CN109685124A - Road disease recognition methods neural network based and device - Google Patents
Road disease recognition methods neural network based and device Download PDFInfo
- Publication number
- CN109685124A CN109685124A CN201811534062.5A CN201811534062A CN109685124A CN 109685124 A CN109685124 A CN 109685124A CN 201811534062 A CN201811534062 A CN 201811534062A CN 109685124 A CN109685124 A CN 109685124A
- Authority
- CN
- China
- Prior art keywords
- disease
- road
- image
- class label
- classification
- 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
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the present invention provides a kind of road disease recognition methods neural network based and device, this method comprises: obtaining the road disease image that vehicle-mounted acquisition equipment is sent, the road disease image is input in preset convolutional neural networks model, the corresponding disease class label of the road disease image is calculated, disease relative position and disease example cut zone, the disease relative position is position of the disease in road disease image, the corresponding disease classification of the road disease image is determined according to the disease class label, the corresponding disease grade of the road disease image is determined according to the disease relative position or the disease example cut zone, disease geo-radar image automatic identification disease classification and disease grade can be passed through, recognition efficiency is high, the accuracy of identification can be improved by preset convolutional neural networks model, it keeps away Manpower-free's recognition efficiency is low, there are problems that security risk.
Description
Technical field
The present embodiments relate to field of computer technology more particularly to a kind of road disease identifications neural network based
Method and device.
Background technique
With constantly improve for China's transport development, road traffic occupies important ground in China's economy and people's lives
Position.However as the construction of road, the later period is maintained into as main problem, how when road disease occurs quickly discovery is asked
It inscribes and takes measures, maintenance of surface expense can be greatlyd save, therefore in the case where not influencing normal traffic order, it is quickly right
The disease in whole section of roadside identify and orientation problem area, becomes a great problem urgently to be resolved.
Currently, traditional carry out road disease to know method for distinguishing mainly by way of artificial scene investigation, that is, exist
During carrying out regular road maintenance, road maintenance personnel drive vehicle and make an inspection tour road, stop when finding disease attached to disease
Closely, disease picture is shot manually and records Damage Types, position, and disease picture and Damage Types, position are uploaded to system.
However, it is found by the inventors that there are the following problems: the effect manually checked for the mode of existing manual identified road disease
Rate is low, and there are larger hidden dangers for the personal safety of road maintenance personnel.
Summary of the invention
The present invention provides a kind of road disease recognition methods neural network based and device, passes through acquisition road disease figure
As being simultaneously input to neural network model, can identify road disease automatically and intelligently, improve road disease identification efficiency and
Safety.
In a first aspect, the present invention provides a kind of road disease recognition methods neural network based, comprising:
Obtain the road disease image that vehicle-mounted acquisition equipment is sent;
The road disease image is input in preset convolutional neural networks model, the road disease is calculated
The corresponding disease class label of image, disease relative position and disease example cut zone, the disease relative position are that disease exists
Position in road disease image, wherein the preset convolutional neural networks model is the road by preset quantity by label
Disease geo-radar image training obtains;
The corresponding disease classification of the road disease image is determined according to the disease class label;
Determine that the road disease image is corresponding according to the disease relative position or the disease example cut zone
Disease grade.
Optionally, described that the road disease image is input in preset convolutional neural networks model, it is calculated
The corresponding disease class label of the road disease image, disease relative position and disease example cut zone, comprising:
Feature extraction is carried out to the road disease image, obtains multilayer feature image;
The candidate frame of multilayer feature figure is extracted, the candidate frame is to be made of the boundary of disease;
Pond is carried out to the candidate frame, the candidate frame of extraction is zoomed to the feature vector of default matrix size;
Described eigenvector is input to the first predicted branches of the preset convolutional neural networks model, is calculated
The corresponding disease class label of the road disease image and disease relative position;
Described eigenvector is input to the second predicted branches of the preset convolutional neural networks model, is calculated
The corresponding disease example cut zone of the road disease image.
Optionally, the disease class label is multiple;It is described that the road disease figure is determined according to the disease class label
As corresponding disease classification, comprising:
The maximum value in the disease class label is obtained, the corresponding disease classification of the maximum value is determined as the road
The corresponding disease classification of disease geo-radar image.
Optionally, if the maximum value in the disease class label is less than preset effective threshold value, determine the disease class
Zhi not be invalid, and re-execute the step of obtaining the road disease image that vehicle-mounted acquisition equipment is sent.
Optionally, the disease classification bulk disease;
The corresponding disease grade of the road disease image is determined according to the disease example cut zone, comprising:
The disease example cut zone is expanded, corrosion treatment, so that the disease example cut zone becomes
One complete and connection region;
The disease area of disease example cut zone after calculation processing;
The corresponding disease grade of the road disease image is determined according to the disease area.
Optionally, the disease classification is linear disease, and the disease relative position is positioned at the road disease image
The maximum boundary rectangle frame on middle disease boundary;
The corresponding disease grade of the road disease image is determined according to the disease relative position, comprising:
Obtain the catercorner length of the rectangle frame;
The corresponding disease grade of the road disease image is determined according to diagonal length.
Optionally, the method also includes:
While obtaining the road disease image that vehicle-mounted acquisition equipment is sent, the geography that vehicle positioning equipment is sent is obtained
Coordinate information;
The road disease classification, disease grade and geographic coordinate information are sent to the terminal device of maintenance personnel.
Second aspect, the present invention provide a kind of road disease identification device neural network based, comprising:
Disease geo-radar image obtains module, the road disease image sent for obtaining vehicle-mounted acquisition equipment;
Processing with Neural Network module, for the road disease image to be input to preset convolutional neural networks model
In, the corresponding disease class label of the road disease image, disease relative position and disease example cut zone, institute is calculated
State disease relative position be position of the disease in road disease image, wherein the preset convolutional neural networks model be by
Preset quantity is obtained by the road disease image training of label;
Disease category determination module, for determining the corresponding disease of the road disease image according to the disease class label
Classification;
Disease level determination module, for determining institute according to the disease relative position or the disease example cut zone
State the corresponding disease grade of road disease image.
Optionally, the Processing with Neural Network module, comprising:
Characteristic image extraction unit obtains multilayer feature image for carrying out feature extraction to the road disease image;
Candidate frame extraction unit, for extracting the candidate frame of multilayer feature figure, the candidate frame be boundary by disease
Composition;
The candidate frame of extraction is zoomed to default matrix for carrying out pond to the candidate frame by candidate frame pond unit
The feature vector of size;
First predicting unit, for described eigenvector to be input to the first of the preset convolutional neural networks model
The corresponding disease class label of the road disease image and disease relative position is calculated in predicted branches;
Second predicting unit, for described eigenvector to be input to the second of the preset convolutional neural networks model
The corresponding disease example cut zone of the road disease image is calculated in predicted branches.
Optionally, the disease class label is multiple;The disease category determination module is specifically used for obtaining the disease
The corresponding disease classification of the maximum value is determined as the corresponding disease class of the road disease image by the maximum value in class label
Not.
Optionally, the disease category determination module is preset if the maximum value being also used in the disease class label is less than
Effective threshold value, then determine that the disease class label is invalid, and re-execute obtain it is vehicle-mounted acquisition equipment send road disease
The step of image.
Optionally, the disease classification bulk disease;The disease level determination module is specifically used for real to the disease
Example cut zone expanded, corrosion treatment, so that the disease example cut zone becomes complete and connection a region;
The disease area of disease example cut zone after calculation processing;
The corresponding disease grade of the road disease image is determined according to the disease area.
Optionally, the disease classification is linear disease, and the disease relative position is positioned at the road disease image
The maximum boundary rectangle frame on middle disease boundary;Disease level determination module, specifically for obtaining the diagonal line length of the rectangle frame
Degree;The corresponding disease grade of the road disease image is determined according to diagonal length.
Optionally, described device further include:
Geographic coordinate information obtain module, for obtain it is vehicle-mounted acquisition equipment send road disease image while,
Obtain the geographic coordinate information that vehicle positioning equipment is sent;
Sending module, for the road disease classification, disease grade and geographic coordinate information to be sent to maintenance personnel
Terminal device.
The third aspect, the present invention provide a kind of road disease identification equipment neural network based, comprising: at least one
Manage device and memory;
The memory stores computer executed instructions;
At least one described processor executes the computer executed instructions of memory storage so that it is described at least one
Processor executes such as the described in any item road disease recognition methods neural network based of first aspect.
Fourth aspect, the present invention provide a kind of computer readable storage medium, deposit in the computer readable storage medium
Computer executed instructions are contained, when processor executes the computer executed instructions, are realized as described in any one of first aspect
Road disease recognition methods neural network based.
Road disease recognition methods neural network based provided in an embodiment of the present invention, this method obtain vehicle-mounted acquisition and set
The road disease image is input in preset convolutional neural networks model, calculates by the road disease image that preparation is sent
To the corresponding disease class label of the road disease image, disease relative position and disease example cut zone, the disease phase
Contraposition is set to position of the disease in road disease image, wherein the preset convolutional neural networks model is by preset quantity
It is obtained by the road disease image training of label, determines that the road disease image is corresponding according to the disease class label
Disease classification determines that the road disease image is corresponding according to the disease relative position or the disease example cut zone
Disease grade, can be by disease geo-radar image automatic identification disease classification and disease grade, and recognition efficiency is high, passes through preset convolution
The accuracy of identification can be improved in neural network model, avoids the problem that manual identified inefficiency, there are security risks.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention without any creative labor, may be used also for those of ordinary skill in the art
To obtain other drawings based on these drawings.
Fig. 1 is the architecture diagram of road disease identifying system provided in an embodiment of the present invention;
Fig. 2 is the flow chart one of road disease recognition methods neural network based provided in an embodiment of the present invention;
Fig. 3 is that the disease relative position of position of the disease provided in an embodiment of the present invention in road disease image is illustrated
Figure;
Fig. 4 is the flowchart 2 of road disease recognition methods neural network based provided in an embodiment of the present invention;
Fig. 5 is the flow chart 3 of road disease recognition methods neural network based provided in an embodiment of the present invention;
Fig. 6 is the flow chart four of road disease recognition methods neural network based provided in an embodiment of the present invention;
The position view of Fig. 7 disease example cut zone provided in an embodiment of the present invention and its maximum boundary rectangle frame;
Fig. 8 is the flow chart five of road disease recognition methods neural network based provided in an embodiment of the present invention;
Fig. 9 is the structural schematic diagram one of road disease identification device neural network based provided in an embodiment of the present invention;
Figure 10 is the structural schematic diagram two of road disease identification device neural network based provided in an embodiment of the present invention;
Figure 11 is the hardware configuration signal that road disease neural network based provided in an embodiment of the present invention identifies equipment
Figure.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Description and claims of this specification and term " first ", " second ", " third " " in above-mentioned attached drawing
The (if present)s such as four " are to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should manage
The data that solution uses in this way are interchangeable under appropriate circumstances, so that the embodiment of the present invention described herein for example can be to remove
Sequence other than those of illustrating or describe herein is implemented.In addition, term " includes " and " having " and theirs is any
Deformation, it is intended that cover it is non-exclusive include, for example, containing the process, method of a series of steps or units, system, production
Product or equipment those of are not necessarily limited to be clearly listed step or unit, but may include be not clearly listed or for this
A little process, methods, the other step or units of product or equipment inherently.
Fig. 1 is the architecture diagram of road disease identifying system provided in an embodiment of the present invention, as shown in Figure 1, the present invention is implemented
Road disease identifying system in example, comprising: mobile unit 100, server 200 and terminal device 300.
Wherein, mobile unit 100 includes the image capture device 110 being arranged on vehicle, can be camera, video camera
Or other image capture devices.Image capture device 110 is used to obtain road disease image in vehicle travel process.It is vehicle-mounted to set
Standby 100 further include the positioning device 120 being arranged on vehicle, can be GPS (Global Positioning System, the whole world
Positioning system) locating module or Beidou positioning module.120 user of positioning device obtains the corresponding geographical location of road disease image
Information.
Server 200 can be a server, or the server cluster consisted of several servers or one
A cloud computing service platform.Server can receive the road disease image of the transmission of image capture device 310, and to road disease
Image is analyzed and processed to obtain the defect information of road disease, while can receive the road disease of the transmission of positioning device 320
Geographical location information, the defect information of road disease and geographical location information are then sent to terminal device 300.
Terminal device 100 is the electronic equipment that maintenance personnel obtain road disease information, including but not limited to: desktop
Brain, laptop, tablet computer, mobile phone, intelligent wearable device etc..Terminal device 100 can be equipped with application client
End, or browser is installed, by the webpage client of browser access application, obtain road disease information and geography
Location information, so that maintenance personnel timely feedback according to road defect information and geographical location information.
Technical solution of the present invention is described in detail with specifically embodiment below.These specific implementations below
Example can be combined with each other, and the same or similar concept or process may be repeated no more in some embodiments.
Fig. 2 is the flow chart one of road disease recognition methods neural network based provided in an embodiment of the present invention, this reality
The executing subject of example is applied as the server 200 in Fig. 1.As shown in Fig. 2, the method for the present embodiment, may include:
S201: the road disease image that vehicle-mounted acquisition equipment is sent is obtained.
Wherein, road disease image includes but is not limited to the disease geo-radar image of pavement of road, road guard-rail etc..Such as road
Disease figure includes disease example, and disease example refers to the ontology of disease, including on pavement of road swag, on road guard-rail
Pit-hole and pavement of road on crack etc..
Acquisition equipment can make to look around picture pick-up device with 360 °.
In the present embodiment, 360 ° of the plurality of pictures for looking around picture pick-up device shooting can be spliced, is obtained a certain
The road disease image of position.Wherein road disease image may include disease or many places disease at one.
Specifically, it when being loaded with vehicle driving to a certain position of vehicle-mounted acquisition equipment, opens 360 ° of camera shooting of looking around and sets
It is standby to shoot multiple disease pictures, multiple disease pictures are spliced according to shooting sequence to the road disease figure for obtaining the position
Picture.
S202: the road disease image is input in preset convolutional neural networks model, the road is calculated
The corresponding disease class label of road disease geo-radar image, disease relative position and disease example cut zone, the disease relative position are
Position of the disease in road disease image, wherein the preset convolutional neural networks model is by preset quantity by marking
The training of road disease image obtain.
Wherein, preset convolutional neural networks model can be the trained depth convolution based on Mask R-CNN
Neural network model.
In one embodiment of the invention, road disease image can be pre-processed: will be in road disease image
Triple channel (BGR) subtract mean value threshold value, such as the BGR of mean value threshold value is respectively (103,116,123).By subtracting threshold value
Pretreatment is so that road disease image is easier to carry out process of convolution.
In the present embodiment, the road disease that multiple road disease images carry out disease classification and disease grade is obtained in advance
The label of image, and the road disease image of the label of preset quantity is chosen, for training initial convolutional neural networks model,
The road disease image of a part label is used to test the error precision of the convolutional neural networks model after training in remainder,
If error precision is met the requirements, the convolutional neural networks model after training is determined as preset convolutional neural networks model;
If error precision is met the requirements, continue the road disease image with another part label in remainder, to convolutional Neural net
Network model continues to train.
In one embodiment of the invention, the depth convolutional neural networks model based on Mask R-CNN is trained
Process are as follows: example calibration is carried out to a large amount of road disease images for having obtained, and utilizes a large amount of road diseases demarcated
Image is trained the initial depth convolutional neural networks model based on Mask R-CNN, obtains trained based on Mask
The depth convolutional neural networks model of R-CNN.
In the present embodiment, obtaining disease class label according to preset convolutional neural networks model can be one or more
A, usual disease class label the greater is the corresponding disease classification of road disease image.
In the present embodiment, disease relative position is position of the disease in road disease image, with reference to Fig. 3, such as institute
State disease relative position is indicated with a rectangle frame A comprising disease, and rectangle frame A is the maximum boundary rectangle frame on disease boundary.
S203: the corresponding disease classification of the road disease image is determined according to the disease class label.
In the present embodiment, disease class label is multiple, each disease class label corresponds to a kind of disease classification, by disease
Class label the maximum is determined as the corresponding disease classification of road disease image.
For example, the disease classification exported in step S202 and the corresponding relationship of disease class label are as shown in table 1.
1. the corresponding relationship of disease classification and disease class label
Disease classification | Swag on pavement of road | Pit-hole on road guard-rail | Crack on pavement of road |
Disease class label | 1.2 | 0.9 | 1.0 |
Reference table 1 will then be determined as it is found that since the corresponding value of the swag on pavement of road is maximum on pavement of road
The corresponding disease classification of road disease geo-radar image.
S204: the road disease image pair is determined according to the disease relative position or the disease example cut zone
The disease grade answered.
In the present embodiment, when disease classification is blocky disease, disease grade, disease can be determined according to disease area
Area is bigger, then disease is more serious, and disease higher grade.It, can be true according to disease length when disease classification is linear disease
Determine disease grade, disease is longer, then disease is more serious, and disease higher grade.
Road disease recognition methods neural network based provided in an embodiment of the present invention obtains vehicle-mounted acquisition equipment and sends
Road disease image, the road disease image is input in preset convolutional neural networks model, is calculated described
The corresponding disease class label of road disease image, disease relative position and disease example cut zone, the disease relative position
For position of the disease in road disease image, wherein the preset convolutional neural networks model is by preset quantity by marking
What the road disease image training of note obtained, the corresponding disease class of the road disease image is determined according to the disease class label
Not, corresponding disease of the road disease image etc. is determined according to the disease relative position or the disease example cut zone
Grade, can be by disease geo-radar image automatic identification disease classification and disease grade, and recognition efficiency is high, passes through preset convolutional Neural net
The accuracy of identification can be improved in network model, avoids the problem that manual identified inefficiency, there are security risks.
Below with reference to a specific embodiment, detailed description is based on preset convolutional neural networks model identification road disease
The process of evil image corresponding disease class label and disease example cut zone.
Fig. 4 is the flowchart 2 of road disease recognition methods neural network based provided in an embodiment of the present invention, such as Fig. 4
Shown, the method for the present embodiment may include:
S401: feature extraction is carried out to the road disease image, obtains multilayer feature image.
In the present embodiment, based on the resnet101 core network in preset convolutional neural networks model, to road disease
Evil image carries out feature extraction, obtains multilayer feature image.Characteristic image refers to the character pixel of road disease for identification,
The corresponding character pixel of each layer of characteristic image.
S402: the candidate frame of multilayer feature figure is extracted, the candidate frame is to be made of the boundary of disease.
In the present embodiment, it using FPN (feature pyramid networks, pyramid network) algorithm, extracts more
The candidate frame of layer characteristic pattern.Candidate frame is to be made of the boundary of disease, with reference to shown in Fig. 3 center A.
S403: pond is carried out to the candidate frame, the candidate frame of extraction is zoomed to the feature vector of default matrix size.
In the present embodiment, it is calculated using ROI Align (Region of interest Align, region of interest alignment)
Candidate frame is carried out pond by method.The candidate frame of extraction is zoomed to the feature vector of 7 × 7 fixed matrixs.
Described eigenvector: being input to the first predicted branches of the preset convolutional neural networks model by S404, meter
Calculation obtains the corresponding disease class label of the road disease image and disease relative position.
In the present embodiment, the first predicted branches are used to calculate the corresponding multiple diseases of road disease image according to feature vector
Evil class label, is denoted as scores;And disease relative position, it is denoted as rect.
Described eigenvector: being input to the second predicted branches of the preset convolutional neural networks model by S405, meter
Calculation obtains the corresponding disease example cut zone of the road disease image.
In the present embodiment, the second predicted branches are used to calculate the corresponding disease of road disease image according to feature vector real
Example cut zone, is denoted as mask.
With reference to Fig. 5, Fig. 5 is the flow chart of road disease recognition methods neural network based provided in an embodiment of the present invention
Three, in the present embodiment, the disease class label is multiple;Institute is determined according to the disease class label described in above-mentioned steps S203
The corresponding disease classification of road disease image is stated, is specifically included:
S501: the maximum value in the disease class label is obtained.
S502: if the maximum value in the disease class label is less than preset effective threshold value, determine the disease classification
Value is invalid, and re-executes the step of obtaining the road disease image that vehicle-mounted acquisition equipment is sent.
In the present embodiment, preset effective threshold value can according to need setting, it is preferred that preset effective threshold value is
0.9。
S503: if the maximum value in the disease class label is greater than or equal to preset effective threshold value, by the maximum
It is worth corresponding disease classification and is determined as the corresponding disease classification of the road disease image.
Through the foregoing embodiment it is found that by the way that effective threshold value is arranged, the only maximum value in the disease class label is big
When preset effective threshold value, the corresponding disease classification of the maximum value is just determined as the road disease image pair
The disease classification answered can be further improved the accuracy of the corresponding disease classification identification of road disease image.
It in one embodiment of the invention, is provided in an embodiment of the present invention neural network based with reference to Fig. 6, Fig. 6
The flow chart four of road disease recognition methods, the disease classification bulk disease, for example including road surface swag class disease, protective fence
Disease, road surface check crack disease and soundproof plate damage disease etc..It is true according to the disease example cut zone in above-mentioned steps S204
Determine the corresponding disease grade of the road disease image, specifically include:
S601: expanding the disease example cut zone, corrosion treatment, so that the disease example cut zone
As complete and connection a region.
In the present embodiment, with reference to Fig. 7, shadow region part is the disease example cut zone after expansion, corrosion treatment.
S602: the disease area of the disease example cut zone after calculation processing.
In the present embodiment, with reference to Fig. 7, the disease area of disease example cut zone, tool are calculated by calculus principle
Steps are as follows for body:
1) area for calculating the maximum boundary rectangle frame of disease example cut zone, is denoted as mask_area.
2) maximum boundary rectangle frame B is divided into multiple small rectangles, wherein two sections are divided by disease example cut zone,
I.e.WithAnd it calculates separatelyWithArea summation
3) the area object_area=mask_area-rect of disease example cut zone is calculatedΔx_area。
4) actual disease face is converted by the area object_area of disease example cut zone according to the ratio of shooting
Product.
S603: the corresponding disease grade of the road disease image is determined according to the disease area.
In the present embodiment, disease area is bigger, then disease is more serious, and disease grade is also higher.It can first disease area
Corresponding disease grade is obtained multiplied by preset disease coefficient.
For example, first disease area being calculated is 10 square meters, second disease area is 5 square meters, respectively multiplied by
Preset disease coefficient 0.2, then the grade of first disease is 10 × 0.02=2, and the grade of second disease is 5 × 0.02=
1, first disease is 1 times higher than the severity of second disease.
Through the foregoing embodiment it is found that being expanded first to disease example cut zone, corrosion treatment, so that the disease
Evil example cut zone becomes complete and connection a region, then the disease face of the disease example cut zone after calculation processing
Product, determines the corresponding disease grade of the road disease image according to disease area, can accurately determine road surface swag class disease
Or the disease severity of protective fence disease.
It in one embodiment of the invention, is provided in an embodiment of the present invention neural network based with reference to Fig. 8, Fig. 8
The flow chart five of road disease recognition methods, the disease classification are linear disease, for example, road surface crackle disease, the disease
Relative position is the maximum boundary rectangle frame on the disease boundary in the road disease image;According to institute in above-mentioned steps S204
It states disease relative position and determines the corresponding disease grade of the road disease image, specifically include:
S801: the catercorner length of the rectangle frame is obtained.
In the present embodiment, the maximum boundary rectangle frame on rectangle frame disease boundary, is indicated with the catercorner length of rectangle frame
The length of road surface crackle disease.I.e. catercorner length is bigger, and the length of road surface crackle disease is bigger, and disease is more serious.
S802: the corresponding disease grade of the road disease image is determined according to diagonal length.
In the present embodiment, can the corresponding diagonal length of first disease multiplied by preset disease coefficient obtain corresponding disease
Grade.
For example, the diagonal length for first disease being calculated is 20 meters, the diagonal length of second disease is 10 meters,
Respectively multiplied by preset disease coefficient 0.1, then the grade of first disease is 20 × 0.1=2, and the grade of second disease is 10
× 0.1=1, first disease are 1 times higher than the severity of second disease.
Through the foregoing embodiment it is found that passing through the diagonal line length for the maximum boundary rectangle frame that will calculate rectangle frame disease boundary
Degree, can rapidly obtain the disease grade of road surface crackle disease, compare and calculate complicated, irregular splitting in the prior art
Length, the method for the present embodiment is more rapidly, efficiently.
In one embodiment of the invention, on the basis of Fig. 2 corresponding embodiment, the method also includes:
While obtaining the road disease image that vehicle-mounted acquisition equipment is sent, the geography that vehicle positioning equipment is sent is obtained
Coordinate information;The road disease classification, disease grade and geographic coordinate information are sent to the terminal device of maintenance personnel.
In the present embodiment, the corresponding geographic coordinate information of road disease image can be obtained by positioning device, then
The corresponding road disease classification of road image, disease grade and the corresponding geographic coordinate information of road disease image are sent to
The terminal device of maintenance personnel, so that maintenance personnel believe according to obtained road disease classification, disease grade and geographical coordinate
Breath is in time handled road disease, improves treatment effeciency, guarantees that road safety is unimpeded.
Fig. 9 is the structural schematic diagram one of road disease identification device neural network based provided in an embodiment of the present invention,
As shown in figure 9, the road disease identification device 900 neural network based of the present embodiment, comprising: disease geo-radar image obtains module
901, Processing with Neural Network module 902, disease category determination module 903 and disease level determination module 904.
Wherein, disease geo-radar image obtains module 901, the road disease image sent for obtaining vehicle-mounted acquisition equipment;
Processing with Neural Network module 902, for the road disease image to be input to preset convolutional neural networks mould
In type, the corresponding disease class label of the road disease image, disease relative position and disease example cut zone is calculated,
The disease relative position is position of the disease in road disease image, wherein the preset convolutional neural networks model is
It is obtained by preset quantity by the road disease image training of label;
Disease category determination module 903, for determining that the road disease image is corresponding according to the disease class label
Disease classification;
Disease level determination module 904, for true according to the disease relative position or the disease example cut zone
Determine the corresponding disease grade of the road disease image.
Road disease identification device neural network based provided in an embodiment of the present invention obtains vehicle-mounted acquisition equipment and sends
Road disease image, the road disease image is input in preset convolutional neural networks model, is calculated described
The corresponding disease class label of road disease image, disease relative position and disease example cut zone, the disease relative position
For position of the disease in road disease image, wherein the preset convolutional neural networks model is by preset quantity by marking
What the road disease image training of note obtained, the corresponding disease class of the road disease image is determined according to the disease class label
Not, corresponding disease of the road disease image etc. is determined according to the disease relative position or the disease example cut zone
Grade, can be by disease geo-radar image automatic identification disease classification and disease grade, and recognition efficiency is high, passes through preset convolutional Neural net
The accuracy of identification can be improved in network model, avoids the problem that manual identified inefficiency, there are security risks.
Figure 10 is the structural schematic diagram two of road disease identification device neural network based provided in an embodiment of the present invention,
As shown in Figure 10, on the basis of embodiment shown in Fig. 9, the Processing with Neural Network module 902, comprising:
Characteristic image extraction unit 9021 obtains multilayer feature for carrying out feature extraction to the road disease image
Image;
Candidate frame extraction unit 9022, for extracting the candidate frame of multilayer feature figure, the candidate frame is to be by disease
Boundary composition;
Candidate frame pond unit 9023 zooms to the candidate frame of extraction default for carrying out pond to the candidate frame
The feature vector of matrix size;
First predicting unit 9024, for described eigenvector to be input to the preset convolutional neural networks model
The corresponding disease class label of the road disease image and disease relative position is calculated in first predicted branches;
Second predicting unit 9025, for described eigenvector to be input to the preset convolutional neural networks model
The corresponding disease example cut zone of the road disease image is calculated in second predicted branches.
In one embodiment of the invention, the disease class label is multiple;The disease category determination module 903,
Specifically for obtaining the maximum value in the disease class label, the corresponding disease classification of the maximum value is determined as the road
The corresponding disease classification of disease geo-radar image.
In one embodiment of the invention, the disease category determination module 903, if being also used to the disease class label
In maximum value be less than preset effective threshold value, then determine that the disease class label is invalid, and re-execute and obtain vehicle-mounted acquisition
The step of road disease image that equipment is sent.
In one embodiment of the invention, the disease classification bulk disease;The disease level determination module 904,
Specifically for being expanded to the disease example cut zone, corrosion treatment so that the disease example cut zone becomes
One complete and connection region;The disease area of disease example cut zone after calculation processing;According to the disease area
Determine the corresponding disease grade of the road disease image.
In one embodiment of the invention, the disease classification is linear disease, the disease relative position be positioned at
The maximum boundary rectangle frame on disease boundary in the road disease image;Disease level determination module 904 is specifically used for obtaining institute
State the catercorner length of rectangle frame;The corresponding disease grade of the road disease image is determined according to diagonal length.
With reference to Fig. 9, described device further include: geographic coordinate information obtains module 905 and sending module 906.
Wherein, geographic coordinate information obtains module 905, in the road disease image for obtaining vehicle-mounted acquisition equipment transmission
While, obtain the geographic coordinate information that vehicle positioning equipment is sent;
Sending module 906, for the road disease classification, disease grade and geographic coordinate information to be sent to maintenance people
The terminal device of member.
Figure 11 is the hardware configuration signal that road disease neural network based provided in an embodiment of the present invention identifies equipment
Figure.As shown in figure 11, road disease identification equipment 1000 neural network based provided in this embodiment includes: at least one
Manage device 1001 and memory 1002.The road disease identification equipment 1000 neural network based further includes communication component 1003.
Wherein, processor 1001, memory 1002 and communication component 1003 are connected by bus 1004.
During specific implementation, at least one processor 1001 executes the computer that the memory 1002 stores and executes
Instruction, so that at least one processor 1001 executes the knowledge of the road disease neural network based in any of the above-described embodiment of the method
Other method.Communication component 1003 with terminal device and/or server for being communicated.
The specific implementation process of processor 1001 can be found in above method embodiment, implementing principle and technical effect class
Seemingly, details are not described herein again for the present embodiment.
In the embodiment shown in above-mentioned Figure 11, it should be appreciated that processor can be central processing unit (English:
Central Processing Unit, referred to as: CPU), can also be other general processors, digital signal processor (English:
Digital Signal Processor, referred to as: DSP), graphics processor (English: Graphics Processing Unit,
Referred to as: GPU) specific integrated circuit (English: Application Specific Integrated Circuit, abbreviation: ASIC)
Deng.General processor can be microprocessor or the processor is also possible to any conventional processor etc..In conjunction with invention institute
The step of disclosed method, can be embodied directly in hardware processor and execute completion, or with the hardware and software mould in processor
Block combination executes completion.
Memory may include high speed RAM memory, it is also possible to and it further include non-volatile memories NVM, for example, at least one
Magnetic disk storage.
Bus can be industry standard architecture (Industry Standard Architecture, ISA) bus, outer
Portion's apparatus interconnection (Peripheral Component, PCI) bus or extended industry-standard architecture (Extended
Industry Standard Architecture, EISA) bus etc..Bus can be divided into address bus, data/address bus, control
Bus etc..For convenient for indicating, the bus in illustrations does not limit only a bus or a type of bus.
The embodiment of the present invention also provides a kind of computer readable storage medium, stores in the computer readable storage medium
There are computer executed instructions, when processor executes the computer executed instructions, realizes in any of the above-described embodiment of the method
Road disease recognition methods neural network based.
Above-mentioned computer readable storage medium, can be by any kind of volatibility or non-volatile memory device or
Person's their combination is realized, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM),
Erasable Programmable Read Only Memory EPROM (EPROM), programmable read only memory (PROM), read-only memory (ROM), magnetic storage
Device, flash memory, disk or CD.It is any available to can be that general or specialized computer can access for readable storage medium storing program for executing
Medium.
A kind of illustrative readable storage medium storing program for executing is coupled to processor, to enable a processor to from the readable storage medium storing program for executing
Information is read, and information can be written to the readable storage medium storing program for executing.Certainly, readable storage medium storing program for executing is also possible to the composition portion of processor
Point.Processor and readable storage medium storing program for executing can be located at specific integrated circuit (Application Specific Integrated
Circuits, referred to as: ASIC) in.Certainly, processor and readable storage medium storing program for executing can also be used as discrete assembly and be present in equipment
In.
Those of ordinary skill in the art will appreciate that: realize that all or part of the steps of above-mentioned each method embodiment can lead to
The relevant hardware of program instruction is crossed to complete.Program above-mentioned can be stored in a computer readable storage medium.The journey
When being executed, execution includes the steps that above-mentioned each method embodiment to sequence;And storage medium above-mentioned include: ROM, RAM, magnetic disk or
The various media that can store program code such as person's CD.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations;To the greatest extent
Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement;And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme.
Claims (16)
1. a kind of road disease recognition methods neural network based characterized by comprising
Obtain the road disease image that vehicle-mounted acquisition equipment is sent;
The road disease image is input in preset convolutional neural networks model, the road disease image is calculated
Corresponding disease class label, disease relative position and disease example cut zone, the disease relative position are disease in road
Position in disease geo-radar image, wherein the preset convolutional neural networks model is the road disease by preset quantity by label
Image training obtains;
The corresponding disease classification of the road disease image is determined according to the disease class label;
The corresponding disease of the road disease image is determined according to the disease relative position or the disease example cut zone
Grade.
2. the method according to claim 1, wherein described be input to preset volume for the road disease image
In product neural network model, the corresponding disease class label of the road disease image, disease relative position and disease is calculated
Example cut zone, comprising:
Feature extraction is carried out to the road disease image, obtains multilayer feature image;
The candidate frame of multilayer feature figure is extracted, the candidate frame is made of the boundary of disease;
Pond is carried out to the candidate frame, the candidate frame of extraction is zoomed to the feature vector of default matrix size;
Described eigenvector is input to the first predicted branches of the preset convolutional neural networks model, is calculated described
The corresponding disease class label of road disease image and disease relative position;
Described eigenvector is input to the second predicted branches of the preset convolutional neural networks model, is calculated described
The corresponding disease example cut zone of road disease image.
3. the method according to claim 1, wherein the disease class label is multiple;It is described according to the disease
Evil class label determines the corresponding disease classification of the road disease image, comprising:
The maximum value in the disease class label is obtained, the corresponding disease classification of the maximum value is determined as the road disease
The corresponding disease classification of image.
4. according to the method described in claim 3, it is characterized by further comprising:
If the maximum value in the disease class label is less than preset effective threshold value, determine that the disease class label is invalid, and
Re-execute the step of obtaining the road disease image that vehicle-mounted acquisition equipment is sent.
5. the method according to claim 1, wherein the disease classification is blocky disease;
The corresponding disease grade of the road disease image is determined according to the disease example cut zone, comprising:
The disease example cut zone is expanded, corrosion treatment, so that the disease example cut zone becomes one
Complete and connection region;
The disease area of disease example cut zone after calculation processing;
The corresponding disease grade of the road disease image is determined according to the disease area.
6. the disease is opposite the method according to claim 1, wherein the disease classification is linear disease
Position is the maximum boundary rectangle frame on the disease boundary in the road disease image;
The corresponding disease grade of the road disease image is determined according to the disease relative position, comprising:
Obtain the catercorner length of the rectangle frame;
The corresponding disease grade of the road disease image is determined according to diagonal length.
7. method according to any one of claims 1 to 6, which is characterized in that the method also includes:
While obtaining the road disease image that vehicle-mounted acquisition equipment is sent, the geographical coordinate that vehicle positioning equipment is sent is obtained
Information;
The road disease classification, disease grade and geographic coordinate information are sent to the terminal device of maintenance personnel.
8. a kind of road disease identification device neural network based characterized by comprising
Disease geo-radar image obtains module, the road disease image sent for obtaining vehicle-mounted acquisition equipment;
Processing with Neural Network module is counted for the road disease image to be input in preset convolutional neural networks model
Calculation obtains the corresponding disease class label of the road disease image, disease relative position and disease example cut zone, the disease
Evil relative position is position of the disease in road disease image, wherein the preset convolutional neural networks model is by presetting
Quantity is obtained by the road disease image training of label;
Disease category determination module, for determining the corresponding disease class of the road disease image according to the disease class label
Not;
Disease level determination module, for determining the road according to the disease relative position or the disease example cut zone
The corresponding disease grade of road disease geo-radar image.
9. the apparatus according to claim 1, which is characterized in that the Processing with Neural Network module, comprising:
Characteristic image extraction unit obtains multilayer feature image for carrying out feature extraction to the road disease image;
Candidate frame extraction unit, for extracting the candidate frame of multilayer feature figure, the candidate frame is to be made of the boundary of disease
's;
The candidate frame of extraction is zoomed to default matrix size for carrying out pond to the candidate frame by candidate frame pond unit
Feature vector;
First predicting unit, for described eigenvector to be input to the first prediction of the preset convolutional neural networks model
The corresponding disease class label of the road disease image and disease relative position is calculated in branch;
Second predicting unit, for described eigenvector to be input to the second prediction of the preset convolutional neural networks model
The corresponding disease example cut zone of the road disease image is calculated in branch.
10. the apparatus according to claim 1, which is characterized in that the disease class label is multiple;The disease classification is true
The corresponding disease classification of the maximum value is determined as by cover half block specifically for obtaining the maximum value in the disease class label
The corresponding disease classification of the road disease image.
11. device according to claim 10, which is characterized in that the disease category determination module, if being also used to described
Maximum value in disease class label is less than preset effective threshold value, then determines that the disease class label is invalid, and re-execute and obtain
It picks up the car and carries the step of acquiring the road disease image that equipment is sent.
12. the apparatus according to claim 1, which is characterized in that the disease classification bulk disease;The disease grade is true
Cover half block, specifically for being expanded to the disease example cut zone, corrosion treatment, so that the disease example cut section
Domain becomes complete and connection a region;The disease area of disease example cut zone after calculation processing;According to the disease
Evil area determines the corresponding disease grade of the road disease image.
13. the apparatus according to claim 1, which is characterized in that the disease classification is linear disease, and the disease is opposite
Position is the maximum boundary rectangle frame on the disease boundary in the road disease image;Disease level determination module, it is specific to use
In the catercorner length for obtaining the rectangle frame;The corresponding disease grade of the road disease image is determined according to diagonal length.
14. according to the described in any item devices of claim 8 to 13, which is characterized in that described device further include:
Geographic coordinate information obtains module, for obtaining while obtaining the road disease image that vehicle-mounted acquisition equipment is sent
The geographic coordinate information that vehicle positioning equipment is sent;
Sending module, for the road disease classification, disease grade and geographic coordinate information to be sent to the end of maintenance personnel
End equipment.
15. a kind of road disease neural network based identifies equipment characterized by comprising at least one processor and deposit
Reservoir;
The memory stores computer executed instructions;
At least one described processor executes the computer executed instructions of the memory storage, so that at least one described processing
Device executes road disease recognition methods neural network based as described in any one of claim 1 to 7.
16. a kind of computer readable storage medium, which is characterized in that be stored with computer in the computer readable storage medium
It executes instruction, when processor executes the computer executed instructions, realizes as described in any one of claim 1 to 7 be based on
The road disease recognition methods of neural network.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811534062.5A CN109685124A (en) | 2018-12-14 | 2018-12-14 | Road disease recognition methods neural network based and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811534062.5A CN109685124A (en) | 2018-12-14 | 2018-12-14 | Road disease recognition methods neural network based and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109685124A true CN109685124A (en) | 2019-04-26 |
Family
ID=66186094
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811534062.5A Pending CN109685124A (en) | 2018-12-14 | 2018-12-14 | Road disease recognition methods neural network based and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109685124A (en) |
Cited By (22)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110189317A (en) * | 2019-05-30 | 2019-08-30 | 上海卡罗网络科技有限公司 | A kind of road image intelligent acquisition and recognition methods based on deep learning |
CN110427911A (en) * | 2019-08-12 | 2019-11-08 | 清华大学苏州汽车研究院(吴江) | A kind of Approach for road detection, device, equipment and storage medium |
CN110992314A (en) * | 2019-11-15 | 2020-04-10 | 广东华路交通科技有限公司 | Pavement defect detection method and device and storage medium |
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 |
CN112598672A (en) * | 2020-11-02 | 2021-04-02 | 坝道工程医院(平舆) | Pavement disease image segmentation method and system based on deep learning |
CN112652015A (en) * | 2020-11-30 | 2021-04-13 | 中国公路工程咨询集团有限公司 | BIM-based pavement disease marking method and device |
CN113096517A (en) * | 2021-04-13 | 2021-07-09 | 北京工业大学 | Pavement damage intelligent detection trolley and sand table display system based on 5G and automatic driving |
CN113096126A (en) * | 2021-06-03 | 2021-07-09 | 四川九通智路科技有限公司 | Road disease detection system and method based on image recognition deep learning |
CN113240623A (en) * | 2021-03-18 | 2021-08-10 | 中国公路工程咨询集团有限公司 | Pavement disease detection method and device |
CN113255605A (en) * | 2021-06-29 | 2021-08-13 | 深圳市城市交通规划设计研究中心股份有限公司 | Pavement disease detection method and device, terminal equipment and storage medium |
CN113420621A (en) * | 2021-06-08 | 2021-09-21 | 广东省大湾区集成电路与系统应用研究院 | Pavement damage degree identification method, device, equipment and storage medium |
CN113505702A (en) * | 2021-07-12 | 2021-10-15 | 北京洞微科技发展有限公司 | Pavement disease identification method and system based on double neural network optimization |
CN113537037A (en) * | 2021-07-12 | 2021-10-22 | 北京洞微科技发展有限公司 | Pavement disease identification method, system, electronic device and storage medium |
CN113780200A (en) * | 2021-09-15 | 2021-12-10 | 安徽理工大学 | Computer vision-based pavement multi-disease area detection and positioning method |
CN113808098A (en) * | 2021-09-14 | 2021-12-17 | 丰图科技(深圳)有限公司 | Road disease identification method and device, electronic equipment and readable storage medium |
CN113963285A (en) * | 2021-09-09 | 2022-01-21 | 济南金宇公路产业发展有限公司 | Road maintenance method and equipment based on 5G |
CN115082802A (en) * | 2022-08-18 | 2022-09-20 | 深圳市城市交通规划设计研究中心股份有限公司 | Road disease identification method, device, equipment and readable storage medium |
CN115272300A (en) * | 2022-09-20 | 2022-11-01 | 中电信数字城市科技有限公司 | Pavement disease detection method, system, device, equipment and medium |
WO2022227405A1 (en) * | 2021-04-26 | 2022-11-03 | 深圳市商汤科技有限公司 | Road distress detection method and apparatus, electronic device, and storage medium |
CN117522175A (en) * | 2024-01-08 | 2024-02-06 | 中国公路工程咨询集团有限公司 | Road maintenance decision method and system |
WO2024060529A1 (en) * | 2022-09-23 | 2024-03-28 | 中路交科科技股份有限公司 | Pavement disease recognition method and system, device, and storage medium |
CN118334038A (en) * | 2024-06-17 | 2024-07-12 | 杭州可见智能技术有限公司 | Pavement disease analysis method, device, equipment, storage medium and related products |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104608799A (en) * | 2014-12-12 | 2015-05-13 | 郑州轻工业学院 | Information fusion technology based train wheel set tread damage online detection and recognition method |
CN106529593A (en) * | 2016-11-08 | 2017-03-22 | 广东诚泰交通科技发展有限公司 | Pavement disease detection method and system |
CN107316064A (en) * | 2017-06-26 | 2017-11-03 | 长安大学 | A kind of asphalt pavement crack classifying identification method based on convolutional neural networks |
CN108898085A (en) * | 2018-06-20 | 2018-11-27 | 安徽大学 | Intelligent road disease detection method based on mobile phone video |
-
2018
- 2018-12-14 CN CN201811534062.5A patent/CN109685124A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104608799A (en) * | 2014-12-12 | 2015-05-13 | 郑州轻工业学院 | Information fusion technology based train wheel set tread damage online detection and recognition method |
CN106529593A (en) * | 2016-11-08 | 2017-03-22 | 广东诚泰交通科技发展有限公司 | Pavement disease detection method and system |
CN107316064A (en) * | 2017-06-26 | 2017-11-03 | 长安大学 | A kind of asphalt pavement crack classifying identification method based on convolutional neural networks |
CN108898085A (en) * | 2018-06-20 | 2018-11-27 | 安徽大学 | Intelligent road disease detection method based on mobile phone video |
Non-Patent Citations (1)
Title |
---|
JANPREET SINGH 等: "Road Damage Detection And Classification In Smartphone Captured Images Using Mask R-CNN", 《ARXIV:1811.04535V1 [CS.CV]》 * |
Cited By (29)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110189317A (en) * | 2019-05-30 | 2019-08-30 | 上海卡罗网络科技有限公司 | A kind of road image intelligent acquisition and recognition methods based on deep learning |
CN110427911A (en) * | 2019-08-12 | 2019-11-08 | 清华大学苏州汽车研究院(吴江) | A kind of Approach for road detection, device, equipment and storage medium |
CN110427911B (en) * | 2019-08-12 | 2023-11-28 | 清华大学苏州汽车研究院(吴江) | Road detection method, device, equipment and storage medium |
CN110992314A (en) * | 2019-11-15 | 2020-04-10 | 广东华路交通科技有限公司 | Pavement defect detection method and device and storage medium |
CN110992314B (en) * | 2019-11-15 | 2024-07-26 | 广东华路交通科技有限公司 | Pavement defect detection method, device and storage medium |
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 |
CN112598672A (en) * | 2020-11-02 | 2021-04-02 | 坝道工程医院(平舆) | Pavement disease image segmentation method and system based on deep learning |
CN112652015A (en) * | 2020-11-30 | 2021-04-13 | 中国公路工程咨询集团有限公司 | BIM-based pavement disease marking method and device |
CN112652015B (en) * | 2020-11-30 | 2023-05-09 | 中国公路工程咨询集团有限公司 | BIM-based pavement disease marking method and device |
CN113240623A (en) * | 2021-03-18 | 2021-08-10 | 中国公路工程咨询集团有限公司 | Pavement disease detection method and device |
CN113240623B (en) * | 2021-03-18 | 2023-11-07 | 中国公路工程咨询集团有限公司 | Pavement disease detection method and device |
CN113096517A (en) * | 2021-04-13 | 2021-07-09 | 北京工业大学 | Pavement damage intelligent detection trolley and sand table display system based on 5G and automatic driving |
WO2022227405A1 (en) * | 2021-04-26 | 2022-11-03 | 深圳市商汤科技有限公司 | Road distress detection method and apparatus, electronic device, and storage medium |
CN113096126A (en) * | 2021-06-03 | 2021-07-09 | 四川九通智路科技有限公司 | Road disease detection system and method based on image recognition deep learning |
CN113420621A (en) * | 2021-06-08 | 2021-09-21 | 广东省大湾区集成电路与系统应用研究院 | Pavement damage degree identification method, device, equipment and storage medium |
CN113255605A (en) * | 2021-06-29 | 2021-08-13 | 深圳市城市交通规划设计研究中心股份有限公司 | Pavement disease detection method and device, terminal equipment and storage medium |
CN113537037A (en) * | 2021-07-12 | 2021-10-22 | 北京洞微科技发展有限公司 | Pavement disease identification method, system, electronic device and storage medium |
CN113505702A (en) * | 2021-07-12 | 2021-10-15 | 北京洞微科技发展有限公司 | Pavement disease identification method and system based on double neural network optimization |
CN113963285A (en) * | 2021-09-09 | 2022-01-21 | 济南金宇公路产业发展有限公司 | Road maintenance method and equipment based on 5G |
CN113963285B (en) * | 2021-09-09 | 2022-06-10 | 山东金宇信息科技集团有限公司 | Road maintenance method and equipment based on 5G |
CN113808098A (en) * | 2021-09-14 | 2021-12-17 | 丰图科技(深圳)有限公司 | Road disease identification method and device, electronic equipment and readable storage medium |
CN113780200A (en) * | 2021-09-15 | 2021-12-10 | 安徽理工大学 | Computer vision-based pavement multi-disease area detection and positioning method |
CN115082802A (en) * | 2022-08-18 | 2022-09-20 | 深圳市城市交通规划设计研究中心股份有限公司 | Road disease identification method, device, equipment and readable storage medium |
CN115272300A (en) * | 2022-09-20 | 2022-11-01 | 中电信数字城市科技有限公司 | Pavement disease detection method, system, device, equipment and medium |
WO2024060529A1 (en) * | 2022-09-23 | 2024-03-28 | 中路交科科技股份有限公司 | Pavement disease recognition method and system, device, and storage medium |
CN117522175A (en) * | 2024-01-08 | 2024-02-06 | 中国公路工程咨询集团有限公司 | Road maintenance decision method and system |
CN117522175B (en) * | 2024-01-08 | 2024-04-02 | 中国公路工程咨询集团有限公司 | Road maintenance decision method and system |
CN118334038A (en) * | 2024-06-17 | 2024-07-12 | 杭州可见智能技术有限公司 | Pavement disease analysis method, device, equipment, storage medium and related products |
CN118334038B (en) * | 2024-06-17 | 2024-08-27 | 杭州可见智能技术有限公司 | Pavement disease analysis method, device, equipment, storage medium and related products |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109685124A (en) | Road disease recognition methods neural network based and device | |
US20210319561A1 (en) | Image segmentation method and system for pavement disease based on deep learning | |
CN111144254A (en) | Urban waterlogging degree identification method and device | |
CN110163057B (en) | Object detection method, device, equipment and computer readable medium | |
CN113808098A (en) | Road disease identification method and device, electronic equipment and readable storage medium | |
CN115984273B (en) | Road disease detection method, device, computer equipment and readable storage medium | |
CN112966665A (en) | Pavement disease detection model training method and device and computer equipment | |
CN109916415B (en) | Road type determination method, device, equipment and storage medium | |
CN114299457A (en) | Accumulated water depth detection method and device | |
CN115439750A (en) | Road disease detection method and device, electronic equipment and storage medium | |
KR20190124436A (en) | Method for searching building based on image and apparatus for the same | |
CN115526846A (en) | Joint detection-based crack detection method and device and computer equipment | |
CN118172291B (en) | Image cloud removing method and device for remote sensing image and electronic equipment | |
CN117078624A (en) | Light-weight concrete crack detection method and device based on multidimensional attention module | |
CN104318216B (en) | Across the identification matching process of blind area pedestrian target in video monitoring | |
CN111553183A (en) | Ship detection model training method, ship detection method and ship detection device | |
CN111199539A (en) | Crack detection method based on integrated neural network | |
CN114639076A (en) | Target object detection method, target object detection device, storage medium, and electronic device | |
CN110046632B (en) | Model training method and device | |
CN107886049B (en) | Visibility recognition early warning method based on camera probe | |
CN111738324B (en) | Multi-frequency and multi-scale fusion automatic crack detection method based on frequency division convolution | |
CN111985497B (en) | Crane operation identification method and system under overhead transmission line | |
JP7180827B2 (en) | General object recognition system | |
CN110688978A (en) | Pedestrian detection method, device, system and equipment | |
CN118164196B (en) | Method and system for monitoring health state of coal conveying belt based on machine vision |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190426 |