CN111898502A - Dangerous goods vehicle identification method and device, computer storage medium and electronic equipment - Google Patents
Dangerous goods vehicle identification method and device, computer storage medium and electronic equipment Download PDFInfo
- Publication number
- CN111898502A CN111898502A CN202010698582.0A CN202010698582A CN111898502A CN 111898502 A CN111898502 A CN 111898502A CN 202010698582 A CN202010698582 A CN 202010698582A CN 111898502 A CN111898502 A CN 111898502A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- neural network
- stage
- dangerous
- convolutional neural
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- 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
-
- 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
- G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/09—Recognition of logos
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Molecular Biology (AREA)
- Evolutionary Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
Dangerous goods vehicle identification method, device, computer storage medium and electronic equipment comprise: acquiring a road image; recognizing dangerous goods vehicles in the road image by utilizing a multi-stage series convolution neural network obtained by pre-training; the multi-stage series convolution neural network comprises a first-stage convolution neural network, a second-stage convolution neural network and a third-stage convolution neural network, wherein the first-stage convolution neural network identifies dangerous goods vehicles and vehicle areas thereof in the road image; the second-stage convolutional neural network identifies the dangerous goods signs of the vehicle area according to the screenshot of the vehicle area; and the third-level convolutional neural network identifies the vehicle type according to the screenshot of the vehicle area, judges the confidence coefficient of the dangerous article mark and obtains the probability of the dangerous article vehicle according to the vehicle type and the confidence coefficient of the dangerous article mark. By adopting the scheme in the application, various dangerous goods marks on the dangerous goods vehicle body can be accurately identified, and full-automatic, accurate and efficient intelligent identification of dangerous goods vehicles is realized.
Description
Technical Field
The application relates to an intelligent transportation technology, in particular to a dangerous goods vehicle identification method and device, a computer storage medium and electronic equipment.
Background
The dangerous goods transport vehicle is a special vehicle for transporting dangerous goods such as petrochemical products, explosives, firecrackers and the like. Because of high requirements on safety conditions and high accident hazard, the control of the system is very strict.
Problems existing in the prior art:
the existing management and control mode mainly depends on manpower, and is high in cost and low in efficiency. Cannot realize large-scale and timely supervision.
Disclosure of Invention
The embodiment of the application provides a method and a device for identifying dangerous goods vehicles, a computer storage medium and electronic equipment, so as to solve the technical problems.
According to a first aspect of the embodiments of the present application, there is provided a method for identifying a hazardous material vehicle, including the steps of:
acquiring a road image;
recognizing dangerous goods vehicles in the road image by utilizing a multi-stage series convolution neural network obtained by pre-training;
the multi-stage series convolution neural network comprises a first-stage convolution neural network, a second-stage convolution neural network and a third-stage convolution neural network, and the first-stage convolution neural network identifies dangerous goods vehicles and vehicle areas thereof in the road image; the second-stage convolutional neural network identifies dangerous article marks of the vehicle area according to the screenshot of the vehicle area returned by the first-stage convolutional neural network; and the third-stage convolutional neural network identifies the vehicle type according to the screenshot of the vehicle area returned by the first-stage convolutional neural network, judges the confidence coefficient of the dangerous article mark, and obtains the probability of the dangerous article vehicle according to the vehicle type and the confidence coefficient of the dangerous article mark.
According to a second aspect of the embodiments of the present application, there is provided a hazardous material vehicle identification device, including:
the acquisition module is used for acquiring a road image;
the identification module is used for identifying the dangerous goods vehicles in the road image by utilizing a multi-stage series convolution neural network obtained by pre-training;
the two-stage series convolution neural network comprises a first-stage convolution neural network, a second-stage convolution neural network and a third-stage convolution neural network, and the first-stage convolution neural network identifies dangerous goods vehicles and vehicle areas thereof in the road image; the second-stage convolutional neural network identifies dangerous article marks of the vehicle area according to the screenshot of the vehicle area returned by the first-stage convolutional neural network; and the third-stage convolutional neural network identifies the vehicle type according to the screenshot of the vehicle area returned by the first-stage convolutional neural network, judges the confidence coefficient of the dangerous article mark, and obtains the probability of the dangerous article vehicle according to the vehicle type and the confidence coefficient of the dangerous article mark.
According to a third aspect of embodiments of the present application, there is provided a computer storage medium having a computer program stored thereon, the computer program, when executed by a processor, implementing the steps of the hazardous material vehicle identification method as described above.
According to a fourth aspect of embodiments herein, there is provided an electronic device comprising one or more processors, and memory for storing one or more programs; the one or more programs, when executed by the one or more processors, implement the hazardous materials vehicle identification method as described above.
By adopting the method and the device for identifying the dangerous goods vehicles, the computer storage medium and the electronic equipment, which are provided by the embodiment of the application, based on the convolutional neural network, various dangerous goods marks on the dangerous goods vehicle body can be accurately identified, and the dangerous goods vehicles can be intelligently identified in a full-automatic, accurate and efficient manner.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
FIG. 1 is a schematic flow chart illustrating an implementation of a method for identifying a hazardous material vehicle according to a first embodiment of the present application;
FIG. 2 is a schematic diagram illustrating a process of determining a probability of a hazardous material vehicle according to a first embodiment of the present application;
FIG. 3 illustrates a vehicle type identification process in accordance with an embodiment of the present application;
FIG. 4 is a diagram illustrating a dangerous goods type score calculation process for a vehicle according to a first embodiment of the present application;
FIG. 5 is a diagram illustrating a self-learning process according to an embodiment of the present application;
fig. 6 is a schematic structural diagram illustrating a hazardous article vehicle identification device according to a second embodiment of the present application;
fig. 7 shows a schematic structural diagram of an electronic device in the fourth embodiment of the present application.
Detailed Description
The scheme in the embodiment of the application can be implemented by adopting various computer languages, such as object-oriented programming language Java and transliterated scripting language JavaScript.
In order to make the technical solutions and advantages of the embodiments of the present application more apparent, the following further detailed description of the exemplary embodiments of the present application with reference to the accompanying drawings makes it clear that the described embodiments are only a part of the embodiments of the present application, and are not exhaustive of all embodiments. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
Example one
Fig. 1 shows a schematic flow chart of an implementation of a method for identifying a hazardous material vehicle according to a first embodiment of the present application.
As shown in the figure, the method for identifying a hazardous material vehicle comprises the following steps:
the multi-stage series convolution neural network comprises a first-stage convolution neural network, a second-stage convolution neural network and a third-stage convolution neural network, and the first-stage convolution neural network identifies dangerous goods vehicles and vehicle areas thereof in the road image; the second-stage convolutional neural network identifies dangerous article marks of the vehicle area according to the screenshot of the vehicle area returned by the first-stage convolutional neural network; and the third-stage convolutional neural network identifies the vehicle type according to the screenshot of the vehicle area returned by the first-stage convolutional neural network, judges the confidence coefficient of the dangerous article mark, and obtains the probability of the dangerous article vehicle according to the vehicle type and the confidence coefficient of the dangerous article mark.
When the method is specifically implemented, the vehicle area in the road image can be identified and intercepted, and then whether relevant signs of dangerous goods exist in the intercepted vehicle area or not is predicted, so that whether the vehicle is a dangerous goods vehicle or not is judged.
Fig. 2 shows a schematic diagram of a process for determining the probability of a hazardous material vehicle in the first embodiment of the present application.
As shown in the figure, the embodiment of the present application designs a series-connection and combination network:
1) the first-level network identifies the vehicle and the position in the whole image returned by the camera;
2) making a vehicle area screenshot of the returned vehicle position;
3) the second-level network predicts the suspected mark position of the dangerous article on the vehicle screenshot;
4) simultaneously sending the vehicle screenshot and the dangerous article suspected mark screenshot into a third-level network;
5) the third-level network is a combined network comprising a plurality of classification models, wherein the first classification model is used for classifying vehicle types (such as a big tank car, a heavy truck, a car, a non-motor vehicle and the like) of the vehicle screenshots to obtain dangerous article type probabilities of different vehicle types; the confidence degree judgment is carried out on two pairs of suspected danger marks;
6) and finally, judging the probability of the dangerous goods vehicle by integrating the vehicle type and the confidence coefficient of the dangerous goods mark, for example: the vehicle probability of the dangerous goods is 0.6 multiplied by the mark probability of the dangerous goods and 0.4 multiplied by the vehicle type probability of the dangerous goods.
By adopting the dangerous goods vehicle identification method provided by the embodiment of the application, the various dangerous goods marks on the dangerous goods vehicle body can be accurately identified based on the convolutional neural network, and the full-automatic, accurate and efficient intelligent identification of the dangerous goods vehicle is realized.
In one embodiment, the acquiring the road image includes:
extracting a monitoring video shot by a camera;
and acquiring each frame of road image according to the monitoring video.
In one embodiment, the hazardous material indicia include one or more of:
dangerous brand, triangular dangerous article mark, dangerous article diamond mark, "fire"/"explode"/"rot" dangerous article subdivision mark.
In one embodiment, the training process of the first stage convolutional neural network comprises:
marking the position information of each vehicle in the training image to obtain training data comprising the position information of the vehicle;
training by means of gradient back transmission according to the training data comprising the vehicle position information to obtain a first-stage convolutional neural network; the input of the first-stage convolutional neural network is a road image, and the output of the first-stage convolutional neural network is corresponding position information of a vehicle in the road image.
In specific implementation, the step of training the first stage convolutional neural network may be as follows:
1. and (5) marking training data. And marking the position information (xmin, ymin, xmax and ymax) of each vehicle in the training picture. Thus, a batch of training data with position information is obtained.
2. And (5) training a convolutional neural network. Using labeled training data, a detection model (without specific limitation to the kind) of the convolutional neural network is trained by means of gradient back propagation. The model is made capable of knowing the vehicle position (xmin, ymin, xmax, ymax) in the image.
3. And deploying the trained model. And providing an input picture of the road, and acquiring vehicle corresponding positions (xmin, ymin, xmax and ymax) in the input picture after the model is executed.
In one embodiment, the training process of the second stage convolutional neural network comprises:
marking the position information and the type information of each dangerous article mark in the training vehicle image to obtain training data comprising the position information and the type information of the dangerous article mark;
training by means of gradient back transmission according to the training data including the position information and the type information of the dangerous article mark to obtain a second-stage convolutional neural network; and the second-stage convolutional neural network inputs the vehicle image and outputs the position information and the type information of each dangerous goods mark in the vehicle image.
In specific implementation, the step of training the second stage convolutional neural network may be as follows:
1. and (5) marking training data. And marking the position information (xmin, ymin, xmax and ymax) and the type information (burning, explosion, corrosion and the like) of each dangerous article mark in the training vehicle picture. Thus, a batch of training data with position and type information is obtained.
2. And (5) training a convolutional neural network. Using labeled training data, a detection model (without specific limitation to the kind) of the convolutional neural network is trained by means of gradient back propagation. The model has the capability of knowing the position (xmin, ymin, xmax, ymax) and the type (burning, explosion, corrosion, etc.) of each dangerous article mark in the image.
3. And deploying the trained model. And providing vehicle picture input, and obtaining the mark positions (xmin, ymin, xmax and ymax) and the types (burning, explosion, corrosion and the like) of the dangerous goods in the input picture after the model is executed.
In one embodiment, the identifying the dangerous goods vehicle in the road image by using the multi-stage series convolution neural network obtained by training in advance comprises:
identifying the position information of the area where the vehicle is located in the road image by utilizing a first-stage convolutional neural network;
intercepting a vehicle area in a matrix of the road image as a vehicle image according to the position information of the area where the vehicle is located;
inputting the vehicle image into a second-stage convolutional neural network, and identifying the position information and the type information of the dangerous goods mark by using the second-stage convolutional neural network;
identifying the type of the vehicle and judging the confidence of the dangerous goods sign by utilizing a third-stage convolutional neural network according to the screenshot of the vehicle area returned by the first-stage convolutional neural network;
and obtaining the probability of the dangerous goods vehicle according to the vehicle type and the confidence coefficient of the dangerous goods mark.
In specific implementation, the working process of the multistage series convolution neural network can be as follows:
1. and the road picture acquired by the monitoring facility passes through the first-stage convolutional neural network to obtain the vehicle position information.
2. And through programming, according to the given picture and the vehicle position information, the vehicle area in the picture matrix is intercepted to be used as a new picture, and the new picture is provided for the second-stage convolutional neural network to use.
3. Inputting the vehicle picture of the screenshot by the second-stage convolutional neural network, and deducing the mark and the type information of the dangerous goods;
4. the third-stage convolutional neural network identifies the type of the vehicle according to the screenshot of the vehicle area returned by the first-stage convolutional neural network and judges the confidence degree of the dangerous goods mark;
5. and obtaining the probability of the dangerous goods vehicle according to the vehicle type and the confidence coefficient of the dangerous goods mark.
In one embodiment, the method further comprises:
and when the dangerous goods vehicle is identified, sending a danger early warning.
During specific implementation, when the dangerous goods vehicle is identified, the road or the position of the dangerous goods vehicle can be located, and the road or the position of the dangerous goods vehicle is identified and tracked on the map according to the road or the position of the dangerous goods vehicle, so that other vehicles around or passing through the road section can be prompted to pay attention to safety.
In one embodiment, the training process of the third-stage convolutional neural network includes:
training a plurality of training vehicle sample data to obtain a vehicle type identification model;
the confidence level of the hazardous material indicator of each type of vehicle is determined from all the sampled hazardous material indicator vehicles.
Fig. 3 illustrates a vehicle type identification process in the first embodiment of the present application.
As shown in the figure, the embodiment of the present application constructs a vehicle type identification model, and the specific construction process may be as follows:
the vehicle type distinction is carried out on a batch of training vehicle data in a manual marking mode, and the types of the training vehicle data can be classified into a large tank car, a heavy truck, a light truck, a car and the like. And then training by using a multi-classification deep neural network to obtain a deep learning model capable of intelligently identifying which type of the vehicle belongs to the training types.
Fig. 4 shows a vehicle dangerous goods type score calculation process in the first embodiment of the present application.
As shown, the embodiment of the present application uses a normalized probability calculation method, assuming that the total number of vehicles with dangerous goods signs of all samples is N, the total number of vehicles with dangerous goods signs of the vehicle types (big tank truck, heavy truck, light truck, sedan) in example 4 are N1, N2, N3, N4, respectively. Then the four vehicle threat category scores are: N1/N, N2/N, N3/N, N4/N. Wherein N is N1+ N2+ N3+ N4.
In one embodiment, the method further comprises:
when the dangerous goods vehicle is a first type of dangerous goods vehicle, acquiring data of the dangerous goods vehicle, and retraining the multilevel convolutional neural network by taking the data of the dangerous goods vehicle as a training sample;
the first type of hazardous goods vehicle is a vehicle type that accounts for less than the second type of hazardous goods vehicle when the multi-stage convolutional neural network is trained.
Considering that the probability that different vehicle types belong to dangerous goods vehicles is different, the embodiment of the application determines the probability that each vehicle type belongs to the dangerous goods vehicle in a real-time self-learning adjustment mode;
the specific implementation method comprises the following steps: counting the probability of detecting the dangerous goods mark of each type of vehicle in the monitoring scene, updating the ratio of the vehicles with the dangerous goods mark in each type of vehicle in real time, and determining the probability of the various types of vehicles belonging to the dangerous goods vehicle according to the ratio. The method provided by the embodiment of the application has self-learning capability on different types of dangerous goods vehicles in various regions, and has good generalization performance.
Fig. 5 shows a schematic diagram of a self-learning process in a first embodiment of the present application.
As shown in the figure, due to the addition of the vehicle type self-learning module, a positive feedback self-lifting effect is formed between the vehicle type self-learning branch and the dangerous article mark identification branch. For a dangerous goods mark recognition model, the recognition effect of the dangerous goods mark is different on various vehicle types, the recognition effect is mainly influenced by the proportional distribution of various vehicle types in training data, and for dangerous goods vehicles, the number of the largest dangerous goods vehicles is more than 80%, and the dangerous goods vehicles probably have sufficient training data, so that the model has better performance. But for pickup trucks, the occupancy may be below 5%, so the training of the model is biased towards more data vehicle types, and the performance on pickup trucks may be very poor. After the self-learning model is introduced, dangerous goods vehicle data of types such as light trucks and the like can be obtained in each vehicle database in a targeted mode by utilizing the vehicle type self-learning module and the dangerous goods mark identification module, so that vehicle type dangerous goods vehicles with a small proportion in training data can be supplemented quickly, and the dangerous goods mark identification module can achieve a good effect on various vehicle types after the data types are balanced. Conversely, after the dangerous goods mark identification module is improved, the accuracy of the occupation ratio of various vehicles in the vehicle type identification module can be improved. And positive feedback is formed, so that the overall frame effect is improved.
In summary, the embodiment of the present application has the following advantages:
1) full-automatic, need not the manpower. Only need the camera that each crossing has been erect, catch the road surface image, this application embodiment can each vehicle in the automatic identification image to and whether the vehicle is the hazardous articles car.
2) Is accurate. By means of the method and the device, the accuracy rate of dangerous goods vehicle identification is nearly 100%, and all dangerous goods vehicles passing through the camera road can be accurately identified and reported.
3) High efficiency. The embodiment of the application can be made into a software package to be deployed on various servers, so that the image processing speed of hundreds of thousands of frames per hour is realized, and the monitoring requirement in a large range is met.
4) And timely. According to the embodiment of the application, dangerous goods vehicles can be timely identified at the first time when the front-end camera returns the image. The potential danger can be found in time, the accident can be prevented, and the problem that the accident can be found after the later-known postnotifier is avoided.
Example two
Based on the same inventive concept, the embodiment of the application provides a dangerous goods vehicle identification device, the principle of the device for solving the technical problem is similar to that of a dangerous goods vehicle identification method, and repeated parts are not repeated.
Fig. 6 shows a schematic structural diagram of a hazardous article vehicle identification device in the second embodiment of the present application.
As shown in the drawing, the hazardous material vehicle identification device includes:
an obtaining module 601, configured to obtain a road image;
the identification module 602 is configured to identify a hazardous vehicle in the road image by using a multi-stage series convolutional neural network obtained through pre-training;
the multi-stage series convolution neural network comprises a first-stage convolution neural network, a second-stage convolution neural network and a third-stage convolution neural network, and the first-stage convolution neural network identifies dangerous goods vehicles and vehicle areas thereof in the road image; the second-stage convolutional neural network identifies dangerous article marks of the vehicle area according to the screenshot of the vehicle area returned by the first-stage convolutional neural network; and the third-stage convolutional neural network identifies the vehicle type according to the screenshot of the vehicle area returned by the first-stage convolutional neural network, judges the confidence coefficient of the dangerous article mark, and obtains the probability of the dangerous article vehicle according to the vehicle type and the confidence coefficient of the dangerous article mark.
By adopting the dangerous goods vehicle identification device provided in the embodiment of the application, various dangerous goods marks on a dangerous goods vehicle body can be accurately identified based on the realization of a convolutional neural network, and full-automatic, accurate and efficient intelligent identification of dangerous goods vehicles is realized.
In one embodiment, the obtaining module includes:
the video extraction unit is used for extracting the monitoring video shot by the camera;
and the image acquisition unit is used for acquiring each frame of road image according to the monitoring video.
In one embodiment, the hazardous material indicia include one or more of:
dangerous brand, triangular dangerous article mark, dangerous article diamond mark, "fire"/"explode"/"rot" dangerous article subdivision mark.
In one embodiment, the training process of the first stage convolutional neural network comprises:
marking the position information of each vehicle in the training image to obtain training data comprising the position information of the vehicle;
training by means of gradient back transmission according to the training data comprising the vehicle position information to obtain a first-stage convolutional neural network; the input of the first-stage convolutional neural network is a road image, and the output of the first-stage convolutional neural network is corresponding position information of a vehicle in the road image.
In one embodiment, the training process of the second stage convolutional neural network comprises:
marking the position information and the type information of each dangerous article mark in the training vehicle image to obtain training data comprising the position information and the type information of the dangerous article mark;
training by means of gradient back transmission according to the training data including the position information and the type information of the dangerous article mark to obtain a second-stage convolutional neural network; and the second-stage convolutional neural network inputs the vehicle image and outputs the position information and the type information of each dangerous goods mark in the vehicle image.
In one embodiment, the identifying the dangerous goods vehicle in the road image by using a pre-trained two-stage series convolution neural network comprises:
identifying the position information of the area where the vehicle is located in the road image by utilizing a first-stage convolutional neural network;
intercepting a vehicle area in a matrix of the road image as a vehicle image according to the position information of the area where the vehicle is located;
inputting the vehicle image into a second-stage convolutional neural network, and identifying the position information and the type information of the dangerous goods mark by using the second-stage convolutional neural network;
identifying the type of the vehicle according to the screenshot of the vehicle area returned by the first-stage convolutional neural network and judging the confidence degree of the dangerous goods mark;
and obtaining the probability of the dangerous goods vehicle according to the vehicle type and the confidence coefficient of the dangerous goods mark.
In one embodiment, the training process of the third-stage convolutional neural network includes:
training a plurality of training vehicle sample data to obtain a vehicle type identification model;
the confidence level of the hazardous material indicator of each type of vehicle is determined from all the sampled hazardous material indicator vehicles.
In one embodiment, the apparatus further comprises:
and the early warning module is used for sending danger early warning when a dangerous goods vehicle is identified.
In one embodiment, the apparatus further comprises:
the self-learning module is used for acquiring data of the dangerous goods vehicle when the dangerous goods vehicle is a first type of dangerous goods vehicle, and retraining the multilevel convolutional neural network by taking the data of the dangerous goods vehicle as a training sample;
the first type of hazardous goods vehicle is a vehicle type that accounts for less than the second type of hazardous goods vehicle when the multi-stage convolutional neural network is trained.
EXAMPLE III
Based on the same inventive concept, embodiments of the present application further provide a computer storage medium, which is described below.
The computer storage medium has a computer program stored thereon, and the computer program, when executed by a processor, implements the steps of the hazardous material vehicle identification method according to an embodiment.
The computer storage medium provided by the embodiment of the application is realized based on the convolutional neural network, various dangerous article marks on a dangerous article vehicle body can be accurately identified, and full-automatic, accurate and efficient intelligent identification of dangerous article vehicles is realized.
Example four
Based on the same inventive concept, the embodiment of the present application further provides an electronic device, which is described below.
Fig. 7 shows a schematic structural diagram of an electronic device in the fourth embodiment of the present application.
As shown, the electronic device includes a memory 701 for storing one or more programs, and one or more processors 702; the one or more programs, when executed by the one or more processors, implement the hazardous substance vehicle identification method of embodiment one.
The electronic equipment provided in the embodiment of the application is realized based on the convolutional neural network, various dangerous article marks on a dangerous article vehicle body can be accurately identified, and full-automatic, accurate and efficient intelligent identification of dangerous article vehicles is realized.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While the preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.
Claims (10)
1. A method for identifying a hazardous vehicle, comprising:
acquiring a road image;
recognizing dangerous goods vehicles in the road image by utilizing a multi-stage series convolution neural network obtained by pre-training;
the multi-stage series convolution neural network comprises a first-stage convolution neural network, a second-stage convolution neural network and a third-stage convolution neural network, and the first-stage convolution neural network identifies dangerous goods vehicles and vehicle areas thereof in the road image; the second-stage convolutional neural network identifies dangerous article marks of the vehicle area according to the screenshot of the vehicle area returned by the first-stage convolutional neural network; and the third-stage convolutional neural network identifies the vehicle type according to the screenshot of the vehicle area returned by the first-stage convolutional neural network, judges the confidence coefficient of the dangerous article mark, and obtains the probability of the dangerous article vehicle according to the vehicle type and the confidence coefficient of the dangerous article mark.
2. The method of claim 1, wherein the acquiring the road image comprises:
extracting a monitoring video shot by a camera;
and acquiring each frame of road image according to the monitoring video.
3. The method of claim 1, wherein the hazardous material indicia comprises one or more of:
dangerous brand, triangular dangerous article mark, dangerous article diamond mark, "fire"/"explode"/"rot" dangerous article subdivision mark.
4. The method of claim 1, wherein the training process of the first stage convolutional neural network comprises:
marking the position information of each vehicle in the training image to obtain training data comprising the position information of the vehicle;
training by means of gradient back transmission according to the training data comprising the vehicle position information to obtain a first-stage convolutional neural network; the input of the first-stage convolutional neural network is a road image, and the output of the first-stage convolutional neural network is corresponding position information of a vehicle in the road image.
5. The method of claim 1, wherein the training process of the second stage convolutional neural network comprises:
marking the position information and the type information of each dangerous article mark in the training vehicle image to obtain training data comprising the position information and the type information of the dangerous article mark;
training by means of gradient back transmission according to the training data including the position information and the type information of the dangerous article mark to obtain a second-stage convolutional neural network; and the second-stage convolutional neural network inputs the vehicle image and outputs the position information and the type information of each dangerous goods mark in the vehicle image.
6. The method of claim 1, wherein the training process of the third-stage convolutional neural network comprises:
training a plurality of training vehicle sample data to obtain a vehicle type identification model;
the confidence level of the hazardous material indicator of each type of vehicle is determined from all the sampled hazardous material indicator vehicles.
7. The method of claim 1, further comprising:
when the dangerous goods vehicle is a first type of dangerous goods vehicle, acquiring data of the dangerous goods vehicle, and retraining the multilevel convolutional neural network by taking the data of the dangerous goods vehicle as a training sample;
the first type of hazardous goods vehicle is a vehicle type that accounts for less than the second type of hazardous goods vehicle when the multi-stage convolutional neural network is trained.
8. A hazardous material vehicle identification device, comprising:
the acquisition module is used for acquiring a road image;
the identification module is used for identifying the dangerous goods vehicles in the road image by utilizing a multi-stage series convolution neural network obtained by pre-training;
the multi-stage series convolution neural network comprises a first-stage convolution neural network, a second-stage convolution neural network and a third-stage convolution neural network, and the first-stage convolution neural network identifies vehicles and vehicle areas thereof in the road image; the second-stage convolutional neural network identifies a dangerous article mark in the vehicle area according to the screenshot of the vehicle area returned by the first-stage convolutional neural network; and the third-stage convolutional neural network identifies the vehicle type according to the screenshot of the vehicle area returned by the first-stage convolutional neural network, judges the confidence coefficient of the dangerous article mark, and obtains the probability of the dangerous article vehicle according to the vehicle type and the confidence coefficient of the dangerous article mark.
9. A computer storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 7.
10. An electronic device comprising one or more processors, and memory for storing one or more programs; the one or more programs, when executed by the one or more processors, implement the method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010698582.0A CN111898502A (en) | 2020-07-20 | 2020-07-20 | Dangerous goods vehicle identification method and device, computer storage medium and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010698582.0A CN111898502A (en) | 2020-07-20 | 2020-07-20 | Dangerous goods vehicle identification method and device, computer storage medium and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111898502A true CN111898502A (en) | 2020-11-06 |
Family
ID=73189255
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010698582.0A Pending CN111898502A (en) | 2020-07-20 | 2020-07-20 | Dangerous goods vehicle identification method and device, computer storage medium and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111898502A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112863190A (en) * | 2020-12-30 | 2021-05-28 | 北京北大千方科技有限公司 | Dangerous goods transport vehicle information acquisition equipment and acquisition method |
| CN113313098A (en) * | 2021-07-30 | 2021-08-27 | 阿里云计算有限公司 | Video processing method, device, system and storage medium |
| CN113313400A (en) * | 2021-06-08 | 2021-08-27 | 中国石油大学(北京) | Accident warning method, accident warning device and storage medium |
| CN113469158A (en) * | 2021-09-06 | 2021-10-01 | 智广海联(天津)大数据技术有限公司 | Method and system for identifying illegal hazardous chemical substance transport vehicle based on convolutional neural network |
| CN113553913A (en) * | 2021-06-29 | 2021-10-26 | 浙江大华技术股份有限公司 | Early warning method, early warning device, electronic device and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105373782A (en) * | 2015-11-16 | 2016-03-02 | 深圳市哈工大交通电子技术有限公司 | Method of automatically recognizing hazardous chemical vehicle from image or video |
| CN108038540A (en) * | 2017-11-08 | 2018-05-15 | 北京卓视智通科技有限责任公司 | A kind of multiple dimensioned neutral net and the image characteristic extracting method based on the network |
| CN208141607U (en) * | 2018-03-26 | 2018-11-23 | 河北交投智能交通技术有限责任公司 | A kind of wisdom card sending system for supporting HAZMAT vehicle control based on video identification technology |
-
2020
- 2020-07-20 CN CN202010698582.0A patent/CN111898502A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105373782A (en) * | 2015-11-16 | 2016-03-02 | 深圳市哈工大交通电子技术有限公司 | Method of automatically recognizing hazardous chemical vehicle from image or video |
| CN108038540A (en) * | 2017-11-08 | 2018-05-15 | 北京卓视智通科技有限责任公司 | A kind of multiple dimensioned neutral net and the image characteristic extracting method based on the network |
| CN208141607U (en) * | 2018-03-26 | 2018-11-23 | 河北交投智能交通技术有限责任公司 | A kind of wisdom card sending system for supporting HAZMAT vehicle control based on video identification technology |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112863190A (en) * | 2020-12-30 | 2021-05-28 | 北京北大千方科技有限公司 | Dangerous goods transport vehicle information acquisition equipment and acquisition method |
| CN113313400A (en) * | 2021-06-08 | 2021-08-27 | 中国石油大学(北京) | Accident warning method, accident warning device and storage medium |
| CN113313400B (en) * | 2021-06-08 | 2023-12-15 | 中国石油大学(北京) | Accident warning method, device and storage medium |
| CN113553913A (en) * | 2021-06-29 | 2021-10-26 | 浙江大华技术股份有限公司 | Early warning method, early warning device, electronic device and storage medium |
| CN113313098A (en) * | 2021-07-30 | 2021-08-27 | 阿里云计算有限公司 | Video processing method, device, system and storage medium |
| CN113313098B (en) * | 2021-07-30 | 2022-01-04 | 阿里云计算有限公司 | Video processing method, device, system and storage medium |
| CN113469158A (en) * | 2021-09-06 | 2021-10-01 | 智广海联(天津)大数据技术有限公司 | Method and system for identifying illegal hazardous chemical substance transport vehicle based on convolutional neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111898502A (en) | Dangerous goods vehicle identification method and device, computer storage medium and electronic equipment | |
| AU2019101142A4 (en) | A pedestrian detection method with lightweight backbone based on yolov3 network | |
| Tian et al. | An automatic car accident detection method based on cooperative vehicle infrastructure systems | |
| Peng et al. | Uncertainty evaluation of object detection algorithms for autonomous vehicles | |
| US20210182713A1 (en) | Explainable artificial intelligence (ai) based image analytic, automatic damage detection and estimation system | |
| CN113033604B (en) | Vehicle detection method, system and storage medium based on SF-YOLOv4 network model | |
| CN110414320B (en) | Method and system for safety production supervision | |
| CN108509954A (en) | A kind of more car plate dynamic identifying methods of real-time traffic scene | |
| CN113052159A (en) | Image identification method, device, equipment and computer storage medium | |
| CN115376101A (en) | Incremental learning method and system for automatic driving environment perception | |
| CN113423063A (en) | Vehicle monitoring method and device based on vehicle-mounted T-BOX, vehicle and medium | |
| CN116645586A (en) | Port container damage detection method and system based on improved YOLOv5 | |
| CN109993058A (en) | The recognition methods of road signs based on multi-tag classification | |
| CN115810134A (en) | Image acquisition quality inspection method, system and device for preventing car insurance from cheating | |
| CN112085001B (en) | Tunnel identification model and method based on multi-scale edge feature detection | |
| Wu et al. | Research on asphalt pavement disease detection based on improved YOLOv5s | |
| Khan | Vehicle and pedestrian detection using YOLOv3 and YOLOv4 for self-driving cars | |
| CN115439450A (en) | Video picture quality evaluation method, device, equipment and storage medium for automatic driving | |
| US11912289B2 (en) | Method and device for checking an AI-based information processing system used in the partially automated or fully automated control of a vehicle | |
| CN114973169A (en) | Vehicle classification counting method and system based on multi-target detection and tracking | |
| CN113591543B (en) | Traffic sign recognition method, device, electronic equipment and computer storage medium | |
| Sikander et al. | Image Classification using CNN for Traffic Signs in Pakistan | |
| CN114359884A (en) | License plate recognition method and device, electronic equipment and storage medium | |
| Alfinnur Charisma | Modified YOLOv5-based License Plate Detection: an Efficient Approach for Automatic Vehicle Identification | |
| Sisias et al. | Preliminary results of a proposed CNN framework for use in motorway applicable detection systems |
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 |