CN112133094A - Road intersection full-factor health diagnosis system based on deep learning technology - Google Patents
Road intersection full-factor health diagnosis system based on deep learning technology Download PDFInfo
- Publication number
- CN112133094A CN112133094A CN202011022329.XA CN202011022329A CN112133094A CN 112133094 A CN112133094 A CN 112133094A CN 202011022329 A CN202011022329 A CN 202011022329A CN 112133094 A CN112133094 A CN 112133094A
- Authority
- CN
- China
- Prior art keywords
- traffic
- intersection
- model
- module
- deep learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a road intersection full-factor health diagnosis system based on a deep learning technology, which comprises the following components: the model establishing module is used for establishing each frame of image input as a video and outputting a deep learning model of a traffic object in each frame of image; the video acquisition module is used for acquiring the video to be detected corresponding to each road junction; the track reconstruction module is used for reconstructing the traffic track of the traffic object based on the video to be detected and the deep learning model; the analysis and recognition module is used for recognizing lane flow and traffic events according to the traffic track of the traffic object and a preset traffic scene; the diagnosis evaluation module is used for inputting the lane flow and the traffic incident into a preset study and judgment model to obtain a diagnosis result of the intersection output by the study and judgment model; and the intersection determining module is used for determining the intersection to be controlled preferentially based on the diagnosis result corresponding to each intersection. The invention realizes the diagnosis of the health condition of the intersection through a plurality of intersection indexes, and is convenient for the overall treatment.
Description
Technical Field
The invention relates to the field of intersection diagnosis, in particular to a road intersection full-factor health diagnosis system based on a deep learning technology.
Background
Deep Learning (DL) is a new research direction in the field of Machine Learning (ML), which is introduced to make Machine Learning closer to the original target, Artificial Intelligence (AI), which is the concept of research on Artificial neural networks.
Deep learning is the intrinsic law and expression level of the learning sample data, and the information obtained in the learning process is very helpful for the interpretation of data such as characters, images and sounds. The final aim of the method is to enable the machine to have the analysis and learning capability like a human, and to recognize data such as characters, images and sounds. Deep learning is a complex machine learning algorithm, and achieves the effect in speech and image recognition far exceeding the prior related art.
In the traffic field, because the data volume is huge, how to use the data to realize the whole element health evaluation of the intersection effect becomes a problem which needs to be solved urgently at the present stage.
Disclosure of Invention
The invention aims to provide a road intersection all-element health diagnosis system based on a deep learning technology, which realizes the diagnosis of the health condition of an intersection through a plurality of intersection indexes and facilitates the overall treatment.
In order to achieve the above object, the present invention provides a deep learning technology-based road intersection all-element health diagnosis system, including: the model establishing module is used for establishing each frame of image input as a video and outputting a deep learning model of a traffic object in each frame of image; the video acquisition module is used for acquiring the video to be detected corresponding to each road junction; the track reconstruction module is used for reconstructing the traffic track of the traffic object based on the video to be detected and the deep learning model; the analysis and recognition module is used for recognizing lane flow and traffic events according to the traffic track of the traffic object and a preset traffic scene; the diagnosis evaluation module is used for inputting the lane flow and the traffic events into a preset study and judgment model to obtain the diagnosis result of the intersection output by the study and judgment model; and the intersection determining module is used for determining the intersection to be controlled preferentially based on the diagnosis result corresponding to each intersection.
Preferably, the model building module comprises: the historical data acquisition submodule is used for acquiring multi-frame historical images and traffic objects existing in the multi-frame historical images; the model building submodule is used for building a basic model which takes the multi-frame images as input and takes the traffic objects as output; and the training submodule is used for training the basic model through the multi-frame historical images and the traffic objects existing in the multi-frame historical images to obtain a deep learning model.
Preferably, the analysis recognition module includes: the scene establishing sub-module is used for establishing various traffic scenes, wherein each traffic scene is configured to contain the traffic track of a specific traffic object and the corresponding characteristic information; and the identification submodule is used for identifying lane flow and traffic events according to the reconstructed traffic track of the traffic object and the acquired characteristic information related to the traffic object.
Preferably, the identifying sub-module for identifying lane traffic and traffic events according to the reconstructed traffic trajectory of the traffic object and the feature information related to the traffic object comprises: when the acquired characteristic information related to the traffic object comprises the traffic track of the specific traffic object of any traffic scene and all the corresponding characteristic information thereof, identifying the traffic event corresponding to any traffic scene; and identifying lane traffic based on the reconstructed traffic trajectory of the traffic object.
Preferably, the traffic scene comprises any one of: the method comprises the following steps of an unreasonable signal lamp timing scene, an intersection knotting overflow scene and an intersection man-machine non-conflict scene.
Preferably, the diagnostic evaluation module comprises: the model acquisition submodule is used for acquiring a trained judging model, wherein the input of the judging model is the historical lane flow and the historical traffic incident of the intersection, and the output of the judging model is the diagnosis result of the intersection; and the result output submodule is used for inputting the lane flow and the traffic incident into the judging model so as to obtain the diagnosis result of the intersection output by the judging model.
Preferably, the intersection determination module includes: a result comparison submodule for comparing intersection scores shown by the diagnosis results of the respective intersections; and the intersection determining submodule is used for determining the intersection with the lowest score as the priority governing intersection.
According to the technical scheme, the deep learning model is established by the model establishing module and used for identifying the traffic objects in the images, the video acquiring module is used for acquiring the videos to be detected corresponding to all intersections, the track reconstructing module is used for reconstructing the traffic tracks of the traffic objects according to the identification results of the traffic objects output by the model and the relation between the multi-frame images, the lane flow and the traffic events are identified through the analyzing and identifying module based on the traffic tracks and the preset traffic scenes, the output of the diagnosis results is finally realized through the diagnosis and evaluation module, and the intersection which needs to be treated in the top priority is finally determined. The whole process realizes the diagnosis of the health condition of the intersection through a plurality of intersection indexes, and finally realizes the overall treatment.
Additional features and advantages of the invention will be set forth in the detailed description which follows.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a block diagram illustrating a deep learning technique based intersection all-element health diagnostic system of the present invention;
FIG. 2 is a block diagram illustrating the model building module of the present invention;
FIG. 3 is a block diagram illustrating an analysis recognition module of the present invention; and
FIG. 4 is a block diagram illustrating the diagnostic evaluation module of the present invention.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the present invention, are given by way of illustration and explanation only, not limitation.
Fig. 1 is a block diagram of a deep learning technology-based intersection all-element health diagnosis system according to the present invention, and as shown in fig. 1, the deep learning technology-based intersection all-element health diagnosis system includes: the model establishing module is used for establishing a deep learning model which is input into each frame of image of a video and output into a traffic object in each frame of image, and the model establishing module is used for establishing the deep learning model which can be trained through a plurality of historical data; the system comprises a video acquisition module, a video processing module and a video processing module, wherein the video acquisition module is used for acquiring videos to be detected corresponding to each intersection; the track reconstruction module is used for reconstructing the traffic track of the traffic object based on the video to be detected and the deep learning model, and the track reconstruction module is used for reconstructing the traffic track of the traffic object, wherein the reconstruction process is a splicing process of each frame of image; the analysis and identification module is used for identifying lane flow and traffic events according to the traffic track of the traffic object and a preset traffic scene, wherein the lane flow is used for reflecting the traffic jam condition, the traffic jam is easy to cause as the flow is larger, and the corresponding green light time is increased; the diagnosis evaluation module is used for inputting the lane flow and the traffic incident into a preset study and judgment model to obtain a diagnosis result of the intersection output by the study and judgment model, wherein the output of the study and judgment model is a diagnosis result which is used for reflecting the health score of the intersection; and the intersection determining module is used for determining the intersection to be controlled preferentially based on the diagnosis result corresponding to each intersection.
Preferably, the model building module may include: the historical data acquisition submodule is used for acquiring multi-frame historical images and traffic objects existing in the historical images, the historical data can be directly acquired from the storage of the monitoring system, and the traffic objects are individuals such as automobiles, non-motor vehicles, people and the like in the images; the model building submodule is used for building a basic model which takes multi-frame images as input and takes a traffic object as output, and is used for building a basic model which is not trained yet and can not be directly used; and the training submodule is used for training the basic model through the multi-frame historical images and the traffic objects in the multi-frame historical images to obtain a deep learning model, and the training submodule is used for training through the training submodule and then obtaining the deep learning model.
Preferably, the analysis recognition module may include: the scene establishing sub-module is used for establishing various traffic scenes, wherein each traffic scene is configured to contain a traffic track of a specific traffic object and corresponding characteristic information, the scene establishing sub-module is used for establishing various traffic scenes, and the specific traffic scenes comprise: the traffic light timing is unreasonable, the crossing is knotted and overflows the scene, the crossing man-machine is not conflicted the scene, every scene includes many characteristic information, for example when the timing is unreasonable, the flow of some lane is often very big and especially congested, and there is no vehicle in the direction of other lanes, or there is abundance in green light, wherein man, machine, non-person, motor vehicle and non-motor vehicle, wherein three should not be in a lane, but because planning and jam of the road, cause and mix together, the serious "conflict", namely there is crossing of the route when the motor vehicle, person or non-motor vehicle is run, very apt to cause the traffic accident; and the identification submodule is used for identifying lane flow and traffic events according to the reconstructed traffic track of the traffic object and the acquired characteristic information related to the traffic object.
Preferably, the identifying sub-module for identifying lane traffic and traffic events according to the reconstructed traffic trajectory of the traffic object and the feature information related to the traffic object comprises: when the acquired characteristic information related to the traffic object comprises the traffic track of the specific traffic object of any traffic scene and all the corresponding characteristic information, for example, the battery car runs to a motorway and stops suddenly, the motor car runs to the motorway and stops suddenly, if the states exceed 2min, a traffic accident is determined to exist, at the moment, the characteristic information related to the traffic object comprises all the characteristics of a man-machine non-conflict scene at an intersection, the traffic event corresponding to any traffic scene is identified, and the traffic event is a collision; and identifying lane traffic based on the reconstructed traffic trajectory of the traffic object.
Preferably, the diagnostic evaluation module comprises: the model acquisition submodule is used for acquiring a trained judging model, wherein the input of the judging model is the historical lane flow and the historical traffic incident of the intersection, and the output of the judging model is the diagnosis result of the intersection; and the result output submodule is used for inputting the lane flow and the traffic incident into the judging model so as to obtain the diagnosis result of the intersection output by the judging model. The study model will score the lane flows and traffic events traffic lane openings based on table 1 below. As can be seen from table 1, different traffic events and different lane flows correspond to different diagnosis results, the diagnosis results are used for reflecting traffic health conditions, and the lower the score is, the more urgent treatment is needed.
TABLE 1
Preferably, the intersection determination module includes: a result comparison submodule for comparing intersection scores shown by the diagnosis results of the respective intersections, as shown in table 1; and an intersection determining submodule for determining that the intersection with the lowest score is determined as the priority treatment intersection, if the intersection is the condition shown in table 1, the intersection which is continuously treated is the intersection 6, the condition is only a part of the characteristics of the invention, the diagnosis result can be influenced by the difference of traffic flow and the difference of traffic events, and in a simple way, the diagnosis result is in inverse proportion to the traffic flow of the lane under the condition of determining the traffic events.
The above are merely examples of the present application and are not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.
Claims (7)
1. The road intersection all-element health diagnosis system based on the deep learning technology is characterized by comprising the following components:
the model establishing module is used for establishing each frame of image input as a video and outputting a deep learning model of a traffic object in each frame of image;
the video acquisition module is used for acquiring the video to be detected corresponding to each road junction;
the track reconstruction module is used for reconstructing the traffic track of the traffic object based on the video to be detected and the deep learning model;
the analysis and recognition module is used for recognizing lane flow and traffic events according to the traffic track of the traffic object and a preset traffic scene;
the diagnosis evaluation module is used for inputting the lane flow and the traffic events into a preset study and judgment model to obtain the diagnosis result of the intersection output by the study and judgment model; and
and the intersection determining module is used for determining the intersection to be treated preferentially based on the diagnosis result corresponding to each intersection.
2. The deep learning technology-based intersection all-element health diagnostic system as claimed in claim 1, wherein the model building module comprises:
the historical data acquisition submodule is used for acquiring multi-frame historical images and traffic objects existing in the multi-frame historical images;
the model building submodule is used for building a basic model which takes the multi-frame images as input and takes the traffic objects as output; and
and the training submodule is used for training the basic model through the multi-frame historical images and the traffic objects existing in the multi-frame historical images to obtain a deep learning model.
3. The deep learning technology-based intersection all-element health diagnostic system according to claim 1, wherein the analysis and identification module comprises:
the scene establishing sub-module is used for establishing various traffic scenes, wherein each traffic scene is configured to contain the traffic track of a specific traffic object and the corresponding characteristic information; and
and the recognition submodule is used for recognizing lane flow and traffic events according to the reconstructed traffic track of the traffic object and the acquired characteristic information related to the traffic object.
4. The deep learning technique-based intersection all-element health diagnostic system of claim 3, wherein the identification sub-module for identifying lane traffic and traffic events from the reconstructed traffic trajectory of the traffic object and the characteristic information associated with the traffic object comprises:
when the acquired characteristic information related to the traffic object comprises the traffic track of the specific traffic object of any traffic scene and all the corresponding characteristic information thereof, identifying the traffic event corresponding to any traffic scene; and
identifying lane traffic based on the reconstructed traffic trajectory of the traffic object.
5. The deep learning technology-based intersection all-element health diagnostic system according to claim 1, wherein the traffic scenario includes any one of: the method comprises the following steps of an unreasonable signal lamp timing scene, an intersection knotting overflow scene and an intersection man-machine non-conflict scene.
6. The deep learning technology-based intersection all-element health diagnostic system according to claim 1, wherein the diagnostic evaluation module comprises:
the model acquisition submodule is used for acquiring a trained judging model, wherein the input of the judging model is the historical lane flow and the historical traffic incident of the intersection, and the output of the judging model is the diagnosis result of the intersection; and
and the result output submodule is used for inputting the lane flow and the traffic incident into the judging model so as to obtain the diagnosis result of the intersection output by the judging model.
7. The deep learning technology-based intersection all-element health diagnosis system as claimed in claim 1, wherein the intersection determination module comprises:
a result comparison submodule for comparing intersection scores shown by the diagnosis results of the respective intersections; and
and the intersection determining submodule is used for determining the intersection with the lowest score as the priority treatment intersection.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011022329.XA CN112133094A (en) | 2020-09-25 | 2020-09-25 | Road intersection full-factor health diagnosis system based on deep learning technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011022329.XA CN112133094A (en) | 2020-09-25 | 2020-09-25 | Road intersection full-factor health diagnosis system based on deep learning technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112133094A true CN112133094A (en) | 2020-12-25 |
Family
ID=73840094
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011022329.XA Pending CN112133094A (en) | 2020-09-25 | 2020-09-25 | Road intersection full-factor health diagnosis system based on deep learning technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112133094A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104157141A (en) * | 2014-08-28 | 2014-11-19 | 石成富 | Intelligent traffic regulation system based on Internet of Things |
| CN106297297A (en) * | 2016-11-03 | 2017-01-04 | 成都通甲优博科技有限责任公司 | Traffic jam judging method based on degree of depth study |
| CN109920251A (en) * | 2019-04-23 | 2019-06-21 | 公安部交通管理科学研究所 | A kind of urban road intersection traffic organization rationality diagnostic analysis method and system |
| KR102021534B1 (en) * | 2018-08-29 | 2019-10-18 | 주식회사 핀텔 | Image analysis apparatus for sensing vehicle and pedestrian based on deep learning and method thereof |
| CN110717433A (en) * | 2019-09-30 | 2020-01-21 | 华中科技大学 | Deep learning-based traffic violation analysis method and device |
| CN111311918A (en) * | 2020-05-12 | 2020-06-19 | 南京云析科技有限公司 | Traffic management method and device based on visual analysis |
| CN111540204A (en) * | 2020-05-08 | 2020-08-14 | 青岛海信网络科技股份有限公司 | Intersection problem diagnosis-oriented traffic running state assessment method and device |
-
2020
- 2020-09-25 CN CN202011022329.XA patent/CN112133094A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104157141A (en) * | 2014-08-28 | 2014-11-19 | 石成富 | Intelligent traffic regulation system based on Internet of Things |
| CN106297297A (en) * | 2016-11-03 | 2017-01-04 | 成都通甲优博科技有限责任公司 | Traffic jam judging method based on degree of depth study |
| KR102021534B1 (en) * | 2018-08-29 | 2019-10-18 | 주식회사 핀텔 | Image analysis apparatus for sensing vehicle and pedestrian based on deep learning and method thereof |
| CN109920251A (en) * | 2019-04-23 | 2019-06-21 | 公安部交通管理科学研究所 | A kind of urban road intersection traffic organization rationality diagnostic analysis method and system |
| CN110717433A (en) * | 2019-09-30 | 2020-01-21 | 华中科技大学 | Deep learning-based traffic violation analysis method and device |
| CN111540204A (en) * | 2020-05-08 | 2020-08-14 | 青岛海信网络科技股份有限公司 | Intersection problem diagnosis-oriented traffic running state assessment method and device |
| CN111311918A (en) * | 2020-05-12 | 2020-06-19 | 南京云析科技有限公司 | Traffic management method and device based on visual analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102637257B (en) | Video-based detection and recognition system and method of vehicles | |
| WO2020253965A1 (en) | Control device, system and method for determining perceptual load of a visual and dynamic driving scene in real time | |
| CN112793576B (en) | Lane change decision method and system based on rule and machine learning fusion | |
| CN115035491A (en) | Driving behavior road condition early warning method based on federal learning | |
| CN112289037B (en) | Motor vehicle illegal parking detection method and system based on high visual angle under complex environment | |
| Ketcham et al. | Recognizing the Illegal Parking Patterns of Cars on the Road in Front of the Bus Stop Using the Support Vector Machine | |
| CN212009589U (en) | Video identification driving vehicle track acquisition device based on deep learning | |
| CN114299473A (en) | Driver behavior identification method based on multi-source information fusion | |
| DE102019124419A1 (en) | SCENE CLASSIFICATION | |
| DE102021003567A1 (en) | Process for the recognition of object relationships and attributions from sensor data | |
| CN113076852A (en) | Vehicle-mounted snapshot processing system occupying bus lane based on 5G communication | |
| CN112133094A (en) | Road intersection full-factor health diagnosis system based on deep learning technology | |
| CN112652174A (en) | Parking service control method and system based on face verification | |
| CN115635969A (en) | Vehicle control method and device and vehicle | |
| DE102015216352A1 (en) | Method for detecting a possible collision of a vehicle with a pedestrian on the basis of high-resolution recordings | |
| CN113850112A (en) | Road condition identification method and system based on twin neural network | |
| Suresh et al. | Intelligent vehicle license plate recognition by deploying deep learning model for smart cities | |
| DE102020202342A1 (en) | Cloud platform for automated mobility and computer-implemented method for providing cloud-based data enrichment for automated mobility | |
| CN110489575A (en) | It is a kind of based on the data cached similar license plate filter method of virtual bayonet | |
| Husen et al. | Continuous Car Driving intention recognition with syntactic pattern approach | |
| CN114674338B (en) | Fine recommendation method for road drivable area based on layered input and output and double-attention jump | |
| Zhang et al. | A Novel Method for the Driver Lane-Changing Intention Recognition | |
| Parthiban et al. | A Deep Learning-Based Accident Detection in Emergency Situation from CCTV Camera Using H-RNN Algorithm | |
| CN116246231A (en) | Real-time traffic violation detection method and system based on deep learning | |
| CN117809269A (en) | Data sampling method and device, readable storage medium and vehicle |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201225 |
|
| RJ01 | Rejection of invention patent application after publication |