CN117671572A - Multi-platform linkage road image model processing system and method - Google Patents
Multi-platform linkage road image model processing system and method Download PDFInfo
- Publication number
- CN117671572A CN117671572A CN202410145468.3A CN202410145468A CN117671572A CN 117671572 A CN117671572 A CN 117671572A CN 202410145468 A CN202410145468 A CN 202410145468A CN 117671572 A CN117671572 A CN 117671572A
- Authority
- CN
- China
- Prior art keywords
- information
- module
- platform
- police
- model
- 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
- 238000012545 processing Methods 0.000 title claims abstract description 101
- 238000000034 method Methods 0.000 title claims description 27
- 238000012544 monitoring process Methods 0.000 claims abstract description 123
- 238000003672 processing method Methods 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 91
- 230000006399 behavior Effects 0.000 claims description 68
- 238000013528 artificial neural network Methods 0.000 claims description 48
- 206010039203 Road traffic accident Diseases 0.000 claims description 20
- 238000010276 construction Methods 0.000 claims description 13
- 238000012706 support-vector machine Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 2
- 238000007726 management method Methods 0.000 description 56
- 238000010586 diagram Methods 0.000 description 4
- 238000012937 correction Methods 0.000 description 3
- 238000000926 separation method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 230000001755 vocal effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/44—Event detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
-
- 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
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- 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
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
-
- 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
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0137—Measuring and analyzing of parameters relative to traffic conditions for specific applications
- G08G1/0141—Measuring and analyzing of parameters relative to traffic conditions for specific applications for traffic information dissemination
-
- 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
- G08G1/04—Detecting movement of traffic to be counted or controlled using optical or ultrasonic detectors
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/133—Protocols for remote procedure calls [RPC]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Analytical Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Databases & Information Systems (AREA)
- Signal Processing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a multi-platform linkage road image model processing system and a multi-platform linkage road image model processing method, which belong to the technical field of security monitoring and comprise a plurality of regional subsystems arranged in different regions and a general control system arranged in a cloud, wherein the regional subsystems comprise a video acquisition module, a primary judgment module, a primary sorting module and a site alarm module, and the general control system comprises a secondary sorting module, a traffic police interface module, a urban pipe interface module and a civil police interface module.
Description
Technical Field
The invention relates to the field of security monitoring, in particular to a multi-platform linkage road image model processing system and method.
Background
The road monitoring system is an important component of the police command system at present, provides the most visual reflection of the site situation, is a basic guarantee for implementing accurate scheduling, and front-end equipment of key places and monitoring points transmits video images to the traffic command center in various modes for storing, processing and publishing information. Many monitoring systems can automatically identify illegal laws of citizens, and help law enforcement personnel such as polices, urban management and the like to acquire field information. The current public security command system is provided with service platforms of various traffic safety comprehensive service systems which are collectively called traffic police platforms, and the service platforms are used for traffic police internal circulation, data storage or flow. Urban management and polices have similar service platforms.
The prior art solutions described above have the following drawbacks: the existing road monitoring system only can upload command centers after illegal information is judged, law enforcement personnel still need to judge processing conditions after receiving photos, then edit law enforcement content and distribute law enforcement personnel, and the processing process is complex, time-consuming and labor-consuming.
Disclosure of Invention
In order to simplify the case processing process of law enforcement personnel by using a road monitoring system, time and labor are saved, and the application provides a multi-platform linkage road image model processing system and method.
On one hand, the multi-platform linkage road image model processing system provided by the application adopts the following technical scheme:
the multi-platform linkage road image model processing system comprises a plurality of regional subsystems arranged in different regions and a general control system arranged at a cloud, wherein the regional subsystems comprise a video acquisition module, a model construction module, a primary judgment module, a primary sorting module and a field alarm module;
the video acquisition module acquires image information with position information and transmits the image information to the primary judgment module;
the model construction module builds an illegal judgment model, builds a first monitoring model and a second monitoring model in the illegal judgment model, the first monitoring model identifies whether pedestrians and/or vehicles exist in the image information, the second monitoring model identifies illegal behaviors of the image information of the pedestrians and/or vehicles, the illegal judgment model is trained, and the model construction module transmits the trained illegal judgment model to the primary judgment module;
The primary judgment module receives the illegal judgment model, judges whether illegal behaviors exist in the image information through the illegal judgment model, frames and selects the areas with illegal behaviors in the image information, judges the type of the illegal behaviors, associates the type of the illegal behaviors with the image information to form primary judgment information, and sends the primary judgment information to the primary sorting module;
the primary sorting module classifies the primary judgment information into processing types of public warning, traffic violation, traffic accident, urban management responsible event and civil police responsible event, sends the primary judgment information under the public warning processing type to the on-site alarm module for processing, and sends other primary judgment information to the master control system;
the on-site alarm module plays the received preliminary judgment information;
and the master control system sends the received preliminary judgment information to a traffic police platform, a city management platform or a civil police platform according to the processing type.
Through adopting above-mentioned scheme, this system can be in judging fast whether take place the illegal action in the picture of taking, can also broadcast the information directly or select separately the platform that the law enforcement personnel of difference used, can carry out more detailed classification to the information in the separation process, simplify the handling process, labour saving and time saving.
Preferably, the model construction module invokes the image information of the video acquisition module and divides the image information into a first training sample and a second training sample, and constructs a first monitoring model, wherein the first monitoring model comprises two classifiers of different types, the two classifiers comprise a support vector machine classifier and a Bayesian classifier, the first training sample is selected, and the first training sample comprises a first classification sample of a man-vehicle road and a second classification sample of an unmanned vehicle road; the road with the pedestrians and/or vehicles is a road with the pedestrians and/or vehicles, the road with the unmanned vehicles is a road without the pedestrians or vehicles, the first classification sample and the second classification sample both comprise identification data of the corresponding road, and the identification data comprise road safety identifications and traffic rule identifications; respectively inputting the first classification sample and the second classification sample into two classifiers for training; taking classification results of the two classifiers as prediction results of the first monitoring model; inputting the identification data into a first monitoring model to obtain a first monitoring result output by the first monitoring model; if the first monitoring result shows that pedestrians and/or vehicles exist in the target road, acquiring image data of the target road; constructing a second monitoring model, wherein the second monitoring model comprises a plurality of neural networks; the plurality of neural networks includes a VGG16 neural network and a U-Net neural network; the second monitoring model takes the U-Net neural network as a main body of the second monitoring model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through imageNet; the process of training the second monitoring model includes: selecting a second training sample, wherein the second training sample comprises image data of a man-vehicle road; the man-vehicle road is a road with pedestrians and/or vehicles; inputting a second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model; inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model, wherein the second monitoring result is various illegal behaviors; the model construction module transmits the trained illegal judgment model to the primary judgment module;
The primary sorting module is preset with a behavior type table, the behavior type table comprises a plurality of illegal behaviors and a plurality of processing types, each illegal behavior is associated with one processing type, the processing types comprise public warning, traffic violations, traffic accidents, urban management responsible events and civil police responsible events, the primary sorting module sorts the primary judgment information into corresponding processing types according to the behavior type table, and the primary sorting module sends the primary judgment information under the public warning processing types to the on-site alarm module and sends other primary judgment information to the master control system;
the on-site alarm module comprises a multimedia device arranged near the regional subsystem, converts the type of illegal and illegal behaviors in the preliminary judgment information into character information after receiving the preliminary judgment information, and plays the character information and the image information in the preliminary judgment information through the multimedia device;
the master control system comprises a secondary sorting module, a traffic police interface module, a city pipe interface module and a civil police interface module;
the secondary sorting module receives the preliminary judgment information and then acquires the processing type corresponding to the preliminary judgment information and the position information corresponding to the preliminary judgment information, combines the preliminary judgment information of the traffic violation processing type and the traffic accident processing type with the position information and then sends the combined information to the traffic police interface module, combines the preliminary judgment information of the urban management responsible event processing type with the position information and then sends the combined information to the urban pipe interface module, and combines the preliminary judgment information of the civil police responsible event processing type with the position information and then sends the combined information to the civil police interface module;
The traffic police interface module is connected with the traffic police platform and sends the received preliminary judgment information to the traffic police platform;
the urban pipe interface module is connected with the urban pipe platform and sends the received preliminary judgment information to the urban pipe platform;
the police interface module is connected with the police platform and sends the received preliminary judgment information to the police platform.
Preferably, the master control system further comprises a traffic judgment module, a city management judgment module and a police judgment module;
the traffic judgment module receives the preliminary judgment information under the traffic accident handling type output by the secondary sorting module, calls out the image information in the preliminary judgment information, identifies the face information of a frame selection area in the image information through face recognition software, uploads the face information and a query request to the traffic police platform, receives the driver information returned by the traffic police platform, associates the driver information with the frame selection area, sends the driver information to the traffic police interface module, and sends the driver information and the preliminary judgment information to the traffic police platform together;
the traffic judgment module sends a position request to the traffic police platform, receives personnel position information returned by the traffic police platform, inquires the nearest personnel position information according to the position information related to the preliminary judgment information, plans a shortest path according to the inquired personnel position information and the position information, sends the personnel position information and the shortest path to the traffic police interface module, and sends the personnel position information and the shortest path to the traffic police platform;
The urban management judgment module receives preliminary judgment information under the event processing type, which is output by the secondary sorting module, and sends a position request to the urban management platform, receives personnel position information returned by the urban management platform, inquires the nearest personnel position information according to the position information related to the preliminary judgment information, plans a shortest path according to the inquired personnel position information and the position information, and sends the personnel position information and the shortest path to the urban management platform;
the police judgment module receives the preliminary judgment information of the police responsible event processing type output by the secondary sorting module, the police judgment module sends a position request to the police platform, receives the personnel position information returned by the police platform, inquires the nearest personnel position information according to the position information related to the preliminary judgment information, plans the shortest path according to the inquired personnel position information and the position information, sends the personnel position information and the shortest path to the police interface module, and the police interface module sends the personnel position information and the shortest path to the police platform.
By adopting the scheme, the system can automatically carry out face recognition on the image information of the traffic accident type, inquires the driver information including the driving license, the driving license and the like, and is convenient for traffic police to rapidly process the accident. The system can quickly inquire recently available post law enforcement personnel and plan a route for the law enforcement personnel platform according to the accident position, and help the platform dispatch the law enforcement personnel.
Preferably, the master control system further comprises a traffic police feedback module, a city management feedback module, a civil police feedback module and a model training module;
the traffic police feedback module is connected with the traffic police platform, receives the misjudgment information related to the preliminary judgment information sent by the traffic police platform, and sends the preliminary judgment information related to the misjudgment information to the model training module;
the urban management feedback module is connected with the urban management platform, receives the misjudgment information related to the preliminary judgment information and sent by the urban management platform, and sends the preliminary judgment information related to the misjudgment information to the model training module;
the police feedback module is connected with the police platform, receives the misjudgment information related to the preliminary judgment information sent by the police platform, and sends the preliminary judgment information related to the misjudgment information to the model training module;
The model training module calls the illegal judgment model of the primary judgment module, the model training module calls the image information in the received primary judgment information, and the illegal judgment model is trained according to the framed area in the image information, so that the illegal judgment model does not recognize the image with the same characteristics as illegal behaviors.
By adopting the scheme, the law enforcement personnel platform can return misjudged image information to the system, and the system carries out correction training on the illegal judgment model.
Preferably, the traffic police feedback module judges whether the received preliminary judgment information is in the warning processing type, if so, the preliminary judgment information is transmitted to the on-site alarm module, and the on-site alarm module cancels the multimedia equipment to play the text information and the image information corresponding to the preliminary judgment information after receiving the preliminary judgment information transmitted by the traffic police feedback module.
By adopting the scheme, the system can quickly cancel the played misjudgment information according to the feedback information.
Preferably, the traffic police interface module receives a call instruction of the traffic police platform, calls image information corresponding to the regional subsystem according to the call instruction and transmits the image information to the traffic police platform;
The urban pipe interface module receives a calling instruction of the urban pipe platform, calls image information corresponding to the regional subsystem according to the calling instruction and transmits the image information to the urban pipe platform;
the police interface module receives a call instruction of the police platform, calls image information corresponding to the regional subsystem according to the call instruction and transmits the image information to the police platform;
the traffic police feedback module receives the image information after the frame selection of the area transmitted by the traffic police platform, and sends the image information after the frame selection of the area to the model training module;
the city management feedback module receives image information of the frame selected area transmitted by the city management platform, and sends the image information of the frame selected area to the model training module;
the police feedback module receives the image information after the area is selected by the frame and transmitted by the police platform, and sends the image information after the area is selected by the frame to the model training module;
and the model training module receives the image information of the frame selected area, then preprocesses the image information of the frame selected area, and adds the preprocessed image information into a database of the illegal judgment model.
By adopting the scheme, the system can actively request the law enforcement personnel platform to return a result when uploading the image information, and if the law enforcement personnel platform returns the image information which is not judged by the system to the system, the system can automatically add the image information into the database of the illegal judgment model, so that the database can be continuously perfected.
Preferably, the on-site alarm module presets a plurality of voice warnings, an image area and a text area on the multimedia device, arranges the received image information and text information in the image area and the text area, selects the voice warnings according to the type of illegal and illegal behaviors in the preliminary judgment information, and plays the voice warnings through the multimedia device.
By adopting the scheme, the system can play the image information under the oral warning type in the form of images, characters and voice, and the voice is used for warning offensive personnel.
On the other hand, the multi-platform linkage road image model processing method provided by the application adopts the following technical scheme:
the multi-platform linkage road image model processing method comprises the road image model processing system, and comprises the following steps:
the regional subsystem presets a behavior category table;
the regional subsystem shoots image information;
the regional subsystem builds an illegal judgment model, invokes image information and divides the image information into a first training sample and a second training sample, builds a first monitoring model, wherein the first monitoring model comprises two different classes of classifiers, the two classifiers comprise a support vector machine classifier and a Bayesian classifier, the first training sample is selected, and the identification data comprises a road safety identification and a traffic rule identification; respectively inputting the first classification sample and the second classification sample into two classifiers for training; taking classification results of the two classifiers as prediction results of the first monitoring model; inputting the identification data into a first monitoring model to obtain a first monitoring result output by the first monitoring model; if the first monitoring result shows that pedestrians and/or vehicles exist in the target road, acquiring image data of the target road; constructing a second monitoring model, wherein the second monitoring model comprises a plurality of neural networks; the plurality of neural networks includes a VGG16 neural network and a U-Net neural network; the second monitoring model takes the U-Net neural network as a main body of the second monitoring model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through imageNet; the process of training the second monitoring model includes: selecting a second training sample, wherein the second training sample comprises image data of a man-vehicle road; inputting a second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model; inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model, wherein the second monitoring result is various illegal behaviors;
The regional subsystem judges whether the illegal behavior exists in the image information through the illegal judging model, and selects the regional frame with the illegal behavior in the image information, and judges the type of the illegal behavior;
the regional subsystem associates the type of the illegal and illegal behaviors with the image information to form preliminary judgment information;
the regional subsystem classifies the preliminary judgment information into corresponding processing types according to the behavior type table;
setting a multimedia device near the regional subsystem;
the regional subsystem selects the preliminary judgment information under the warning processing type, converts the illegal behavior type in the preliminary judgment information into character information, and plays the character information and the image information in the preliminary judgment information through the multimedia equipment;
the regional subsystem sends other preliminary judgment information to the master control system;
the master control system acquires the processing type corresponding to the preliminary judgment information and the position information corresponding to the preliminary judgment information;
the primary judgment information of the traffic violation processing type and the traffic accident processing type is combined with the position information and then sent to a traffic police platform, the primary judgment information of the urban management responsible event processing type is combined with the position information and then sent to the urban management platform, and the primary judgment information of the civil police responsible event processing type is combined with the position information and then sent to the civil police platform.
Through adopting above-mentioned scheme, can be in judging fast whether take place the illegal action in the picture of taking, can also broadcast the information directly or select separately the platform that the law enforcement personnel of difference used, can carry out the more detailed classification to the information in the separation process, simplify the handling process, labour saving and time saving.
Preferably, the method further comprises the following steps:
after receiving the preliminary judgment information under the traffic accident handling type, the master control system calls out the image information in the preliminary judgment information, and the face information and the query request are uploaded to a traffic police platform by recognizing the face information of the frame-selected area in the image information through face recognition software;
the traffic control system receives the driver information returned by the traffic police platform, the traffic judgment module correlates the driver information with the frame selection area, the driver information is sent to the traffic police interface module, and the traffic police interface module sends the driver information and the preliminary judgment information to the traffic police platform;
the master control system sends a position request to the traffic police platform;
the general control system receives personnel position information returned by the traffic police platform, plans a shortest path according to the inquired personnel position information and the position information, and sends the personnel position information and the shortest path to the traffic police platform;
After receiving the preliminary judgment information of the type of the event processing in charge of the urban management, the master control system sends a position request to the urban management platform;
the main control system receives personnel position information returned by the urban management platform, the urban management judging module inquires the latest personnel position information according to the position information related to the preliminary judging information, and a shortest path is planned according to the inquired personnel position information and the position information, and the personnel position information and the shortest path are sent to the urban management platform;
after receiving the preliminary judgment information of the type of the event processing responsible by the police, the master control system sends a position request to the police platform;
the master control system receives the personnel position information returned by the police platform, the police judgment module inquires the nearest personnel position information according to the position information related to the preliminary judgment information, plans the shortest path according to the inquired personnel position information and the position information, and sends the personnel position information and the shortest path to the police platform.
Through adopting above-mentioned scheme, can carry out face identification to the image information of traffic accident type automatically, inquire including driver's license, driving license etc. driver's information, make things convenient for traffic police to handle the accident fast, can inquire recently can the sentry law enforcement personnel and plan the route for law enforcement personnel platform fast according to the position of taking place the accident simultaneously, help the platform dispatch law enforcement personnel.
Preferably, the method further comprises the following steps:
the general control system receives misjudgment information which is sent by the traffic police platform, the urban management platform and the civil police platform and is related to the preliminary judgment information, the related preliminary judgment information is called according to the misjudgment information, then image information in the received preliminary judgment information is called, and the illegal judgment model is trained according to a framed area in the image information, so that the illegal judgment model does not recognize images with the same characteristics as illegal behaviors;
after receiving a call instruction sent by the traffic police platform, the urban management platform or the police platform, the master control system calls image information corresponding to the regional subsystem according to the call instruction and transmits the image information to the traffic police platform, the urban management platform or the police platform;
after the general control system receives the image information of the frame selected area transmitted by the traffic police platform, the urban management platform or the civil police platform, preprocessing the image information of the frame selected area, and adding the preprocessed image information into a database of the illegal judgment model.
By adopting the scheme, the law enforcement personnel platform can return misjudged image information to the system, the system carries out correction training on the illegal judgment model, the law enforcement personnel platform can be actively requested to return a result when uploading the image information, if the law enforcement personnel platform returns the image information which is not judged by the system to the system, the system can automatically add the image information to the database of the illegal judgment model, so that the database can be continuously perfected.
In summary, the invention has the following beneficial effects:
1. the method has the advantages that whether illegal behaviors occur in the shot pictures or not is judged rapidly, information can be directly broadcast or sorted to platforms used by different law enforcement officers, the information can be sorted in more detail in the sorting process, the processing process is simplified, and time and labor are saved.
Drawings
Fig. 1 is an overall system block diagram of an embodiment of the present application.
Fig. 2 is a block diagram of a regional subsystem and a central control system according to an embodiment of the present application.
FIG. 3 is a block diagram of a model building block of an embodiment of the present application;
fig. 4 is a block diagram of a master control system according to a first embodiment of the present application.
Reference numerals illustrate:
1. a regional subsystem; 11. a video acquisition module; 12. a primary judgment module; 13. a primary sorting module; 14. a field alarm module; 141. a multimedia device; 15. a model building module; 2. a master control system; 21. a secondary sorting module; 22. a traffic police interface module; 23. a city pipe interface module; 24. a police interface module; 25. a traffic judgment module; 251. a traffic police feedback module; 26. a city management judging module; 261. a city management feedback module; 27. a police judgment module; 271. a civil police feedback module; 28. a model training module; 3. a traffic police platform; 4. a city management platform; 5. a civil police platform.
Detailed Description
In the first embodiment, the embodiment of the application discloses a road image model processing system with multi-platform linkage, as shown in fig. 1, the road image model processing system comprises a plurality of regional subsystems 1 arranged in different regions and a general control system 2 arranged in a cloud, wherein the regional subsystems 1 comprise a video acquisition module 11, a primary judgment module 12, a primary sorting module 13 and a field alarm module 14. The master control system 2 comprises a secondary sorting module 21, a traffic police interface module 22, a city pipe interface module 23, a policing interface module 24, a traffic judgment module 25, a traffic police feedback module 251, a city pipe judgment module 26, a city pipe feedback module 261, a policing judgment module 27, a policing feedback module 271 and a model training module 28.
As shown in fig. 1 and 2, the video acquisition module 11 includes a camera provided on the street, and the video acquisition module 11 captures image information with position information and transmits the image information to the primary judgment module 12.
As shown in fig. 2 and 3, the model construction module 15 constructs an illegal judgment model, the model construction module 15 invokes the image information of the video acquisition module 11 and divides the image information into a first training sample and a second training sample, constructs a first monitoring model, and the first monitoring model comprises two different classes of classifiers, wherein the two classifiers comprise a support vector machine classifier and a bayesian classifier, and selects the first training sample, and the first training sample comprises a first classification sample of a man-vehicle road and a second classification sample of an unmanned vehicle road. The road with the pedestrians and/or vehicles is a road with the pedestrians and/or vehicles, the road with the unmanned vehicles is a road without the pedestrians or vehicles, the first classification sample and the second classification sample both comprise identification data of the corresponding road, and the identification data comprise road safety identifications and traffic rule identifications. And respectively inputting the first classification sample and the second classification sample into two classifiers for training. The classification results of the two classifiers are used as prediction results of the first monitoring model. And inputting the identification data into a first monitoring model to obtain a first monitoring result output by the first monitoring model. If the first monitoring result shows that pedestrians and/or vehicles exist in the target road, acquiring image data of the target road; constructing a second monitoring model, wherein the second monitoring model comprises a plurality of neural networks; the plurality of neural networks includes a VGG16 neural network and a U-Net neural network. The second monitoring model takes the U-Net neural network as a main body of the second monitoring model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through imageNet; the process of training the second monitoring model includes: and selecting a second training sample, wherein the second training sample comprises image data of a road of a person or a vehicle. The man-vehicle road is a road with pedestrians and/or vehicles; and inputting a second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model. Inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model, wherein the second monitoring result is various illegal behaviors; the model construction module 15 transmits the trained illicit decision model to the primary decision module 12. The primary judgment module 12 presets an illegal judgment model, judges whether illegal behaviors exist in the image information through the illegal judgment model, selects the area with the illegal behaviors in the image information, judges the type of the illegal behaviors, associates the type of the illegal behaviors with the image information to form primary judgment information, and sends the primary judgment information to the primary sorting module 13.
As shown in fig. 2, the primary sorting module 13 is preset with a behavior type table, the behavior type table includes a plurality of illegal behaviors and a plurality of processing types, each of the illegal behaviors is associated with one processing type, the processing types include a public warning, a traffic violation, a traffic accident, a city management responsible event and a civil police responsible event, the primary sorting module 13 sorts the preliminary judgment information into the corresponding processing types according to the behavior type table, the primary sorting module 13 sends the preliminary judgment information under the public warning processing type to the on-site warning module, and other preliminary judgment information to the secondary sorting module 21.
As shown in fig. 2, the field alarm module 14 includes a multimedia device 141 disposed near the regional subsystem 1, and the multimedia device 141 may be a liquid crystal screen, an audio device, or the like. The on-site alarm module 14 receives the preliminary judgment information, converts the type of illegal and illegal behaviors in the preliminary judgment information into text information, and plays the text information and the image information in the preliminary judgment information through the multimedia device 141. The on-site alarm module 14 presets various voice warnings, image areas and text areas on the multimedia device 141, arranges the received image information and text information in the image areas and the text areas, and the on-site alarm module 14 selects the voice warnings according to the type of illegal and illegal behaviors in the preliminary judgment information and plays the voice warnings through the multimedia device 141. The system is capable of playing image information under the verbal warning type in the form of images, text and speech for alerting offensive personnel.
As shown in fig. 2, the secondary sorting module 21 receives the preliminary judgment information, obtains the position information corresponding to the preliminary judgment information and corresponding to the preliminary judgment information, combines the preliminary judgment information of the traffic violation processing type and the traffic accident processing type with the position information, sends the combined information to the traffic police interface module 22, combines the preliminary judgment information of the urban management responsible event processing type with the position information, sends the combined information to the urban pipe interface module 23, and sends the combined information of the preliminary judgment information of the police responsible event processing type to the police interface module 24.
As shown in fig. 2, the traffic police interface module 22 is connected to the traffic police platform 3, and sends the received preliminary judgment information to the traffic police platform 3. The city pipe interface module 23 is connected to the city pipe platform 4 and transmits the received preliminary judgment information to the city pipe platform 4. The police interface module 24 is connected with the police platform 5 and sends the received preliminary judgment information to the police platform 5.
As shown in fig. 2, the traffic judgment module 25 receives the preliminary judgment information under the traffic accident handling type output by the secondary sorting module 21, calls out the image information in the preliminary judgment information, identifies the face information of the frame selection area in the image information through face recognition software, uploads the face information and the query request to the traffic police platform 3, receives the driver information returned by the traffic police platform 3, the traffic judgment module 25 associates the driver information with the frame selection area, sends the driver information to the traffic police interface module 22, and the traffic police interface module 22 sends the driver information and the preliminary judgment information to the traffic police platform 3 together. The system can automatically carry out face recognition on the image information of the traffic accident type, inquires the driver information including the driving license, the driving license and the like, and is convenient for traffic police to rapidly process accidents. The traffic judgment module 25 sends a position request to the traffic police platform 3, receives the personnel position information returned by the traffic police platform 3, queries the nearest personnel position information according to the position information related to the preliminary judgment information, plans the shortest path according to the queried personnel position information and the position information, sends the personnel position information and the shortest path to the traffic police interface module 22, and the traffic police interface module 22 sends the personnel position information and the shortest path to the traffic police platform 3.
As shown in fig. 2, the urban management judging module 26 receives the preliminary judging information under the type of the urban management responsible event processing output by the secondary sorting module 21, the urban management judging module 26 sends a position request to the urban management platform 4, receives the personnel position information returned by the urban management platform 4, the urban management judging module 26 inquires the nearest personnel position information according to the position information related to the preliminary judging information, plans the shortest path according to the inquired personnel position information and the position information, sends the personnel position information and the shortest path to the urban management platform 23, and the urban management platform 4 sends the personnel position information and the shortest path to the urban management platform 23.
As shown in fig. 2, the police judgment module 27 receives the preliminary judgment information under the police responsible event processing type output by the secondary sorting module 21, the police judgment module 27 sends a position request to the police platform 5, receives the personnel position information returned by the police platform 5, the police judgment module 27 inquires the nearest personnel position information according to the position information related to the preliminary judgment information, plans the shortest path according to the inquired personnel position information and the position information, sends the personnel position information and the shortest path to the police interface module 24, and the police interface module 24 sends the personnel position information and the shortest path to the police platform 5. The system can quickly inquire recently available post law enforcement personnel and plan a route for the law enforcement personnel platform according to the accident position, and help the platform dispatch the law enforcement personnel.
As shown in fig. 2 and fig. 4, the traffic police interface module 22 receives a call instruction of the traffic police platform 3, calls image information corresponding to the regional subsystem 1 according to the call instruction, and transmits the image information to the traffic police platform 3. The metro interface module 23 receives the call instruction of the metro platform 4, calls the image information corresponding to the regional subsystem 1 according to the call instruction, and transmits the image information to the metro platform 4. The police interface module 24 receives the call instruction of the police platform 5, calls the image information corresponding to the regional subsystem 1 according to the call instruction, and transmits the image information to the police platform 5. The traffic police feedback module 251 is connected to the traffic police platform 3, receives the misjudgment information associated with the preliminary judgment information sent by the traffic police platform 3, and sends the preliminary judgment information associated with the misjudgment information to the model training module 28. The traffic police feedback module 251 judges whether the received preliminary judgment information is in the warning processing type, if so, the preliminary judgment information is transmitted to the on-site alarm module 14, and the on-site alarm module 14 cancels the multimedia device 141 to play the text information and the image information corresponding to the preliminary judgment information after receiving the preliminary judgment information transmitted by the traffic police feedback module 251. The traffic police feedback module 251 receives the image information after the frame selection of the area transmitted by the traffic police platform 3, and the traffic police feedback module 251 sends the image information after the frame selection of the area to the model training module 28. The city management feedback module 261 is connected with the city management platform 4, receives the misjudgment information related to the preliminary judgment information sent by the city management platform 4, and sends the preliminary judgment information related to the misjudgment information to the model training module 28. The metropolitan area feedback module 261 receives the image information after the frame selection area transmitted by the metropolitan area platform 4, and the metropolitan area feedback module 261 sends the image information after the frame selection area to the model training module 28. The police feedback module 271 is connected with the police platform 5, receives the misjudgment information associated with the preliminary judgment information sent by the police platform 5, and the police feedback module 271 sends the preliminary judgment information associated with the misjudgment information to the model training module 28. The police feedback module 271 receives the image information after the area is selected by the frame and transmitted by the police platform 5, and the police feedback module 271 sends the image information after the area is selected by the frame to the model training module 28.
As shown in fig. 2 and 4, the model training module 28 invokes the illegal judgment model of the primary judgment module 12, and the model training module 28 invokes the image information in the received primary judgment information, trains the illegal judgment model according to the framed area in the image information, so that the illegal judgment model no longer recognizes the image with the same characteristics as illegal behaviors. The model training module 28 receives the image information of the frame selected region, then preprocesses the image information of the frame selected region, and adds the preprocessed image information into the database of the illegal judgment model. The law enforcement personnel platform can return misjudgment image information to the system, and the system carries out correction training on the illegal judgment model. When uploading the image information, the system can actively request the law enforcement personnel platform to return a result, and if the law enforcement personnel platform returns the image information which is not judged by the system to the system, the system can automatically add the image information into a database of the illegal judgment model, so that the database can be continuously perfected.
The implementation principle of the multi-platform linkage road image model processing system is as follows: the system can rapidly judge whether illegal actions occur in the shot pictures, and can directly broadcast or sort information to platforms used by different law enforcement officers, so that the information can be classified in more detail in the sorting process, the processing process is simplified, and time and labor are saved.
In a second embodiment, the embodiment of the application discloses a method for processing a road image model by multi-platform linkage, which specifically comprises the following steps:
the regional subsystem 1 presets a behavior class table. A multimedia device 141 is disposed in proximity to regional subsystem 1.
The area subsystem 1 captures image information.
The regional subsystem 1 builds an illegal judgment model, invokes image information and divides the image information into a first training sample and a second training sample, builds a first monitoring model, wherein the first monitoring model comprises two different classes of classifiers, the two classifiers comprise a support vector machine classifier and a Bayesian classifier, the first training sample is selected, and the identification data comprises a road safety identification and a traffic rule identification; respectively inputting the first classification sample and the second classification sample into two classifiers for training; taking classification results of the two classifiers as prediction results of the first monitoring model; inputting the identification data into a first monitoring model to obtain a first monitoring result output by the first monitoring model; if the first monitoring result shows that pedestrians and/or vehicles exist in the target road, acquiring image data of the target road; constructing a second monitoring model, wherein the second monitoring model comprises a plurality of neural networks; the plurality of neural networks includes a VGG16 neural network and a U-Net neural network; the second monitoring model takes the U-Net neural network as a main body of the second monitoring model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through imageNet; the process of training the second monitoring model includes: selecting a second training sample, wherein the second training sample comprises image data of a man-vehicle road; inputting a second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model; and inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model, wherein the second monitoring result is various illegal behaviors.
The regional subsystem 1 judges whether the illegal and illegal behaviors exist in the image information through the illegal judging model, and selects the regional frame with the illegal and illegal behaviors in the image information, and judges the type of the illegal and illegal behaviors.
The regional subsystem 1 associates the type of the illegal and illegal actions with the image information to form preliminary judgment information.
The regional subsystem 1 classifies the preliminary judgment information into corresponding processing types according to the behavior type table.
The area subsystem 1 selects the preliminary judgment information under the warning processing type, converts the illegal behavior type in the preliminary judgment information into character information, and plays the character information and the image information in the preliminary judgment information through the multimedia device 141.
The regional subsystem 1 sends other preliminary judgment information to the master control system 2.
The general control system 2 acquires the processing type corresponding to the preliminary judgment information and the position information corresponding to the preliminary judgment information.
The primary judgment information of the traffic violation processing type and the traffic accident processing type is combined with the position information and then sent to the traffic police platform 3, the primary judgment information of the urban management responsible event processing type is combined with the position information and then sent to the urban management platform 4, and the primary judgment information of the police responsible event processing type is combined with the position information and then sent to the police platform 5.
After receiving the preliminary judgment information under the traffic accident handling type, the master control system 2 calls out the image information in the preliminary judgment information, and the face information and the query request are uploaded to the traffic police platform 3 by recognizing the face information of the frame-selected area in the image information through face recognition software.
The central control system 2 receives the driver information returned by the traffic police platform 3, the traffic judgment module 25 associates the driver information with the frame selection area, the driver information is sent to the traffic police interface module 22, and the traffic police interface module 22 sends the driver information and the preliminary judgment information to the traffic police platform 3.
The central control system 2 sends a location request to the traffic police platform 3.
The master control system 2 receives the personnel position information returned by the traffic police platform 3, plans the shortest path according to the inquired personnel position information and the position information, and sends the personnel position information and the shortest path to the traffic police platform 3.
After receiving the preliminary judgment information of the type of the event processing in charge of the urban management, the master control system 2 sends a position request to the urban management platform 4.
The central control system 2 receives the personnel position information returned by the urban management platform 4, the urban management judging module 26 inquires the latest personnel position information according to the position information related to the preliminary judging information, plans the shortest path according to the inquired personnel position information and the position information, and sends the personnel position information and the shortest path to the urban management platform 4.
After receiving the preliminary judgment information of the type of the event processing responsible by the police, the master control system 2 sends a position request to the police platform 5.
The master control system 2 receives the personnel position information returned by the police platform 5, the police judgment module 27 inquires the latest personnel position information according to the position information related to the preliminary judgment information, plans the shortest path according to the inquired personnel position information and the position information, and sends the personnel position information and the shortest path to the police platform 5.
The general control system 2 receives the misjudgment information which is sent by the traffic police platform 3, the urban management platform 4 and the civil police platform 5 and is related to the preliminary judgment information, calls out the related preliminary judgment information according to the misjudgment information, calls out the image information in the received preliminary judgment information, trains the illegal judgment model according to the framed area in the image information, and enables the illegal judgment model not to recognize the image with the same characteristics as illegal behaviors.
After receiving the call instruction sent by the traffic police platform 3, the urban management platform 4 or the civil police platform 5, the master control system 2 calls the image information corresponding to the regional subsystem 1 according to the call instruction and transmits the image information to the traffic police platform 3, the urban management platform 4 or the civil police platform 5.
After receiving the image information of the frame selected area transmitted by the traffic police platform 3, the urban management platform 4 or the civil police platform 5, the master control system 2 pre-processes the image information of the frame selected area, and adds the pre-processed image information into a database of the illegal judgment model.
The embodiments of the present invention are all preferred embodiments of the present invention, and are not intended to limit the scope of the present invention in this way, therefore: all equivalent changes in structure, shape and principle of the invention should be covered in the scope of protection of the invention.
Claims (10)
1. A multi-platform linkage road image model processing system is characterized in that: the system comprises a plurality of regional subsystems (1) arranged in different regions and a general control system (2) arranged in a cloud, wherein the regional subsystems (1) comprise a video acquisition module (11), a model construction module (15), a primary judgment module (12), a primary sorting module (13) and a field alarm module (14);
the video acquisition module (11) acquires image information with position information and transmits the image information to the primary judgment module (12);
the model construction module (15) constructs an illegal judgment model, a first monitoring model and a second monitoring model are constructed in the illegal judgment model, the first monitoring model identifies whether pedestrians and/or vehicles exist in image information, the second monitoring model identifies illegal behaviors of the pedestrians and/or vehicles on the image information, the illegal judgment model is trained, and the model construction module (15) transmits the trained illegal judgment model to the primary judgment module (12);
The primary judging module (12) receives the illegal judging model, judges whether illegal behaviors exist in the image information through the illegal judging model, boxes and selects the areas with illegal behaviors in the image information, judges the type of the illegal behaviors, associates the type of the illegal behaviors with the image information to form primary judging information, and sends the primary judging information to the primary sorting module (13);
the primary sorting module (13) sorts the primary judgment information into the processing types of public warning, traffic violation, traffic accident, urban management responsible event and civil police responsible event, sends the primary judgment information under the public warning processing type to the on-site alarm module (14) for processing, and sends other primary judgment information to the master control system (2);
the on-site alarm module (14) plays the received preliminary judgment information;
and the master control system (2) sends the received preliminary judgment information to the traffic police platform (3), the urban management platform (4) or the civil police platform (5) according to the processing type.
2. The multi-platform linked road image model processing system of claim 1, wherein: the model construction module (15) calls the image information of the video acquisition module (11) and divides the image information into a first training sample and a second training sample, a first monitoring model is constructed, the first monitoring model comprises two classifiers of different types, wherein the two classifiers comprise a support vector machine classifier and a Bayesian classifier, the first training sample is selected, and the first training sample comprises a first classification sample of a man-vehicle road and a second classification sample of an unmanned vehicle road; the road with the pedestrians and/or vehicles is a road with the pedestrians and/or vehicles, the road with the unmanned vehicles is a road without the pedestrians or vehicles, the first classification sample and the second classification sample both comprise identification data of the corresponding road, and the identification data comprise road safety identifications and traffic rule identifications; respectively inputting the first classification sample and the second classification sample into two classifiers for training; taking classification results of the two classifiers as prediction results of the first monitoring model; inputting the identification data into a first monitoring model to obtain a first monitoring result output by the first monitoring model; if the first monitoring result shows that pedestrians and/or vehicles exist in the target road, acquiring image data of the target road; constructing a second monitoring model, wherein the second monitoring model comprises a plurality of neural networks; the plurality of neural networks includes a VGG16 neural network and a U-Net neural network; the second monitoring model takes the U-Net neural network as a main body of the second monitoring model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through imageNet; the process of training the second monitoring model includes: selecting a second training sample, wherein the second training sample comprises image data of a man-vehicle road; the man-vehicle road is a road with pedestrians and/or vehicles; inputting a second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model; inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model, wherein the second monitoring result is various illegal behaviors; the model construction module (15) transmits the trained illegal judgment model to the primary judgment module (12);
The primary sorting module (13) is preset with a behavior type table, the behavior type table comprises a plurality of illegal behaviors and a plurality of processing types, each illegal behavior is associated with one processing type, the processing types comprise a public warning, a traffic violation, a traffic accident, a urban management responsible event and a civil police responsible event, the primary sorting module (13) sorts primary judgment information into corresponding processing types according to the behavior type table, the primary sorting module (13) sends primary judgment information under the public warning processing types to the on-site alarm module (14), and other primary judgment information is sent to the master control system (2);
the on-site alarm module (14) comprises a multimedia device (141) arranged near the regional subsystem (1), the on-site alarm module (14) receives the preliminary judgment information and then converts the illegal behavior type in the preliminary judgment information into character information, and the character information and the image information in the preliminary judgment information are played through the multimedia device (141);
the master control system (2) comprises a secondary sorting module (21), a traffic police interface module (22), a city pipe interface module (23) and a civil police interface module (24);
the secondary sorting module (21) receives the preliminary judgment information and then acquires the processing type corresponding to the preliminary judgment information and the position information corresponding to the preliminary judgment information, combines the preliminary judgment information of the traffic violation processing type and the traffic accident processing type with the position information and then sends the combined information to the traffic police interface module (22), combines the preliminary judgment information of the urban management responsible event processing type with the position information and then sends the combined information to the urban pipe interface module (23), and combines the preliminary judgment information of the police responsible event processing type with the position information and then sends the combined information to the police interface module (24);
The traffic police interface module (22) is connected with the traffic police platform (3) and sends the received preliminary judgment information to the traffic police platform (3);
the urban pipe interface module (23) is connected with the urban pipe platform (4) and sends the received preliminary judgment information to the urban pipe platform (4);
the police interface module (24) is connected with the police platform (5) and sends the received preliminary judgment information to the police platform (5).
3. The multi-platform linked road image model processing system of claim 2, wherein: the general control system (2) further comprises a traffic judgment module (25), a city management judgment module (26) and a police judgment module (27);
the traffic judgment module (25) receives the preliminary judgment information under the traffic accident handling type output by the secondary sorting module (21), calls out the image information in the preliminary judgment information, identifies the face information of a frame selection area in the image information through face recognition software, uploads the face information and a query request to the traffic police platform (3), receives the driver information returned by the traffic police platform (3), the traffic judgment module (25) associates the driver information with the frame selection area, sends the driver information to the traffic police interface module (22), and the traffic police interface module (22) sends the driver information and the preliminary judgment information to the traffic police platform (3);
The traffic judgment module (25) sends a position request to the traffic police platform (3), receives personnel position information returned by the traffic police platform (3), inquires the nearest personnel position information according to the position information related to the preliminary judgment information, plans a shortest path according to the inquired personnel position information and the position information, sends the personnel position information and the shortest path to the traffic police interface module (22), and the traffic police interface module (22) sends the personnel position information and the shortest path to the traffic police platform (3);
the urban management judgment module (26) receives preliminary judgment information under the type of event processing of the urban management responsible output by the secondary sorting module (21), the urban management judgment module (26) sends a position request to the urban management platform (4), receives personnel position information returned by the urban management platform (4), inquires the nearest personnel position information according to the position information related to the preliminary judgment information, plans a shortest path according to the inquired personnel position information and the position information, and sends the personnel position information and the shortest path to the urban management platform (4), and the urban management module (23) sends the personnel position information and the shortest path to the urban management platform (4);
the police judgment module (27) receives preliminary judgment information under the police responsible event processing type output by the secondary sorting module (21), the police judgment module (27) sends a position request to the police platform (5), receives personnel position information returned by the police platform (5), the police judgment module (27) inquires the nearest personnel position information according to the position information related to the preliminary judgment information, plans a shortest path according to the inquired personnel position information and the position information, sends the personnel position information and the shortest path to the police interface module (24), and the police interface module (24) sends the personnel position information and the shortest path to the police platform (5).
4. The multi-platform linked road image model processing system of claim 2, wherein: the general control system (2) also comprises a traffic police feedback module (251), a city management feedback module (261), a civil police feedback module (271) and a model training module (28);
the traffic police feedback module (251) is connected with the traffic police platform (3), receives the misjudgment information related to the preliminary judgment information sent by the traffic police platform (3), and sends the preliminary judgment information related to the misjudgment information to the model training module (28);
the urban management feedback module (261) is connected with the urban management platform (4), and is used for receiving the misjudgment information related to the preliminary judgment information sent by the urban management platform (4), and the urban management feedback module (261) is used for sending the preliminary judgment information related to the misjudgment information to the model training module (28);
the police feedback module (271) is connected with the police platform (5), receives the misjudgment information related to the preliminary judgment information sent by the police platform (5), and sends the preliminary judgment information related to the misjudgment information to the model training module (28);
the model training module (28) calls the illegal judgment model of the primary judgment module (12), the model training module (28) calls the image information in the received primary judgment information, and trains the illegal judgment model according to the framed area in the image information, so that the illegal judgment model does not recognize the image with the same characteristics as illegal behaviors.
5. The multi-platform linked road image model processing system of claim 4, wherein: the traffic police feedback module (251) judges whether the received preliminary judgment information is in the warning processing type, if so, the preliminary judgment information is transmitted to the on-site alarm module (14), and the on-site alarm module (14) cancels the multimedia equipment (141) to play the text information and the image information corresponding to the preliminary judgment information after receiving the preliminary judgment information transmitted by the traffic police feedback module (251).
6. The multi-platform linked road image model processing system of claim 4, wherein: the traffic police interface module (22) receives a call instruction of the traffic police platform (3), calls image information corresponding to the regional subsystem (1) according to the call instruction and transmits the image information to the traffic police platform (3);
the urban pipe interface module (23) receives a calling instruction of the urban pipe platform (4), calls image information corresponding to the regional subsystem (1) according to the calling instruction and transmits the image information to the urban pipe platform (4);
the police interface module (24) receives a calling instruction of the police platform (5), calls image information corresponding to the regional subsystem (1) according to the calling instruction and transmits the image information to the police platform (5);
The traffic police feedback module (251) receives the image information of the frame-selected region transmitted by the traffic police platform (3), and the traffic police feedback module (251) sends the image information of the frame-selected region to the model training module (28);
the urban management feedback module (261) receives the image information of the frame-selected region transmitted by the urban management platform (4), and the urban management feedback module (261) sends the image information of the frame-selected region to the model training module (28);
the police feedback module (271) receives the image information after the frame selection of the area transmitted by the police platform (5), and the police feedback module (271) sends the image information after the frame selection of the area to the model training module (28);
and the model training module (28) receives the image information after the frame selection of the area, then preprocesses the image information after the frame selection of the area, and adds the preprocessed image information into a database of the illegal judgment model.
7. The multi-platform linked road image model processing system of claim 2, wherein: the on-site alarm module (14) presets various voice warnings, image areas and text areas on the multimedia equipment (141), arranges the received image information and text information in the image areas and the text areas, and the on-site alarm module (14) selects the voice warnings according to the type of illegal and illegal behaviors in the preliminary judgment information and plays the voice warnings through the multimedia equipment (141).
8. A multi-platform linked road image model processing method, comprising the road image model processing system according to any one of claims 1-7, characterized by comprising the steps of:
the regional subsystem (1) presets a behavior type table;
the regional subsystem (1) shoots image information;
the regional subsystem (1) builds an illegal judgment model, invokes image information and divides the image information into a first training sample and a second training sample, builds a first monitoring model, wherein the first monitoring model comprises two classifiers of different types, the two classifiers comprise a support vector machine classifier and a Bayesian classifier, the first training sample is selected, and the identification data comprise road safety identifications and traffic rule identifications; respectively inputting the first classification sample and the second classification sample into two classifiers for training; taking classification results of the two classifiers as prediction results of the first monitoring model; inputting the identification data into a first monitoring model to obtain a first monitoring result output by the first monitoring model; if the first monitoring result shows that pedestrians and/or vehicles exist in the target road, acquiring image data of the target road; constructing a second monitoring model, wherein the second monitoring model comprises a plurality of neural networks; the plurality of neural networks includes a VGG16 neural network and a U-Net neural network; the second monitoring model takes the U-Net neural network as a main body of the second monitoring model, replaces an encoder of the U-Net neural network with the VGG16 neural network, and initializes the weight of the VGG16 neural network through imageNet; the process of training the second monitoring model includes: selecting a second training sample, wherein the second training sample comprises image data of a man-vehicle road; inputting a second training sample into the second monitoring model for training to obtain a prediction result of the second monitoring model; inputting the image data into a second monitoring model to obtain a second monitoring result output by the second monitoring model, wherein the second monitoring result is various illegal behaviors;
The regional subsystem (1) judges whether the illegal and illegal behaviors exist in the image information through the illegal judging model, and selects a regional frame with the illegal and illegal behaviors in the image information, and judges the type of the illegal and illegal behaviors;
the regional subsystem (1) associates the type of the illegal and illegal behaviors with the image information to form preliminary judgment information;
the regional subsystem (1) classifies the preliminary judgment information into corresponding processing types according to the behavior type table;
-providing a multimedia device (141) in proximity of the regional subsystem (1);
the regional subsystem (1) selects the preliminary judgment information under the warning processing type, converts the illegal behavior type in the preliminary judgment information into character information, and plays the character information and the image information in the preliminary judgment information through the multimedia equipment (141);
the regional subsystem (1) sends other preliminary judgment information to the master control system (2);
the general control system (2) acquires the processing type corresponding to the preliminary judgment information and the position information corresponding to the preliminary judgment information;
the primary judgment information of the traffic violation processing type and the traffic accident processing type is combined with the position information and then sent to the traffic police platform (3), the primary judgment information of the urban management responsible event processing type is combined with the position information and then sent to the urban management platform (4), and the primary judgment information of the police responsible event processing type is combined with the position information and then sent to the police platform (5).
9. The method for processing the multi-platform linked road image model according to claim 8, further comprising the steps of:
after receiving the preliminary judgment information under the traffic accident handling type, the master control system (2) calls out the image information in the preliminary judgment information, and the face information and the query request are uploaded to the traffic police platform (3) by identifying the face information of a frame-selected area in the image information through face recognition software;
the general control system (2) receives the driver information returned by the traffic police platform (3), the traffic judgment module (25) correlates the driver information with the frame selection area, the driver information is sent to the traffic police interface module (22), and the traffic police interface module (22) sends the driver information and the preliminary judgment information to the traffic police platform (3);
the master control system (2) sends a position request to the traffic police platform (3);
the general control system (2) receives personnel position information returned by the traffic police platform (3), plans a shortest path according to the inquired personnel position information and the position information, and sends the personnel position information and the shortest path to the traffic police platform (3);
after receiving the preliminary judgment information of the type of the event processing in charge of the urban management, the master control system (2) sends a position request to the urban management platform (4);
The general control system (2) receives the personnel position information returned by the urban management platform (4), the urban management judging module (26) inquires the latest personnel position information according to the position information related to the preliminary judging information, and the shortest path is planned according to the inquired personnel position information and the position information, and the personnel position information and the shortest path are sent to the urban management platform (4);
after receiving the preliminary judgment information of the type of the event processing responsible by the police, the master control system (2) sends a position request to the police platform (5);
the general control system (2) receives personnel position information returned by the police platform (5), the police judgment module (27) inquires the nearest personnel position information according to the position information related to the preliminary judgment information, and the personnel position information and the shortest path are sent to the police platform (5) according to the inquired personnel position information and the position information to plan the shortest path.
10. The method for processing the multi-platform linked road image model according to claim 8, further comprising the steps of:
the general control system (2) receives misjudgment information which is transmitted by the traffic police platform (3), the urban management platform (4) and the civil police platform (5) and is related to the preliminary judgment information, the related preliminary judgment information is called according to the misjudgment information, then image information in the received preliminary judgment information is called, and the illegal judgment model is trained according to a framed area in the image information, so that the illegal judgment model does not recognize images with the same characteristics as illegal behaviors;
After the general control system (2) receives a call instruction sent by the traffic police platform (3), the urban management platform (4) or the civil police platform (5), the image information corresponding to the regional subsystem (1) is called according to the call instruction and is transmitted to the traffic police platform (3), the urban management platform (4) or the civil police platform (5);
after the general control system (2) receives the image information of the frame selected area transmitted by the traffic police platform (3), the urban management platform (4) or the civil police platform (5), preprocessing the image information of the frame selected area, and adding the preprocessed image information into a database of the illegal judgment model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410145468.3A CN117671572A (en) | 2024-02-02 | 2024-02-02 | Multi-platform linkage road image model processing system and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410145468.3A CN117671572A (en) | 2024-02-02 | 2024-02-02 | Multi-platform linkage road image model processing system and method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117671572A true CN117671572A (en) | 2024-03-08 |
Family
ID=90086694
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202410145468.3A Pending CN117671572A (en) | 2024-02-02 | 2024-02-02 | Multi-platform linkage road image model processing system and method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117671572A (en) |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102263937A (en) * | 2011-07-26 | 2011-11-30 | 华南理工大学 | Driver's driving behavior monitoring device and monitoring method based on video detection |
CN109309673A (en) * | 2018-09-18 | 2019-02-05 | 南京方恒信息技术有限公司 | A kind of DNS private communication channel detection method neural network based |
CN110189524A (en) * | 2019-06-19 | 2019-08-30 | 刘先进 | A kind of active violation recognition methods, system and electronic equipment violating the regulations |
CN110363426A (en) * | 2019-07-15 | 2019-10-22 | 软通动力信息技术有限公司 | A kind of long-range enforcement approach of cloud platform, device, server and storage medium |
CN112381859A (en) * | 2020-11-20 | 2021-02-19 | 公安部第三研究所 | System, method, device, processor and storage medium for realizing intelligent analysis, identification and processing for video image data |
CN112988212A (en) * | 2021-03-24 | 2021-06-18 | 厦门吉比特网络技术股份有限公司 | Method, apparatus, system and storage medium for online incremental update of neural network model |
CN113919655A (en) * | 2021-09-17 | 2022-01-11 | 深圳技术大学 | Law enforcement personnel scheduling method, system, computer device and storage medium |
CN114187625A (en) * | 2021-11-18 | 2022-03-15 | 天津市国瑞数码安全系统股份有限公司 | Video detection method based on video source automatic detection technology |
CN115601717A (en) * | 2022-10-19 | 2023-01-13 | 中诚华隆计算机技术有限公司(Cn) | Deep learning-based traffic violation classification detection method and SoC chip |
CN116109818A (en) * | 2023-04-11 | 2023-05-12 | 成都中医药大学 | Traditional Chinese medicine pulse condition distinguishing system, method and device based on facial video |
-
2024
- 2024-02-02 CN CN202410145468.3A patent/CN117671572A/en active Pending
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102263937A (en) * | 2011-07-26 | 2011-11-30 | 华南理工大学 | Driver's driving behavior monitoring device and monitoring method based on video detection |
CN109309673A (en) * | 2018-09-18 | 2019-02-05 | 南京方恒信息技术有限公司 | A kind of DNS private communication channel detection method neural network based |
CN110189524A (en) * | 2019-06-19 | 2019-08-30 | 刘先进 | A kind of active violation recognition methods, system and electronic equipment violating the regulations |
CN110363426A (en) * | 2019-07-15 | 2019-10-22 | 软通动力信息技术有限公司 | A kind of long-range enforcement approach of cloud platform, device, server and storage medium |
CN112381859A (en) * | 2020-11-20 | 2021-02-19 | 公安部第三研究所 | System, method, device, processor and storage medium for realizing intelligent analysis, identification and processing for video image data |
CN112988212A (en) * | 2021-03-24 | 2021-06-18 | 厦门吉比特网络技术股份有限公司 | Method, apparatus, system and storage medium for online incremental update of neural network model |
CN113919655A (en) * | 2021-09-17 | 2022-01-11 | 深圳技术大学 | Law enforcement personnel scheduling method, system, computer device and storage medium |
CN114187625A (en) * | 2021-11-18 | 2022-03-15 | 天津市国瑞数码安全系统股份有限公司 | Video detection method based on video source automatic detection technology |
CN115601717A (en) * | 2022-10-19 | 2023-01-13 | 中诚华隆计算机技术有限公司(Cn) | Deep learning-based traffic violation classification detection method and SoC chip |
CN116109818A (en) * | 2023-04-11 | 2023-05-12 | 成都中医药大学 | Traditional Chinese medicine pulse condition distinguishing system, method and device based on facial video |
Non-Patent Citations (2)
Title |
---|
TURBO_COME: "Vgg16+Unet介绍", pages 1 - 2, Retrieved from the Internet <URL:https://blog.csdn.net/turbo_come/article/details/104602250> * |
我是一条小青蛇: "Keras 手动搭建 VGG 卷积神经网络识别 ImageNet 1000 种常见分类", pages 4, Retrieved from the Internet <URL:https://cloud.tencent.com/developer/article/1525554> * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109919140B (en) | Automatic determination method, system, equipment and storage medium for vehicle collision accident responsibility | |
CN109800633B (en) | Non-motor vehicle traffic violation judgment method and device and electronic equipment | |
US20120148092A1 (en) | Automatic traffic violation detection system and method of the same | |
CN110718058B (en) | Expressway emergency lane occupation detection and disposal method based on active safety terminal | |
CN113724416A (en) | Vehicle-mounted video processing method and device based on vehicle-road cooperative system and storage medium | |
US20050073436A1 (en) | Method and system for alerting a patrol officer of a wanted vehicle | |
CN110321804A (en) | One kind is combined based on face and electronic Car license recognition and carries out identity authentication method | |
CN102142193A (en) | Automatic detection system and method for illegal traffic behaviors | |
CN112381859B (en) | System, method, device, processor and storage medium for implementing intelligent analysis and identification processing for video image data | |
CN109584133A (en) | Net about vehicle method of controlling security, device, computer equipment and storage medium | |
CN112802344A (en) | Vehicle-mounted intelligent networking real-time traffic violation monitoring device and system | |
CN110395252A (en) | A kind of traffic light intersection automatic control vehicle pass-through system and method | |
CN114394100B (en) | Unmanned patrol car control system and unmanned car | |
CN115035491A (en) | Driving behavior road condition early warning method based on federal learning | |
CN108257400A (en) | The intelligent management system and its management method of a kind of vehicle drive | |
CN109671270A (en) | Driving accident processing method and processing device, storage medium | |
KR101066081B1 (en) | Smart information detection system mounted on the vehicle and smart information detection method using the same | |
CN113870551B (en) | Road side monitoring system capable of identifying dangerous and non-dangerous driving behaviors | |
US20220156504A1 (en) | Audio/video capturing device, vehicle mounted device, control centre system, computer program and method | |
KR102309106B1 (en) | Situation linkage type image analysis device | |
CN112907979B (en) | System and method for tracking illegal running track of motor vehicle based on multiple cameras | |
CN110807929A (en) | Traffic police auxiliary command information analysis and research and judgment system | |
CN109003457B (en) | Method and device for recording behaviors of multiple motor vehicles illegally occupying emergency lane | |
CN117894187A (en) | Highway intellectual detection system early warning system | |
CN109191856A (en) | The method of vehicle tracking system and tracking vehicle based on big data |
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 |