describe every conceivable combination of components or methodologies for purposes of describing the aforementioned embodiments, but one of ordinary skill in the art may recognize that many further combinations and permutations of various embodiments are possible. Accordingly, the embodiments described herein are intended to embrace all such alterations, modifications and variations that fall within the scope of the appended claims. Furthermore, to the extent that the term "includes" is used in either the detailed description or the claims, such term is intended to be inclusive in a manner similar to the term "comprising" as "comprising" is interpreted when employed as a transitional word in a claim. Furthermore, any use of the term "or" in the specification of the claims is intended to mean a "non-exclusive or".
Those of skill in the art will further appreciate that the various illustrative logical blocks, units, and steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate the interchangeability of hardware and software, various illustrative components, elements, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design requirements of the overall system. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present embodiments.
The various illustrative logical blocks, or elements, described in connection with the embodiments disclosed herein may be implemented or performed with a general purpose processor, a digital signal processor, an Application Specific Integrated Circuit (ASIC), a field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general-purpose processor may be a microprocessor, but in the alternative, the processor may be any conventional processor, controller, microcontroller, or state machine. A processor may also be implemented as a combination of computing devices, e.g., a digital signal processor and a microprocessor, a plurality of microprocessors, one or more microprocessors in conjunction with a digital signal processor core, or any other similar configuration.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may be stored in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. For example, a storage medium may be coupled to the processor such the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium may be integral to the processor. The processor and the storage medium may reside in an ASIC, which may be located in a user terminal. In the alternative, the processor and the storage medium may reside in different components in a user terminal.
In one or more exemplary designs, the functions described above in connection with the embodiments of the invention may be implemented in hardware, software, firmware, or any combination of the three. If implemented in software, the functions may be stored on or transmitted over as one or more instructions or code on a computer-readable medium. Computer-readable media includes both computer storage media and communication media that facilitate transfer of a computer program from one place to another. Storage media may be any available media that can be accessed by a general purpose or special purpose computer. For example, such computer-readable media can include, but is not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to carry or store program code in the form of instructions or data structures and which can be read by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. Additionally, any connection is properly termed a computer-readable medium, and, thus, is included if the software is transmitted from a website, server, or other remote source via a coaxial cable, fiber optic cable, twisted pair, Digital Subscriber Line (DSL), or wirelessly, e.g., infrared, radio, and microwave. Such discs (disk) and disks (disc) include compact disks, laser disks, optical disks, DVDs, floppy disks and blu-ray disks where disks usually reproduce data magnetically, while disks usually reproduce data optically with lasers. Combinations of the above may also be included in the computer-readable medium.
The above-mentioned embodiments are intended to illustrate the objects, technical solutions and advantages of the present invention in further detail, and it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present invention, and are not intended to limit the scope of the present invention, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (10)
1. A method for identifying an abnormal behavior state of a vehicle, the method comprising:
acquiring a monitoring area image acquired in real time, and acquiring a position coordinate of a vehicle detection frame from the monitoring area image according to a preset vehicle detection identification model;
acquiring a cutting frame coordinate corresponding to a vehicle state identification frame according to the position coordinate of the vehicle detection frame, and processing the monitored area image according to the cutting frame coordinate to obtain a monitored area image sub-picture set;
processing the monitoring area image sub-map set according to a preset vehicle state identification model to obtain a vehicle state identification vector corresponding to the monitoring area image, wherein the preset vehicle state identification model is used for identifying the vehicle door opening probability, the trunk opening probability and the probability of people around the vehicle, and the vehicle state identification vector comprises the vehicle door opening probability and/or the trunk opening probability;
and comparing the door opening probability and/or trunk opening probability in the vehicle state identification vector with a preset state judgment threshold value to confirm whether the vehicle has abnormal state behaviors.
2. The method for identifying the abnormal behavior state of the vehicle according to claim 1, wherein the step of acquiring the monitoring area image acquired in real time is preceded by the method further comprising:
acquiring a vehicle detection frame from a monitored area image stored in a database through the preset vehicle detection and identification model, and cutting and storing a local target area of the monitored area image data according to the preset size of the length and width extension of the vehicle detection frame to be used as a training image data set;
marking the training picture data set according to the ternary vectors corresponding to the classification type sequence to obtain model training data, wherein the categories of the ternary vectors corresponding to the classification type sequence comprise vehicle door state information, trunk state information and state information of whether people exist around the vehicle;
and inputting the model training data into a preset classification model to obtain a three-component vector Pv ═ P _ do, P _ to and P _ per as the preset vehicle state identification model, wherein P _ do is the vehicle door opening probability, P _ to is the trunk opening probability, and P _ per is the probability of existence of people near the vehicle.
3. The method for identifying the abnormal behavior state of the vehicle according to claim 1 or 2, wherein the step of obtaining the coordinates of the trimming frame corresponding to the vehicle state identification frame based on the position of the vehicle detection frame comprises:
and expanding the length and width dimensions of the vehicle detection frame by preset dimensions to obtain the coordinates of the cutting frame corresponding to the vehicle state identification frame.
4. The method for identifying the abnormal behavior state of the vehicle according to claim 2, wherein the step of processing the sub-image set of the monitored area image according to a preset vehicle state identification model to obtain the vehicle state identification vector corresponding to the monitored area image comprises the steps of:
acquiring vehicle door state information according to a formula P _ do _ new ═ a1 ═ P _ do + a2 × (P _ per), wherein P _ do _ new is a door opening probability considering human presence, a1 ═ exp (P _ do)/(exp (P _ do) + exp (P _ per)), and a2 ═ exp (P _ per)/(exp (P _ do) + exp (P _ per));
vehicle trunk status information is obtained according to the formula P _ to _ new (1-P _ per), wherein P _ to _ new takes into account the trunk opening probability after human presence.
5. The method for identifying the abnormal behavior state of the vehicle according to claim 4, wherein the step of determining whether the vehicle has the abnormal behavior state according to the comparison between the door opening probability and/or the trunk opening probability in the vehicle state identification vector and a preset state determination threshold comprises:
obtaining a vehicle state indication vector state [ S _ d, S _ t ] according to a preset state determination threshold thr [ [ thr _ do, thr _ to ] and a vehicle state identification vector Pv _ new [ [ P _ do _ new, P _ to _ new ], wherein S _ d ═ (P _ do _ new > thr _ do), S _ t ═ P _ to _ new > thr _ to);
if the value of the vehicle state indication vector is 1, confirming that the vehicle has abnormal state behaviors;
and outputting alarm information carrying the vehicle state information and the vehicle target detection identification result.
6. A system for identifying an abnormal behavior state of a vehicle, the system comprising:
the system comprises an acquisition unit, a processing unit and a display unit, wherein the acquisition unit is used for acquiring a monitoring area image acquired in real time and acquiring a vehicle detection frame position coordinate from the monitoring area image according to a preset vehicle detection identification model;
the acquisition unit is further used for acquiring a cutting frame coordinate corresponding to the vehicle state identification frame according to the vehicle detection frame position coordinate, and processing the monitored area image according to the cutting frame coordinate to obtain a monitored area image sub-atlas;
the processing unit is used for processing the monitoring area image sub-atlas according to a preset vehicle state identification model to obtain a vehicle state identification vector corresponding to the monitoring area image, the preset vehicle state identification model is used for identifying the vehicle door opening probability, the trunk opening probability and the probability of people around the vehicle, and the vehicle state identification vector comprises the vehicle door opening probability and/or the trunk opening probability;
and the determining unit is used for comparing the door opening probability and/or trunk opening probability in the vehicle state identification vector with a preset state judgment threshold value to confirm whether the vehicle has abnormal state behaviors.
7. The system for identifying an abnormal behavior state of a vehicle according to claim 6, further comprising:
the cutting unit is used for acquiring a vehicle detection frame from a monitoring area image stored in a database through the preset vehicle detection recognition model, and cutting and storing a local target area of the monitoring area image data according to the preset size of the length and width extension of the vehicle detection frame to serve as a training image data set;
the marking unit is used for acquiring a vehicle detection frame from a monitored area image stored in a database through the preset vehicle detection recognition model, and cutting and storing a local target area of the monitored area picture data according to the preset size of the length and width extension of the vehicle detection frame to serve as a training picture data set;
and the generating unit is used for inputting the model training data into a preset classification model to obtain a three-component vector Pv [ P _ do, P _ to, P _ per ] as the preset vehicle state identification model, wherein P _ do is the door opening probability, P _ to is the trunk opening probability, and P _ per is the existence probability of people near the vehicle.
8. A vehicle abnormal behavior state recognition system according to claim 6 or 7,
the obtaining unit is specifically configured to expand the length and width dimensions of the vehicle detection frame by preset dimensions to obtain coordinates of the cropping frame corresponding to the vehicle state identification frame.
9. The system for identifying an abnormal behavior state of a vehicle according to claim 7,
the processing unit is specifically configured to obtain vehicle door state information according to a formula P _ do _ new ═ a1 ═ P _ do + a2 × P _ per, where P _ do _ new is a door opening probability in consideration of human presence, a1 ═ exp (P _ do)/(exp (P _ do) + exp (P _ per)), and a2 ═ exp (P _ per)/(exp (P _ do) + exp (P _ per)); vehicle trunk status information is obtained according to the formula P _ to _ new (1-P _ per), wherein P _ to _ new takes into account the trunk opening probability after human presence.
10. The system for identifying an abnormal behavior state of a vehicle according to claim 9,
the confirming unit is specifically configured to obtain a vehicle state indication vector state ═ S _ d, S _ t according to a preset state determination threshold thr ═ thr _ do, thr _ to ] and a vehicle state identification vector Pv _ new ═ P _ do _ new, P _ to _ new ], where S _ d ═ (P _ do _ new > thr _ do), and S _ t ═ P _ to _ new > thr _ to; if the value of the vehicle state indication vector is 1, confirming that the vehicle has abnormal state behaviors; and outputting alarm information carrying the vehicle state information and the vehicle target detection identification result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111515694.9A CN114220074A (en) | 2021-12-13 | 2021-12-13 | Method and system for identifying abnormal behavior state of vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111515694.9A CN114220074A (en) | 2021-12-13 | 2021-12-13 | Method and system for identifying abnormal behavior state of vehicle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114220074A true CN114220074A (en) | 2022-03-22 |
Family
ID=80701186
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111515694.9A Pending CN114220074A (en) | 2021-12-13 | 2021-12-13 | Method and system for identifying abnormal behavior state of vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114220074A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115082571A (en) * | 2022-07-20 | 2022-09-20 | 深圳云游四海信息科技有限公司 | Anomaly detection method and system for in-road parking camera |
-
2021
- 2021-12-13 CN CN202111515694.9A patent/CN114220074A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115082571A (en) * | 2022-07-20 | 2022-09-20 | 深圳云游四海信息科技有限公司 | Anomaly detection method and system for in-road parking camera |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3806064B1 (en) | Method and apparatus for detecting parking space usage condition, electronic device, and storage medium | |
| CN111931627A (en) | Vehicle re-identification method and device based on multi-mode information fusion | |
| Abdullah et al. | YOLO-based three-stage network for Bangla license plate recognition in Dhaka metropolitan city | |
| US11113582B2 (en) | Method and system for facilitating detection and identification of vehicle parts | |
| CN110580808B (en) | Information processing method and device, electronic equipment and intelligent traffic system | |
| CN103020624A (en) | Intelligent marking, searching and replaying method and device for surveillance videos of shared lanes | |
| CN102170563A (en) | Intelligent person capture system and person monitoring management method | |
| CN110619277A (en) | Multi-community intelligent deployment and control method and system | |
| US11321571B2 (en) | Automatic license plate recognition (ALPR) and vehicle identification profile methods and systems | |
| CN112633255B (en) | Target detection method, device and equipment | |
| CN108805184B (en) | Image recognition method and system for fixed space and vehicle | |
| CN110490127A (en) | Close rule recognition methods and device, storage medium, electronic device in position | |
| CN114220074A (en) | Method and system for identifying abnormal behavior state of vehicle | |
| CN113473077A (en) | Multi-dimension-based roadside reverse parking judgment method and system | |
| CN103761345A (en) | Video retrieval method based on OCR character recognition technology | |
| EP4064218A1 (en) | Vehicle identification profile methods and systems at the edge | |
| CN111291722A (en) | Vehicle weight recognition system based on V2I technology | |
| Roy et al. | Vehicle number plate recognition and parking system | |
| WO2022135242A1 (en) | Virtual positioning method and apparatus, and virtual positioning system | |
| Baviskar et al. | Auto Number Plate Recognition | |
| Eleuch et al. | Feature-based vehicle tracking at roundabouts in visual sensor networks | |
| Gupta et al. | An Automated Number Plate Recognition System based on Optical Character Recognition | |
| CN113283303A (en) | License plate recognition method and device | |
| KR100855590B1 (en) | System and method for multi car recognition of multi lens and multi view point | |
| CN113449624B (en) | Method and device for determining vehicle behavior based on pedestrian re-identification |
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 |