CN109815912B - Highway safety inspection system based on artificial intelligence - Google Patents
Highway safety inspection system based on artificial intelligence Download PDFInfo
- Publication number
- CN109815912B CN109815912B CN201910080044.2A CN201910080044A CN109815912B CN 109815912 B CN109815912 B CN 109815912B CN 201910080044 A CN201910080044 A CN 201910080044A CN 109815912 B CN109815912 B CN 109815912B
- Authority
- CN
- China
- Prior art keywords
- image
- vehicle
- data
- image data
- road
- 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.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Traffic Control Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an artificial intelligence-based highway safety inspection system, which comprises a data set model module, a training model module and a model application module, wherein the data set model module comprises image data, GPS data and image identification data of a high-speed camera; the training model module is used for carrying out classification learning on the data of the data set model module and storing the learning result into a database; and the model application module is used for analyzing and processing the road information acquired by the monitoring vehicle, storing the analysis and processing result into a database and sending the analysis and processing result to a PC (personal computer) end or a mobile terminal. The road traffic monitoring system can comprehensively identify road information, effectively monitor and report vehicles on the road in time, can monitor and early warn for 24 hours, reminds passing vehicles, reduces the number of times of manual inspection, improves inspection efficiency, ensures that the working environment of workers is safer, the working efficiency is more efficient, manpower and material resources are saved, and the road driving safety is improved.
Description
Technical Field
The invention relates to the field of highway safety inspection, in particular to a highway safety inspection system based on artificial intelligence.
Background
Traditional road inspection mode needs road administration and maintenance department to go on the road many times a day and patrol, meets abnormal conditions, needs get off the bus and shoot, lets the staff be in dangerous position, and shoots the unable timely report of photo information. The mode has the problems of high labor cost, multiple inspection times, low inspection efficiency and the like. And highway traffic lines are many, and traditional patrol means is consuming time and energy, and the danger that faces can cause the secondary accident, often can influence the normal use of traffic because the accident data are collected too slowly to can not accomplish effective early warning to other vehicles trip.
Disclosure of Invention
The invention aims to solve the defects of the prior art and provides an artificial intelligence-based highway safety inspection system.
The technical scheme adopted by the invention for solving the technical problems is as follows:
an artificial intelligence-based highway safety inspection system comprises a data set model module, a training model module and a model application module,
the data set model module comprises image data of a high-speed camera, GPS data and image identification data; the image data and the GPS data of the high-speed camera are acquired by a monitoring vehicle;
the training model module is used for carrying out classified learning on the data of the data set model module and storing the learning result into a database;
and the model application module is used for analyzing and processing the road information acquired by the monitoring vehicle through the training model module, storing the analysis and processing result into the database and sending the analysis and processing result to the PC end or the mobile terminal.
Preferably, the image data of the high-speed camera includes road surface image data, vehicle image data and sign image data,
the road surface image data comprise a normal road surface image set with dry ground, a hollow road surface image set, a road surface image set containing foreign matters and a road surface image set containing ice and snow;
the vehicle image data comprises vehicle type image data and vehicle state image data, and the vehicle type image data comprises a small passenger vehicle image set, a freight transport vehicle image set and a special vehicle image set; the vehicle state image data comprises a vehicle image set in a normal driving state, a vehicle image set in a stopping state and a vehicle image set in an abnormal driving state, wherein the vehicle image set in the abnormal driving state comprises a vehicle image set for driving in the wrong direction, speeding and occupying an emergency lane for a long time;
the sign image data comprises a road sign image set, a vehicle license plate image set and a road section stake mark image set.
Preferably, the image recognition data is obtained by recognizing road sign image data, vehicle license plate image data and road section stake mark image data through an OCR recognition mode, and includes recognized road sign image data, vehicle license plate image data and road section stake mark image data.
Furthermore, the GPS data includes coordinates of a start point and a finish point of the road section to be monitored, and coordinates corresponding to each acquired image.
Further, the coordinates of each image are obtained by using the RTK dynamic measurement technique of the GPS.
Preferably, the training Model module comprises an artificial intelligence recognition system adopting a high-performance distributed deep learning framework PaddlePaddlePaddle, a matched grid structure and hyper-parameters are intelligently selected through an automatic Model selection mode (Auto Model Search), the effect of the Model is improved, an Early Stopping strategy is adopted, the overfitting risk is reduced, and the recognition accuracy is improved; according to the artificial image labeling, image semantic segmentation is carried out through an image semantic segmentation technology of an ICNet image cascade network, balance between accuracy of high-resolution images and efficiency of a low-complexity network is obtained, and learning results are stored in a database.
Preferably, the monitoring vehicle comprises a GPS module, a high-speed camera and a network transmission module, wherein the GPS module is used for positioning coordinates of images shot by the high-speed camera and positioning coordinates of a starting point and a finishing point of a road section to be monitored; the high-speed camera is used for shooting road surface images, vehicle images and sign images; and the network transmission module is used for uploading the image shot by the high-speed camera and the coordinate positioned by the GPS module to the server control platform.
Further, the server control platform performs image recognition on the signboard image shot by the high-speed camera through an OCR recognition mode.
The road traffic monitoring system can comprehensively identify road information, effectively monitor and report vehicles on the road in time, can monitor and early warn for 24 hours, and remind passing vehicles, so that traffic traveling is safer, manual inspection times are reduced, inspection efficiency is improved, working environment of workers is safer, working efficiency is more efficient, manpower and material resources are saved, and road driving safety is improved.
Drawings
The accompanying drawings are included to provide a further understanding of the invention. In the drawings:
fig. 1 is a flow chart of the working process of the artificial intelligence-based highway safety inspection system.
FIG. 2 is a flow chart of the training model phase of the artificial intelligence based highway safety patrol system of the present invention.
Fig. 3 is a schematic structural view of the monitoring vehicle of the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1-3, an artificial intelligence based highway safety patrol system comprises a data set model module, a training model module and a model application module,
the data set model module comprises image data of a high-speed camera, GPS data and image identification data; the image data and the GPS data of the high-speed camera are acquired by a monitoring vehicle;
the image data of the high-speed camera comprises road surface image data, vehicle image data, sign image data and road information data, and because a large amount of training data needs to be provided based on an artificial intelligence recognition technology, the required recognition data is divided into the following sets;
the road surface image data comprise a normal road surface image set with dry ground, a hollow road surface image set, a road surface image set containing foreign matters and a road surface image set containing ice and snow;
the vehicle image data comprises vehicle type image data and vehicle state image data, and the vehicle type image data comprises a small passenger vehicle image set, a freight transport vehicle image set and a special vehicle image set; the vehicle state image data comprises a vehicle image set in a normal driving state, a vehicle image set in a stopping state and a vehicle image set in an abnormal driving state, wherein the vehicle image set in the abnormal driving state comprises a vehicle image set for driving in the wrong direction, speeding and occupying an emergency lane for a long time;
the sign image data comprises a road sign image set, a vehicle license plate image set and a road section stake mark image set. The method comprises the steps of adopting a mode combining GPS and OCR mode recognition, setting starting and ending GPS coordinates according to a road section to be monitored, when a monitoring vehicle runs on a road, simultaneously recording the GPS coordinates where a photo is located when one photo is collected, carrying out OCR mode recognition on each photo, and simultaneously recording recognized label information and corresponding GPS data information into a data set. The image recognition data is obtained by recognizing road sign image data, vehicle license plate image data and road section pile number image data through an OCR recognition mode, and comprises the recognized road sign image data, the recognized vehicle license plate image data, the recognized road section pile number image data and road property and road right information of a road section to be monitored, which is marked in advance. By adopting the OCR mode recognition function and the vehicle monitoring function, the vehicle driving standard of the inspection road section is automatically formulated according to the content of the road signboard in the inspection process, the vehicle operation problems can be effectively and simultaneously inspected under the condition of monitoring the road surface road condition, vehicles with the driving problems are timely reported and timely disposed, the combination of the road surface inspection and the vehicle inspection is realized, and the road driving safety is greatly improved. The prior artificial intelligence patrol system does not have the function.
The GPS data comprises the coordinates of the starting point and the ending point of the road section to be monitored and the coordinates corresponding to each collected image. The coordinates of each image are acquired using the RTK dynamic measurement technique of the GPS and stored in a data set.
The training model module is used for carrying out classification learning on the data of the data set model module and storing a learning result into a database; the key point of the artificial intelligence technology is that a large amount of learning is carried out on the system, the learning information comprises correct information and wrong information, and the larger the learning information amount is, the greater the accuracy rate of the judged things is. The training Model module adopts an artificial intelligence recognition system of a high-performance distributed deep learning framework PaddlePaddley, intelligently selects the most appropriate grid structure and hyper-parameters through an automatic Model selection mode (Auto Model Search), improves the Model effect, and solves the problems that the artificial intelligence recognition technology of the prior method can only manually select the recognition grid mode and set the parameters, so that the recognition rate is low and the like. The Early Stopping strategy is adopted, the overfitting risk is reduced, the identification accuracy is improved, and the image classification of the artificial intelligent identification system exceeds the Top-1Accuracy of 67% of models by 90%; most models Top-1Accuracy >.
In the training mode, the acquired image data is uploaded to an artificial intelligence recognition technology server through a network transmission technology according to the data set classification form of a data set model module for classification learning. The artificial intelligence recognition technology continuously searches the same points of different photos of the uploaded image data, and then achieves the effect of image semantic segmentation according to artificially defined meanings, namely artificial image annotation. The image semantic segmentation is to group and segment image pixels according to different expressed semantic meanings, and the image semantic meaning is to understand the image content. Compared with other methods, the method has the advantages of higher speed and higher accuracy, and the idea is to convert an input Image into different resolutions, then calculate the input with different resolutions by using sub-networks with different computational complexity, and then combine the results. The ICNet is composed of three sub-networks, the network with high computational complexity processes the low resolution input, the network with low computational complexity processes the network with high resolution, a balance is obtained between the accuracy of the high resolution image and the efficiency of the low complexity network by this way, and the learning result is stored in the database.
And the model application module is used for analyzing and processing the road information acquired by the monitoring vehicle through the training model module, storing the analysis and processing result into the database and sending the analysis and processing result to the PC end or the mobile terminal.
The monitoring vehicle comprises a GPS module, a high-speed camera and a network transmission module, wherein the GPS module is used for positioning the coordinates of images shot by the high-speed camera and positioning the coordinates of a starting point and an end point of a road section to be monitored; the high-speed camera is used for shooting road surface images, vehicle images and sign images; and the network transmission module is used for uploading the image shot by the high-speed camera and the coordinate positioned by the GPS module to the server control platform. And the server control platform performs image recognition on the label image shot by the high-speed camera through an OCR recognition mode.
Firstly, basic information including coordinates of a starting point and an ending point, road property and road right information, a road section pile number and the like is input into a measured road section. Then adopt the monitoring vehicle to gather road information, upload to server control platform through mobile unit control cabinet control network transmission module, discern image information by artificial intelligence identification system, road information includes road surface image data, vehicle image data, sign image data and other road information data, according to established rule, to the road surface situation, vehicle illegal information, sign information corresponds with GPS information, GPS information corresponds with highway section stake number, and the terminal of giving down road manager after handling completion with these information in the system includes PC terminal, mobile terminal etc.. The problem of road surface is handled in time to the convenience of managers, reduces the influence of road surface to the driver. The monitoring vehicle is adopted to collect data, the server processes the data, and the user terminal machine is integrated with road routing inspection for receiving the data, so that a manager can find and dispose the problems in time, can monitor and early warn in 24 hours in all weather, and the problem of road driving safety caused by untimely disposition due to unsmooth information is solved; the method improves the inspection efficiency, saves manpower and material resources, and solves the problems that the prior method only depends on manual investigation, and has low fault tolerance rate, low efficiency, high cost and the like.
The system establishes a full life cycle image record for the road, processes the road patrol video through an artificial intelligence technology, automatically identifies and records information such as the road surface, guardrails, meteorology, facilities, vehicles, labels, road property and road right, provides real-time and comprehensive road condition information for passing vehicles, operation and traffic management departments, and provides a road historical information archive for owners.
Although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described in the foregoing embodiments, or equivalents may be substituted for elements thereof.
Claims (4)
1. An artificial intelligence-based highway safety inspection system comprises a data set model module, a training model module and a model application module,
the data set model module comprises image data of a high-speed camera, GPS data and image identification data; the image data and the GPS data of the high-speed camera are acquired by a monitoring vehicle;
the training model module is used for carrying out classification learning on the data of the data set model module and storing a learning result into a database;
the model application module is used for analyzing and processing road information acquired by the monitoring vehicle through the training model module, storing an analysis processing result into a database and sending the analysis processing result to a PC (personal computer) end or a mobile terminal;
the image data of the high-speed camera includes road surface image data, vehicle image data and sign image data,
the road surface image data comprises a normal road surface image set with dry ground, a hollow road surface image set, a road surface image set containing foreign matters and a road surface image set containing ice and snow;
the vehicle image data comprises vehicle type image data and vehicle state image data, and the vehicle type image data comprises a small passenger vehicle image set, a freight transport vehicle image set and a special vehicle image set; the vehicle state image data comprises a vehicle image set in a normal driving state, a vehicle image set in a stopping state and a vehicle image set in an abnormal driving state, wherein the vehicle image set in the abnormal driving state comprises a vehicle image set for driving in the wrong direction, speeding and occupying an emergency lane for a long time;
the signboard image data comprises a road signboard image set, a vehicle license plate image set and a road section stake mark image set;
the image recognition data is obtained by recognizing road sign image data, vehicle license plate image data and road section pile number image data through an OCR recognition mode, and comprises recognized road sign image data, vehicle license plate image data and road section pile number image data;
the GPS data comprises a starting point coordinate and a finishing point coordinate of the road section to be monitored and a coordinate corresponding to each collected image;
the training model module comprises an artificial intelligence recognition system adopting a high-performance distributed deep learning framework PaddlePaddlePaddley, the matched grid structure and the hyper-parameters are intelligently selected through an automatic model selection mode, the effect of the model is improved, and an EarlyStopping strategy is adopted, so that the overfitting risk is reduced, and the recognition accuracy is improved; according to the artificial image labeling, image semantic segmentation is carried out through an image semantic segmentation technology of an ICNet image cascade network, balance between accuracy of high-resolution images and efficiency of a low-complexity network is obtained, and learning results are stored in a database.
2. The artificial intelligence based highway safety patrol system according to claim 1, wherein: the coordinates of each image are acquired using RTK dynamic measurement techniques of GPS.
3. The artificial intelligence based highway safety patrol system according to claim 1, wherein: the monitoring vehicle comprises a GPS module, a high-speed camera and a network transmission module, wherein the GPS module is used for positioning the coordinates of images shot by the high-speed camera and positioning the coordinates of a starting point and an end point of a road section to be monitored; the high-speed camera is used for shooting road surface images, vehicle images and sign images; and the network transmission module is used for uploading the image shot by the high-speed camera and the coordinates positioned by the GPS module to the server control platform.
4. The artificial intelligence based highway safety patrol system according to claim 3, wherein: and the server control platform performs image recognition on the label image shot by the high-speed camera through an OCR recognition mode.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910080044.2A CN109815912B (en) | 2019-01-28 | 2019-01-28 | Highway safety inspection system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910080044.2A CN109815912B (en) | 2019-01-28 | 2019-01-28 | Highway safety inspection system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109815912A CN109815912A (en) | 2019-05-28 |
| CN109815912B true CN109815912B (en) | 2023-04-07 |
Family
ID=66605480
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910080044.2A Active CN109815912B (en) | 2019-01-28 | 2019-01-28 | Highway safety inspection system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109815912B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110085029B (en) * | 2019-05-31 | 2020-08-04 | 重庆交通大学 | Highway inspection system and method based on rail type inspection robot |
| CN111126802A (en) * | 2019-12-10 | 2020-05-08 | 福建省高速公路集团有限公司 | Highway inspection and evaluation method and system based on artificial intelligence |
| CN111666833A (en) * | 2020-01-20 | 2020-09-15 | 深圳大学 | Road damage detection method based on vehicle formation in intelligent networking environment |
| CN113096395B (en) * | 2021-03-31 | 2022-03-25 | 武汉理工大学 | Road traffic safety evaluation system based on positioning and artificial intelligence recognition |
| CN113588664A (en) * | 2021-08-02 | 2021-11-02 | 安徽省通途信息技术有限公司 | Vehicle-mounted road defect rapid inspection and analysis system |
| CN114842649B (en) * | 2022-07-01 | 2022-09-06 | 无锡软美信息科技有限公司 | Traffic control method and system for expressway |
| CN115914298B (en) * | 2022-11-21 | 2023-10-31 | 常州天涛通信设备有限公司 | Communication system for highway remote terminal |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN205388829U (en) * | 2016-02-19 | 2016-07-20 | 李彦玮 | Highway monitoring devices that breaks rules and regulations based on unmanned aerial vehicle |
| WO2018045731A1 (en) * | 2016-09-07 | 2018-03-15 | 东南大学 | Expressway emergency lane occupation supervision and punishment method and system |
| CN108510750A (en) * | 2018-04-25 | 2018-09-07 | 济南浪潮高新科技投资发展有限公司 | A method of the unmanned plane inspection parking offense based on neural network model |
-
2019
- 2019-01-28 CN CN201910080044.2A patent/CN109815912B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN205388829U (en) * | 2016-02-19 | 2016-07-20 | 李彦玮 | Highway monitoring devices that breaks rules and regulations based on unmanned aerial vehicle |
| WO2018045731A1 (en) * | 2016-09-07 | 2018-03-15 | 东南大学 | Expressway emergency lane occupation supervision and punishment method and system |
| CN108510750A (en) * | 2018-04-25 | 2018-09-07 | 济南浪潮高新科技投资发展有限公司 | A method of the unmanned plane inspection parking offense based on neural network model |
Non-Patent Citations (1)
| Title |
|---|
| 基于OpenCV+Python的道路结冰检测系统的设计;李颀等;《陕西科技大学学报(自然科学版)》;20170425(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109815912A (en) | 2019-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109815912B (en) | Highway safety inspection system based on artificial intelligence | |
| CN109686088B (en) | Traffic video alarm method, equipment and system | |
| CN111833598B (en) | Automatic traffic incident monitoring method and system for unmanned aerial vehicle on highway | |
| CN106373426A (en) | Computer vision-based parking space and illegal lane occupying parking monitoring method | |
| CN109782364B (en) | Traffic sign board missing detection method based on machine vision | |
| CN114489122B (en) | UAV and matching airport-based automatic highway inspection method and system | |
| CN111723706A (en) | Box type freight car door opening monitoring device and system based on raspberry group | |
| CN110021167A (en) | A kind of highway cloud inspection system and data processing method | |
| CN116434525A (en) | Intelligent management early warning system for expressway | |
| CN113903008A (en) | Ramp exit vehicle violation identification method based on deep learning and trajectory tracking | |
| CN113588664A (en) | Vehicle-mounted road defect rapid inspection and analysis system | |
| CN111209880A (en) | Vehicle behavior identification method and device | |
| CN114723271A (en) | Power transmission project quality detection method and system based on image recognition | |
| CN103528531A (en) | Intelligent Internet of Things image detection system for small vehicle parameters | |
| CN113487877A (en) | Road vehicle illegal parking monitoring method | |
| CN114783188A (en) | Inspection method and device | |
| CN109934161B (en) | Vehicle identification and detection method and system based on convolutional neural network | |
| CN113256014B (en) | Intelligent detection system for 5G communication engineering | |
| CN116721552B (en) | Non-motor vehicle overspeed identification recording method, device, equipment and storage medium | |
| CN117523449A (en) | Vehicle road co-location system and method for underground coal mine auxiliary transportation robot | |
| CN116631187A (en) | Intelligent acquisition and analysis system for case on-site investigation information | |
| CN114743376B (en) | Intelligent traffic system for non-lamp-controlled intersection | |
| CN114333331B (en) | Method and system for identifying vehicle passing information and vehicle weight of multi-lane bridge | |
| CN111582191A (en) | Pouring amount estimation method in concrete pouring based on artificial intelligence video analysis | |
| CN115273474A (en) | RPA patrols and examines robot and patrols and examines management system based on artificial intelligence |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |