CN114998713A - Pavement disease identification method, device, system, electronic equipment and storage medium - Google Patents
Pavement disease identification method, device, system, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN114998713A CN114998713A CN202210941820.5A CN202210941820A CN114998713A CN 114998713 A CN114998713 A CN 114998713A CN 202210941820 A CN202210941820 A CN 202210941820A CN 114998713 A CN114998713 A CN 114998713A
- Authority
- CN
- China
- Prior art keywords
- disease
- image
- road surface
- pavement
- information
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/54—Extraction of image or video features relating to texture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30132—Masonry; Concrete
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Processing (AREA)
Abstract
The invention relates to the technical field of computer vision, and provides a pavement disease identification method, a device, a system, electronic equipment and a storage medium. Moreover, the disease recognition model is obtained by training a disease pavement image sample carrying a pavement information label, the adopted disease pavement image sample comprises a real disease pavement image sample and a synthesized disease pavement image sample, and the synthesized disease pavement image sample is obtained by fusing a texture characteristic image and a healthy pavement image according to the apparent information of the texture characteristic image corresponding to the disease area in the real disease pavement image sample, so that the diversity of disease forms in the training sample can be improved, and the universality of the disease recognition model is further improved.
Description
Technical Field
The invention relates to the technical field of computer vision, in particular to a pavement damage identification method, a device, a system, electronic equipment and a storage medium.
Background
With the gradual improvement of road networks, the traffic capacity, technical conditions and service level of the existing roads are urgently needed to be maintained through reinforced maintenance.
Pavement disease identification is an important ring for road maintenance and is a key factor for determining the quality of road maintenance. At present, the pavement disease identification method mainly comprises an artificial visual detection method and an automatic identification method. The manual visual inspection method mainly realizes inspection in a mode of observing by human eyes of workers. In the automatic identification method, usually, a road image with an actual disease is used as a training sample to train an identification model, and the trained identification model is used for automatic identification.
In the existing method, the identification efficiency of the manual visual detection method is low, the identification rate of tiny damage is low, and the method has the problems of strong subjectivity, large workload, high human resource consumption and the like and has certain uncertainty. In the automatic identification method, because the road surface image with the actual disease is limited, various different disease forms cannot be covered, which leads to poor universality of the identification model.
Disclosure of Invention
The invention provides a pavement damage identification method, a device, a system, electronic equipment and a storage medium, which are used for overcoming the defects in the prior art.
The invention provides a pavement disease identification method, which comprises the following steps:
acquiring a road surface image to be identified;
inputting the pavement image to be identified into a disease identification model to obtain pavement information in the pavement image to be identified, which is output by the disease identification model, wherein the category of the pavement information comprises disease information;
the disease identification model is obtained by training based on a disease road surface image sample carrying a road surface information label, the disease road surface image sample comprises a real disease road surface image sample and a synthesized disease road surface image sample, and the synthesized disease road surface image sample is obtained by fusing a texture characteristic image and a healthy road surface image based on the apparent information of the texture characteristic image corresponding to a disease area in the real disease road surface image sample.
According to the pavement disease identification method provided by the invention, the determination method of the healthy pavement image comprises the following steps:
determining a non-diseased area image except the diseased area in the real diseased road image sample based on the road information label carried by the real diseased road image sample, wherein the non-diseased area image comprises a blank area corresponding to the diseased area;
and performing image restoration on the blank area based on the non-diseased area image to obtain the healthy pavement image.
According to the pavement disease identification method provided by the invention, the texture feature image determination method comprises the following steps:
extracting the disease area based on the disease information label carried by the real disease pavement image sample to obtain a disease area image;
and extracting texture features in the disease area image based on a local binary pattern algorithm to obtain the texture feature image.
According to the method for identifying the road surface diseases provided by the invention, the step of inputting the road surface image to be identified into a disease identification model to obtain the road surface information in the road surface image to be identified, which is output by the disease identification model, specifically comprises the following steps:
respectively inputting the pavement image to be recognized into a first feature extraction layer and a second feature extraction layer of the disease recognition model to obtain the detailed features of the pavement image to be recognized with the first resolution output by the first feature extraction layer and the semantic features of the pavement image to be recognized output by the second feature extraction layer;
inputting the detail features and the semantic features into a fusion layer of the disease identification model to obtain fusion features of the detail features and the semantic features output by the fusion layer;
and inputting the fusion characteristics to a classification layer of the disease identification model to obtain the pavement information output by the classification layer.
According to the method for identifying the pavement diseases, provided by the invention, the pavement image to be identified is input into a first feature extraction layer of the disease identification model, so that the detailed features of the pavement image to be identified with a first resolution ratio output by the first feature extraction layer are obtained, and the method specifically comprises the following steps:
adjusting the resolution of the road surface image to be identified to obtain a first image of the first resolution;
inputting the first image into the first feature extraction layer to obtain the detail features output by the first feature extraction layer;
and/or the presence of a gas in the gas,
inputting the road surface image to be recognized to a second feature extraction layer of the disease recognition model to obtain semantic features of the road surface image to be recognized, which are output by the second feature extraction layer, and the method specifically comprises the following steps:
adjusting the resolution of the road surface image to be identified to obtain a second image with a second resolution;
and inputting the second image into the second feature extraction layer to obtain the semantic features output by the second feature extraction layer.
According to the method for identifying the road surface diseases provided by the invention, the step of inputting the fusion characteristics into the classification layer of the disease identification model to obtain the road surface information output by the classification layer specifically comprises the following steps:
inputting the fusion features into the classification layer, performing pixel-by-pixel classification on the road surface image to be identified by the classification layer based on the fusion features to obtain the road surface information, determining a road surface color image corresponding to the road surface information, and outputting the road surface information and the road surface color image by the classification layer.
The present invention also provides a pavement damage identifying device, including:
the image acquisition module is used for acquiring a road surface image to be identified;
the disease identification module is used for inputting the road surface image to be identified into a disease identification model to obtain road surface information in the road surface image to be identified, which is output by the disease identification model, wherein the category of the road surface information comprises disease information;
the disease identification model is obtained by training based on a disease road surface image sample carrying a road surface information label, the disease road surface image sample comprises a real disease road surface image sample and a synthesized disease road surface image sample, and the synthesized disease road surface image sample is obtained by fusing a texture characteristic image and a healthy road surface image based on the apparent information of the texture characteristic image corresponding to a disease area in the real disease road surface image sample.
The invention also provides a pavement disease identification system, comprising: the device comprises an image acquisition device and the pavement damage recognition device, wherein the image acquisition device is connected with the pavement damage recognition device;
the image acquisition device is arranged on a vehicle running on the road surface to be identified and is used for acquiring the road surface image to be identified.
The invention also provides an electronic device, which comprises a memory, a processor and a computer program which is stored on the memory and can run on the processor, wherein when the processor executes the program, the pavement damage identification method is realized according to any one of the above methods.
The present invention also provides a non-transitory computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements a pavement disease identification method as recited in any of the above.
The present invention also provides a computer program product comprising a computer program which, when executed by a processor, implements a method of identifying a road surface condition as defined in any one of the above.
According to the pavement disease identification method, the device, the system, the electronic equipment and the storage medium, firstly, a pavement image to be identified is obtained; and then, inputting the road surface image to be identified into a disease identification model to obtain the road surface information in the road surface image to be identified, which is output by the disease identification model. Due to the adoption of the disease identification model, the automatic identification of the pavement diseases can be realized, the identification rate of the pavement diseases can be improved, the workload of workers is reduced, and the human resources are saved. Moreover, the disease identification model is obtained by training a disease pavement image sample carrying a pavement information label, the adopted disease pavement image sample comprises a real disease pavement image sample and a synthesized disease pavement image sample, and the synthesized disease pavement image sample is obtained by fusing a texture characteristic image and a healthy pavement image through the apparent information of the texture characteristic image corresponding to the disease area in the real disease pavement image sample, so that the diversity of disease forms in the training sample can be improved, and the universality of the disease identification model is further improved.
Drawings
In order to more clearly illustrate the present invention or the technical solutions in the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a schematic flow chart of a pavement damage identification method provided by the present invention;
FIG. 2 is a schematic diagram of a training process of a disease identification model in the pavement disease identification method provided by the invention;
FIG. 3 is a schematic structural diagram of a disease identification model in the pavement disease identification method provided by the invention;
FIG. 4 is a schematic view of a process flow of a fusion sublayer of a disease identification model in the pavement disease identification method provided by the invention;
FIG. 5 is a schematic structural diagram of a pavement damage recognition device provided by the present invention;
FIG. 6 is a schematic structural diagram of a pavement damage recognition system provided by the present invention;
fig. 7 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
At present, in the existing pavement disease identification method, the artificial vision detection method has the problems of strong subjectivity, large workload, high human resource consumption and the like, and has certain uncertainty. And the automatic identification method has poor universality. Therefore, the embodiment of the invention provides a pavement disease identification method.
Fig. 1 is a schematic flow chart of a pavement damage identification method provided in an embodiment of the present invention, and as shown in fig. 1, the method includes:
s1, acquiring a road surface image to be identified;
s2, inputting the road surface image to be identified into a disease identification model to obtain road surface information in the road surface image to be identified, which is output by the disease identification model, wherein the category of the road surface information comprises disease information;
the disease identification model is obtained by training based on a disease road surface image sample carrying a road surface information label, the disease road surface image sample comprises a real disease road surface image sample and a synthesized disease road surface image sample, and the synthesized disease road surface image sample is obtained by fusing a texture characteristic image and a healthy road surface image based on the apparent information of the texture characteristic image corresponding to a disease area in the real disease road surface image sample.
Specifically, in the pavement damage identification method provided in the embodiment of the present invention, the main execution body is a pavement damage identification device, the device may be configured in a server, the server may be a local server or a cloud server, the local server may specifically be a computer, a tablet computer, a smart phone, and the like, and the implementation of the method is not particularly limited in the embodiment of the present invention.
Step S1 is first executed to acquire a road surface image to be recognized. The road surface image to be identified is a road surface image of the road surface to be identified, the road surface to be identified may be a road surface with diseases or a healthy road surface, and the judgment needs to be performed according to the road surface image to be identified. The categories of the diseases may include cracks, protrusions, depressions, and the like.
The road surface image to be identified can be acquired by image acquisition devices such as a common camera, an industrial linear camera, a binocular camera and the like, and the image acquisition devices can be installed on street lamp poles and can also be installed on moving vehicles, and the road surface image to be identified is not specifically limited in the position.
And then, step 2 is executed, the road surface image to be identified is input into the disease identification model, the road surface image to be identified is identified through the disease identification model, and then the road surface information in the road surface image to be identified, which is output by the disease identification model, is obtained. Here, the road surface information is used to characterize the road surface health condition of the road surface to be identified, and the category thereof may include health information as well as disease information. The health information may include position information of a healthy area of the road surface to be identified, and the disease information may include position information of a diseased area of the road surface to be identified and disease category information corresponding to the diseased area.
In general, the health information of the road surface to be identified is not null, and the disease information may or may not be null. And if the pavement to be identified is a healthy pavement, the disease information is empty, and if the pavement to be identified is a pavement with diseases, the disease information is not empty.
In the embodiment of the invention, the adopted disease identification model can be obtained by training the initial model through the disease pavement image sample carrying the pavement information label. The initial model may be a neural network model, such as a deep convolutional neural network model, or may also be a dual-flow branch network combining a cross attention mechanism, which is not limited herein.
The adopted diseased pavement image samples can comprise real diseased pavement image samples and synthesized diseased pavement image samples, the real diseased pavement image samples refer to images of diseased pavement samples with actual diseases, which are acquired by an image acquisition device, and the images are not synthesized. The pavement information label carried by the real diseased pavement image sample can be obtained by performing pixel-level labeling on the real diseased pavement image sample, and can include health information and disease information in the real diseased pavement image sample. The road information label carried by the real diseased road image sample can be a semantic segmentation label, namely, the label comprises a healthy semantic region of the diseased road sample in the real diseased road image sample, position information of the diseased semantic region and a diseased semantic category corresponding to the diseased region.
The synthetic diseased pavement image sample is not acquired by an image acquisition device, but acquired by a real diseased pavement image sample and a healthy pavement image. When the composite diseased pavement image sample is determined, a diseased area in the real diseased pavement image sample can be determined firstly, and the diseased area can be determined through a pavement information label carried by the real diseased pavement image sample. And then, extracting the texture features of the disease area through a texture feature extraction algorithm to further determine a texture feature image corresponding to the disease area, wherein the texture feature image can be used for representing complete detail information of the disease area.
The healthy pavement image refers to an image of a healthy pavement sample without diseases, and the healthy pavement image can be an image of the healthy pavement sample acquired by the image acquisition device or an image of the healthy pavement sample with the diseases, and can also be acquired by processing the image sample of the real diseased pavement, so that the image acquisition pressure of the image acquisition device can be reduced.
The apparent information of the texture feature image can comprise information such as the size and the shape of the disease area, and the texture feature image and the healthy road surface image can be fused according to the apparent information to obtain a synthetic disease road surface image sample. The disease form can include information such as the position, size and shape of the disease, so that the disease form in the training sample can be greatly increased by synthesizing the disease pavement image sample.
And an image fusion algorithm can be adopted for fusion during fusion. The image fusion algorithm may be selected as desired, for example a poisson image fusion algorithm.
In the embodiment of the invention, the size of the texture feature image can be adjusted to obtain an adjusted image; and then, selecting a target fusion position on the healthy road image, and fusing the adjusted image to the target fusion position on the healthy road image by adopting an image fusion algorithm. The size of the texture feature image is not adjusted, but the texture feature image is directly fused to a target fusion position on the image of the healthy road surface by adopting an image fusion algorithm.
In addition, in order to further increase the disease forms in the training samples, an image fusion algorithm can be adopted for fusion, namely, the size of the texture feature image is adjusted firstly to obtain an adjusted image; then, selecting a random fusion position on the healthy road image, and fusing the adjusted image to the random fusion position on the healthy road image by adopting an image fusion algorithm; or, the size of the texture feature image is not adjusted, and the texture feature image is directly fused to a random fusion position on the healthy road image by adopting an image fusion algorithm.
In training the initial model, a back propagation algorithm can be adopted. The disease recognition model obtained by training can adopt a forward propagation principle to realize the whole process from the input of the road surface image to be recognized to the output of the road surface information.
The pavement disease identification method provided by the embodiment of the invention comprises the steps of firstly obtaining a pavement image to be identified; and then, inputting the road surface image to be identified into a disease identification model to obtain the road surface information in the road surface image to be identified, which is output by the disease identification model. Due to the adoption of the disease identification model, the automatic identification of the pavement diseases can be realized, the identification rate of the pavement diseases can be improved, the workload of workers is reduced, and the human resources are saved. Moreover, the disease recognition model is obtained by training a disease pavement image sample carrying a pavement information label, the adopted disease pavement image sample comprises a real disease pavement image sample and a synthesized disease pavement image sample, and the synthesized disease pavement image sample is obtained by fusing a texture characteristic image and a healthy pavement image according to the apparent information of the texture characteristic image corresponding to the disease area in the real disease pavement image sample, so that the diversity of disease forms in the training sample can be improved, and the universality of the disease recognition model is further improved.
On the basis of the above embodiment, in the pavement disease identification method provided in the embodiment of the present invention, the method for determining a healthy pavement image includes:
determining a non-diseased area image except the diseased area in the real diseased road image sample based on the road information label carried by the real diseased road image sample, wherein the non-diseased area image comprises a blank area corresponding to the diseased area;
and performing image restoration on the blank area based on the non-diseased area image to obtain the healthy pavement image.
Specifically, in the embodiment of the present invention, when determining the healthy road surface image, the non-damaged area image except the damaged area in the real damaged road surface image sample may be determined by using the road surface information label carried by the real damaged road surface image sample. For example, the damaged area in the real damaged pavement image sample can be removed based on the pavement information label, so that the damaged area in the real damaged pavement image sample becomes a blank area, and the non-damaged area image is obtained.
Then, image restoration can be performed on the blank area in the non-damaged area image according to the other areas except the blank area in the non-damaged area image, and the healthy pavement image can be obtained. The image restoration can be realized by adopting a Navier-Stokes image restoration method.
In the embodiment of the invention, the healthy pavement image can be obtained by directly removing the damaged area and repairing the image of the real damaged pavement image sample without adopting an image acquisition device for acquisition, so that the acquisition pressure of the image acquisition device can be reduced, the image transmission time delay between the image acquisition device and the pavement damage recognition device can be avoided, and the training efficiency of the damage recognition model is improved.
On the basis of the foregoing embodiment, in the pavement disease identification method provided in the embodiment of the present invention, the method for determining the texture feature image includes:
extracting the disease area based on the disease information label carried by the real disease pavement image sample to obtain a disease area image;
and extracting texture features in the disease area image based on a local binary pattern algorithm to obtain the texture feature image.
Specifically, in the embodiment of the present invention, when determining the texture feature image, the disease area in the real disease road surface image sample may be extracted according to the disease information tag carried by the real disease road surface image sample, so as to obtain the disease area image.
For example, the disease area in the real disease road surface image sample can be extracted based on the disease information label, and other areas except the disease area in the real disease road surface image sample can be deleted, so that the disease area image can be obtained. In this case, the apparent information such as the size and shape of the damaged area image is the same as the apparent information of the texture feature image, and the size and shape of the damage of the damaged road surface sample are also the same.
Then, according to a Local Binary Pattern (LBP) algorithm, extracting texture features in the disease area image to obtain a texture feature image. The LBP operator in the LBP algorithm may be one of a base LBP operator, a circle LBP operator, an equivalent pattern LBP operator, and a rotation invariant LBP operator. The LBP algorithm obtains a texture characteristic value of each pixel point in the disease area image through LBP operator calculation, and the texture characteristic value is used as the pixel value to obtain the texture characteristic image.
Taking the example of calculating the texture feature value by using a circular LBP operator, the sampling window is circular, and the radius may be one or more pixel points. In the running process of the algorithm, any pixel point in the image of the disease area can be used as the circle center c of the sampling window. Sampling is performed on the circumference of the sampling window, and P sampling points can be obtained. The value of P may be set as needed, and is not specifically limited herein, and may be, for example, 8.
For any sampling point p, if the sampling point p falls in a certain pixel point a in the image of the disease area, the pixel value of the sampling point p is the pixel value of the pixel point a; and if the sampling point p falls on the pixel point boundary in the image of the disease area, calculating the pixel value of the sampling point p through bilinear interpolation.
After pixel values of n sampling points are obtained, the pixel values are respectively compared with the pixel value of the circle center c, and if the pixel value of a certain sampling point is more than or equal to the pixel value of the circle center c, the pixel value is set to be 1; if the pixel value of a certain sampling point is smaller than the pixel value of the circle center c, the value is set to 0. Thus, a set of n-bit binary numbers is obtained and converted into a decimal number, i.e., an LBP code. The LBP code can be used as the texture feature value of the center c to reflect the texture information of the center c. The texture feature value of the circle center c can be expressed as:
wherein P represents the total number of sampling points on the circumference, R is the radius of the circumference, P is the current sampling point,
the pixel value representing the current sample point p,
the pixel value of the center of the circle is,
is the coordinate of the center c of the circle,
the texture feature value of the circle center c.
Order to
Then, then
The calculation formula of (a) is as follows:
in the embodiment of the invention, the texture characteristic image can be obtained by extracting the disease area and extracting the texture characteristic from the real disease pavement image sample, the texture characteristic image can represent the texture information of the disease area, and the texture information of the disease area in the synthetic disease pavement image sample obtained by fusion can be ensured to be contained.
On the basis of the foregoing embodiment, the method for identifying a road surface disease provided in an embodiment of the present invention includes that the road surface image to be identified is input to a disease identification model, and road surface information in the road surface image to be identified output by the disease identification model is obtained, and specifically includes:
respectively inputting the pavement image to be recognized to a first feature extraction layer and a second feature extraction layer of the disease recognition model to obtain detailed features of a first resolution of the pavement image to be recognized output by the first feature extraction layer and semantic features of the pavement image to be recognized output by the second feature extraction layer;
inputting the detail features and the semantic features into a fusion layer of the disease identification model to obtain fusion features of the detail features and the semantic features output by the fusion layer;
and inputting the fusion characteristics to a classification layer of the disease identification model to obtain the pavement information output by the classification layer.
Specifically, in the embodiment of the present invention, the disease identification model may include an input layer, a first feature extraction layer, a second feature extraction layer, a fusion layer, and a classification layer, where the input layer is connected to the first feature extraction layer and the second feature extraction layer, the first feature extraction layer and the second feature extraction layer are respectively connected to the fusion layer, and the fusion layer is connected to the classification layer. The input layer is used for receiving a road surface image to be recognized, the first feature extraction layer and the second feature extraction layer are used as double-flow branches and are respectively used for extracting detail features and semantic features of a first resolution ratio of the road surface image to be recognized, the fusion layer is used for fusing the detail features and the semantic features to obtain fusion features, and the classification layer is used for classifying the fusion features to further recognize road surface information in the road surface image to be recognized.
The first resolution may be a resolution having a value greater than a first preset threshold, i.e. a high resolution. The first preset threshold may be set as required, and may be, for example, 2K, which means a resolution level of up to 2000 pixels in horizontal resolution. The network structure of the first feature extraction layer may be the same as the HRNetw18 network structure, the network structure of the second feature extraction layer may be the same as the DeepLabV3 network structure based on ResNet-101, and the semantic features may be contextual semantic features.
In the fusion layer, the detail features and the semantic features can be subjected to multi-stage feature fusion through a pixel-by-pixel cross attention mechanism to obtain fusion features.
The classification layer may be composed of two convolution layers, and may determine road information in the road image to be recognized on a pixel-by-pixel basis.
In the embodiment of the invention, the feature extraction of the double-flow branch network is realized through the first feature extraction layer and the second feature extraction layer, different features can be extracted from the road image to be identified, the identification of the road information in the road image to be identified is facilitated, and the identification result can be more accurate.
On the basis of the foregoing embodiment, the method for identifying a road surface disease provided in the embodiment of the present invention is implemented by inputting the road surface image to be identified into a first feature extraction layer of the disease identification model, and obtaining a detailed feature of the road surface image to be identified with a first resolution output by the first feature extraction layer, and specifically includes:
adjusting the resolution of the road surface image to be identified to obtain a first image of the first resolution;
inputting the first image into the first feature extraction layer to obtain the detail features output by the first feature extraction layer;
and/or the presence of a gas in the gas,
inputting the road surface image to be recognized to a second feature extraction layer of the disease recognition model to obtain semantic features of the road surface image to be recognized, which are output by the second feature extraction layer, and the method specifically comprises the following steps:
adjusting the resolution of the road surface image to be identified to obtain a second image with a second resolution;
and inputting the second image into the second feature extraction layer to obtain the semantic features output by the second feature extraction layer.
Specifically, in the embodiment of the present invention, before the road surface image to be recognized is input to the first feature extraction layer, the resolution of the road surface image to be recognized may be adjusted to obtain the first image with the first resolution. After that, the first image can be input into the first feature extraction layer, and the detail features output by the first feature extraction layer can be obtained, so that the calculation amount of the first feature extraction layer can be reduced.
Similarly, before the road surface image to be recognized is input to the second feature extraction layer, the resolution of the road surface image to be recognized may also be adjusted to obtain a second image with a second resolution. After that, the second image can be input to the second feature extraction layer to obtain the semantic features output by the second feature extraction layer, so that the calculation amount of the second feature extraction layer can be reduced. Here, the second resolution may be a resolution less than a second preset threshold, i.e., a low resolution. The second preset threshold may be set as needed, and may be 512 × 512, for example.
On the basis of the foregoing embodiment, the method for identifying a road surface disease provided in the embodiment of the present invention is configured to input the fusion feature to a classification layer of the disease identification model to obtain the road surface information output by the classification layer, and specifically includes:
inputting the fusion features into the classification layer, performing pixel-by-pixel classification on the road surface image to be recognized by the classification layer based on the fusion features to obtain the road surface information, determining a road surface color image corresponding to the road surface information, and outputting the disease information and the road surface color image by the classification layer.
Specifically, in the embodiment of the present invention, after the fusion feature is input to the classification layer, the classification layer may perform pixel-by-pixel classification on the road surface image to be recognized according to the fusion feature to obtain the road surface information, where the road surface information may include road surface information corresponding to each pixel point in the road surface image to be recognized, and the road surface information corresponding to each pixel point may include that the pixel point corresponds to a healthy area, or include that the pixel point corresponds to a diseased area and a disease category corresponding to the diseased area.
For example, for any pixel point in the road surface image to be identified, if the pixel point corresponds to the healthy area, the road surface information corresponding to the pixel point can be represented by a first value; if the pixel point corresponds to the disease area and the disease category corresponding to the disease area is a crack, the pavement information corresponding to the pixel point can be represented by a second value; if the pixel point corresponds to the disease area and the disease category corresponding to the disease area is convex, the pavement information corresponding to the pixel point can be represented by a third value; by analogy, the description is omitted here. The first value, the second value, the third value, and the like may be set as needed, for example, may be set to 0, 1, 2, and the like, and may also be directly expressed in the form of RGB values for facilitating determination of a subsequent road color image.
And then, according to the first value, the second value, the third value and the like, determining a road surface color image corresponding to the road surface information, and outputting the disease information and the road surface color image by the classification layer. For example, if the first value, the second value, the third value, and the like are not related to the RGB values, the RGB values may be considered as gray values, and at this time, the RGB values corresponding to the gray values may be determined according to the preset correspondence between the gray values and the RGB values, and then the road color image may be obtained by using the RGB values as the pixel values of the road color image.
In the embodiment of the invention, the classification layer not only can output the road surface information, but also can output the road surface color image corresponding to the road surface information, so that a user can quickly determine the road surface information in the road surface image to be identified, and the visual experience of the user is improved.
Fig. 2 is a schematic diagram of a training process of a disease recognition model in the pavement disease recognition method provided in the embodiment of the present invention, and as shown in fig. 2, the method includes:
respectively obtaining first real disease pavement image samples carrying pavement information labels
And a second truly damaged pavement image sample
(ii) a The first real damaged pavement image sample and the second real damaged pavement image sample can be the same or different;
pavement information label carried by first damaged pavement image sample
Get rid of
Obtaining non-diseased area image
(ii) a Pavement information label carried by second damaged pavement image sample
Extracting of
Obtaining an image of the diseased area
；
To pair
The blank area in the image restoration device is subjected to image restoration to obtain a healthy pavement image
(ii) a Based on LBP algorithm, extracting
Obtaining the texture feature image
；
Through a Poisson image fusion algorithm, a healthy pavement image is obtained
And texture feature images
Fusing to obtain a synthetic disease pavement image sample
；
Pavement image sample based on first real diseases
And the second real disease pavement image sample
And synthesizing the damaged pavement image sample
And training the initial model to obtain a disease identification model.
Fig. 3 is a schematic structural diagram of a disease identification model in the pavement disease identification method provided in the embodiment of the present invention, and as shown in fig. 3, the disease identification model may include an input layer 1, a first feature extraction layer 2, a second feature extraction layer 3, a fusion layer 4, and a classification layer 5, where the input layer 1 is connected to the first feature extraction layer and the second feature extraction layer, the first feature extraction layer 2 and the second feature extraction layer 3 are respectively connected to the fusion layer 4, and the fusion layer 4 is connected to the classification layer 5. The input layer 1 is used for receiving a road surface image to be identified, the first feature extraction layer 2 comprises a plurality of first feature extraction sublayers which are connected in sequence, the number of channels of each first feature extraction sublayer is different or the number of downsamples of each first feature extraction sublayer is different, the second feature extraction layer 3 comprises a plurality of second feature extraction sublayers which are connected in sequence, and the number of channels of each second feature extraction sublayer is different or the number of downsamples of each second feature extraction sublayer is different.
The fusion layer 4 includes a plurality of fusion sublayers, the number of the first feature extraction sublayer, the second feature extraction sublayer, and the number of the fusion sublayers are all the same, and fig. 3 only shows the case where the number of the first feature extraction sublayer, the number of the second feature extraction sublayer, and the number of the fusion sublayers are all 5.
Each fusion sublayer is used for fusing the features output by the previous first feature extraction sublayer and the previous second feature extraction sublayer which are connected with each other, taking the obtained fusion result as the input of the next first feature extraction sublayer and the next second feature extraction sublayer, and taking the feature output by the last fusion sublayer as the fusion feature. In this way, the fusion layer 4 realizes multi-stage feature fusion of the detail features obtained by the first feature extraction layer 2 and the semantic features obtained by the second feature extraction layer 3 through a plurality of fusion sublayers, and obtains fusion features. Each fusion sub-layer performs a stage of feature fusion. And obtaining and outputting road information and road color images after the fusion features pass through the classification layer 5.
FIG. 4 shows the disease identification in the pavement disease identification method provided in the embodiment of the present inventionThe fusion sub-layer processing flow of the model is schematic, and the features in fig. 4 are all represented in the form of feature maps. As shown in fig. 4, the input of the ith fusion sublayer is the first feature map X output by the ith first feature extraction sublayer H1 And a second feature map X output by the ith second feature extraction sublayer C1 。
First characteristic diagram X H1 Size of c × H × W, second feature diagram X C1 The size is C × H × W, C and C are the number of feature map channels, H and H are the feature map height, and W and W are the feature map width. First characteristic diagram X H1 And a second characteristic diagram X C1 Respectively performing channel transformation on the convolution layers to obtain a third characteristic diagram X with the size of C × H × W H2 And a fourth feature map X of size Cxh w C2 . Third characteristic diagram X H2 And a fourth characteristic diagram X C2 After matrix multiplication, a first similarity matrix S with the size of HW × HW is obtained.
And performing column-by-column normalization (softmax) on the first similarity matrix S to obtain a normalized second similarity matrix S1. And carrying out line-by-line normalization on the first similarity matrix S to obtain a normalized third similarity matrix S2. The second similarity matrix S1 and the third feature map X H2 Matrix multiplication is carried out to obtain a fifth feature map X enhanced by detail features C3 Size of Cxh × w, fifth characteristic diagram X C3 Outputting a sixth feature diagram X enhanced by detail features after channel conversion through a layer of convolution layer of 1X1 C4 。
Third similarity matrix S2 and fourth feature map X C2 Matrix multiplication is carried out to obtain a seventh feature map X enhanced by semantic features H3 Size of C × H × W, seventh feature diagram X H3 Then, the eighth feature map X enhanced by the semantic features is output after the passage conversion through a layer of 1 multiplied by 1 convolution layer H4 。
In summary, the pavement disease identification method provided in the embodiment of the present invention has the following advantages: (1) a synthetic disease pavement image sample is obtained by fusing the texture feature image and the healthy pavement image, so that the identification generalization of the disease identification model to various pavements and various disease forms can be enhanced; (2) the detailed features of the first resolution ratio are extracted through the first feature extraction layer, so that the detection rate and accuracy of the disease identification model for the fine diseases can be enhanced; (3) semantic features are extracted through the second feature extraction layer, so that the semantic expression capability of the disease identification model on various diseases can be enhanced; (4) through a pixel-by-pixel cross attention mechanism, the detail expression capability and the semantic expression capability of the final output features of the disease identification model are improved, so that the prediction accuracy and the detection rate of the disease identification model are improved.
As shown in fig. 5, on the basis of the above embodiment, an embodiment of the present invention provides a road surface damage identifying device, including:
the image acquisition module 51 is used for acquiring a road surface image to be identified;
the disease identification module 52 is configured to input the road surface image to be identified to a disease identification model, so as to obtain road surface information in the road surface image to be identified, which is output by the disease identification model, where the category of the road surface information includes disease information;
the disease identification model is obtained by training on the basis of a disease pavement image sample carrying a pavement information label, the disease pavement image sample comprises a real disease pavement image sample and a synthesized disease pavement image sample, and the synthesized disease pavement image sample is obtained by fusing a texture characteristic image and a healthy pavement image on the basis of the apparent information of the texture characteristic image corresponding to a disease area in the real disease pavement image sample.
On the basis of the above embodiment, the road surface defect recognition apparatus provided in the embodiment of the present invention further includes a healthy road surface image determination module, configured to:
determining a non-diseased area image except the diseased area in the real diseased road image sample based on the road information label carried by the real diseased road image sample, wherein the non-diseased area image comprises a blank area corresponding to the diseased area;
and performing image restoration on the blank area based on the non-diseased area image to obtain the healthy pavement image.
On the basis of the foregoing embodiment, an embodiment of the present invention provides a road surface defect recognition apparatus, further including a texture feature image determination module, configured to:
extracting the disease area based on the disease information label carried by the real disease pavement image sample to obtain a disease area image;
and extracting texture features in the disease area image based on a local binary pattern algorithm to obtain the texture feature image.
On the basis of the foregoing embodiment, in the pavement damage identifying device provided in the embodiment of the present invention, the damage identifying module specifically includes:
the feature extraction submodule is used for respectively inputting the pavement image to be identified to a first feature extraction layer and a second feature extraction layer of the disease identification model to obtain the detail features of the first resolution of the pavement image to be identified output by the first feature extraction layer and the semantic features of the pavement image to be identified output by the second feature extraction layer;
the fusion submodule is used for inputting the detail features and the semantic features into a fusion layer of the disease identification model to obtain fusion features of the detail features and the semantic features output by the fusion layer;
and the classification submodule is used for inputting the fusion characteristics to a classification layer of the disease identification model to obtain the pavement information output by the classification layer.
On the basis of the foregoing embodiment, in the pavement disease recognition apparatus provided in the embodiment of the present invention, the feature extraction submodule is specifically configured to:
adjusting the resolution of the road surface image to be identified to obtain a first image of the first resolution;
inputting the first image into the first feature extraction layer to obtain the detail features output by the first feature extraction layer;
and/or the presence of a gas in the gas,
inputting the road surface image to be recognized to a second feature extraction layer of the disease recognition model to obtain semantic features of the road surface image to be recognized, which are output by the second feature extraction layer, and the method specifically comprises the following steps:
adjusting the resolution of the road surface image to be identified to obtain a second image with a second resolution;
and inputting the second image into the second feature extraction layer to obtain the semantic features output by the second feature extraction layer.
On the basis of the foregoing embodiment, in the pavement damage identifying device provided in the embodiment of the present invention, the classification submodule is specifically configured to:
inputting the fusion features into the classification layer, performing pixel-by-pixel classification on the road surface image to be identified by the classification layer based on the fusion features to obtain the road surface information, determining a road surface color image corresponding to the road surface information, and outputting the road surface information and the road surface color image by the classification layer.
Specifically, the functions of the modules in the pavement damage identifying device provided in the embodiment of the present invention correspond to the operation flows of the steps in the embodiments of the methods one to one, and the implementation effects are also consistent.
Fig. 6 is a schematic structural diagram of a pavement damage recognition system provided in an embodiment of the present invention, and as shown in fig. 6, the system includes: the road surface damage detection device comprises an image acquisition device 61 and a road surface damage identification device 62 provided in the embodiments, wherein the image acquisition device 61 is connected with the road surface damage identification device 62; the image acquisition device 61 is mounted on a vehicle running on the road surface to be identified, and is used for dynamically acquiring the road surface image to be identified. Therefore, the disease identification of the long-distance pavement to be identified can be realized.
Fig. 7 illustrates a physical structure diagram of an electronic device, and as shown in fig. 7, the electronic device may include: a Processor (Processor) 710, a communication Interface 720, a Memory (Memory) 730 and a communication bus 740, wherein the Processor 710, the communication Interface 720 and the Memory 730 communicate with each other via the communication bus 740. The processor 710 may call the logic instructions in the memory 730 to execute the pavement damage identification method provided in the foregoing embodiments, where the method includes: acquiring a road surface image to be identified; inputting the pavement image to be identified into a disease identification model to obtain pavement information in the pavement image to be identified, which is output by the disease identification model, wherein the category of the pavement information comprises disease information; the disease identification model is obtained by training based on a disease road surface image sample carrying a road surface information label, the disease road surface image sample comprises a real disease road surface image sample and a synthesized disease road surface image sample, and the synthesized disease road surface image sample is obtained by fusing a texture characteristic image and a healthy road surface image based on the apparent information of the texture characteristic image corresponding to a disease area in the real disease road surface image sample.
In addition, the logic instructions in the memory 730 can be implemented in the form of software functional units and stored in a computer readable storage medium when the software functional units are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, where the computer program product includes a computer program, the computer program may be stored on a non-transitory computer-readable storage medium, and when the computer program is executed by a processor, a computer can execute the pavement damage identification method provided by the above methods, and the method includes: acquiring a road surface image to be identified; inputting the pavement image to be recognized into a disease recognition model to obtain pavement information in the pavement image to be recognized, which is output by the disease recognition model, wherein the type of the pavement information comprises disease information; the disease identification model is obtained by training based on a disease road surface image sample carrying a road surface information label, the disease road surface image sample comprises a real disease road surface image sample and a synthesized disease road surface image sample, and the synthesized disease road surface image sample is obtained by fusing a texture characteristic image and a healthy road surface image based on the apparent information of the texture characteristic image corresponding to a disease area in the real disease road surface image sample.
In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, the computer program being implemented by a processor to execute the method for identifying a road surface defect provided by the above methods, the method including: acquiring a road surface image to be identified; inputting the pavement image to be identified into a disease identification model to obtain pavement information in the pavement image to be identified, which is output by the disease identification model, wherein the category of the pavement information comprises disease information; the disease identification model is obtained by training based on a disease road surface image sample carrying a road surface information label, the disease road surface image sample comprises a real disease road surface image sample and a synthesized disease road surface image sample, and the synthesized disease road surface image sample is obtained by fusing a texture characteristic image and a healthy road surface image based on the apparent information of the texture characteristic image corresponding to a disease area in the real disease road surface image sample.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A pavement disease identification method is characterized by comprising the following steps:
acquiring a road surface image to be identified;
inputting the pavement image to be identified into a disease identification model to obtain pavement information in the pavement image to be identified, which is output by the disease identification model, wherein the category of the pavement information comprises disease information;
the disease identification model is obtained by training based on a disease road surface image sample carrying a road surface information label, the disease road surface image sample comprises a real disease road surface image sample and a synthesized disease road surface image sample, and the synthesized disease road surface image sample is obtained by fusing a texture characteristic image and a healthy road surface image based on the apparent information of the texture characteristic image corresponding to a disease area in the real disease road surface image sample.
2. The method for identifying a road surface disease according to claim 1, characterized in that the method for determining the image of the healthy road surface includes:
determining a non-diseased area image except the diseased area in the real diseased road image sample based on the road information label carried by the real diseased road image sample, wherein the non-diseased area image comprises a blank area corresponding to the diseased area;
and performing image restoration on the blank area based on the non-diseased area image to obtain the healthy pavement image.
3. The method for identifying a pavement disease according to claim 1, wherein the method for determining the texture feature image includes:
extracting the disease area based on the disease information label carried by the real disease pavement image sample to obtain a disease area image;
and extracting texture features in the disease area image based on a local binary pattern algorithm to obtain the texture feature image.
4. The method for identifying a road surface disease according to any one of claims 1 to 3, wherein the step of inputting the road surface image to be identified into a disease identification model to obtain the road surface information in the road surface image to be identified output by the disease identification model specifically comprises the steps of:
respectively inputting the pavement image to be recognized to a first feature extraction layer and a second feature extraction layer of the disease recognition model to obtain detailed features of a first resolution of the pavement image to be recognized output by the first feature extraction layer and semantic features of the pavement image to be recognized output by the second feature extraction layer;
inputting the detail features and the semantic features into a fusion layer of the disease identification model to obtain fusion features of the detail features and the semantic features output by the fusion layer;
and inputting the fusion characteristics to a classification layer of the disease identification model to obtain the pavement information output by the classification layer.
5. The method for identifying the road surface diseases according to claim 4, wherein the step of inputting the road surface image to be identified into a first feature extraction layer of the disease identification model to obtain the detailed features of the road surface image to be identified with a first resolution output by the first feature extraction layer specifically comprises the steps of:
adjusting the resolution of the road surface image to be identified to obtain a first image of the first resolution;
inputting the first image into the first feature extraction layer to obtain the detail features output by the first feature extraction layer;
and/or the presence of a gas in the gas,
inputting the road surface image to be recognized to a second feature extraction layer of the disease recognition model to obtain semantic features of the road surface image to be recognized, which are output by the second feature extraction layer, and the method specifically comprises the following steps:
adjusting the resolution of the road surface image to be identified to obtain a second image with a second resolution;
and inputting the second image into the second feature extraction layer to obtain the semantic features output by the second feature extraction layer.
6. The method for identifying a road surface disease according to claim 4, wherein the inputting the fusion features into a classification layer of the disease identification model to obtain the road surface information output by the classification layer specifically comprises:
inputting the fusion features into the classification layer, performing pixel-by-pixel classification on the road surface image to be identified by the classification layer based on the fusion features to obtain the road surface information, determining a road surface color image corresponding to the road surface information, and outputting the road surface information and the road surface color image by the classification layer.
7. A pavement damage recognition device, comprising:
the image acquisition module is used for acquiring a road surface image to be identified;
the disease identification module is used for inputting the road surface image to be identified into a disease identification model to obtain road surface information in the road surface image to be identified, which is output by the disease identification model, wherein the category of the road surface information comprises disease information;
the disease identification model is obtained by training based on a disease road surface image sample carrying a road surface information label, the disease road surface image sample comprises a real disease road surface image sample and a synthesized disease road surface image sample, and the synthesized disease road surface image sample is obtained by fusing a texture characteristic image and a healthy road surface image based on the apparent information of the texture characteristic image corresponding to a disease area in the real disease road surface image sample.
8. A pavement disease recognition system, comprising: the pavement damage identification device of claim 7 and an image acquisition device connected to the pavement damage identification device;
the image acquisition device is arranged on a vehicle running on the road surface to be identified and is used for acquiring the road surface image to be identified.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the pavement damage identification method according to any one of claims 1 to 6 when executing the program.
10. A non-transitory computer-readable storage medium on which a computer program is stored, wherein the computer program, when executed by a processor, implements the method for identifying a road surface damage according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210941820.5A CN114998713B (en) | 2022-08-08 | 2022-08-08 | Pavement disease identification method, device and system, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210941820.5A CN114998713B (en) | 2022-08-08 | 2022-08-08 | Pavement disease identification method, device and system, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114998713A true CN114998713A (en) | 2022-09-02 |
| CN114998713B CN114998713B (en) | 2022-11-04 |
Family
ID=83022970
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210941820.5A Active CN114998713B (en) | 2022-08-08 | 2022-08-08 | Pavement disease identification method, device and system, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114998713B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116630899A (en) * | 2023-07-21 | 2023-08-22 | 四川公路工程咨询监理有限公司 | Highway side slope disease monitoring and early warning system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112330664A (en) * | 2020-11-25 | 2021-02-05 | 腾讯科技（深圳）有限公司 | Pavement disease detection method and device, electronic equipment and storage medium |
| CN113706472A (en) * | 2021-07-30 | 2021-11-26 | 中国公路工程咨询集团有限公司 | Method, device and equipment for detecting road surface diseases and storage medium |
| CN114120086A (en) * | 2021-10-29 | 2022-03-01 | 北京百度网讯科技有限公司 | Pavement disease recognition method, image processing model training method, device and electronic equipment |
| US20220237928A1 (en) * | 2020-11-02 | 2022-07-28 | BeSTDR Infrastructure Hospital(Pingyu) | Method for detecting road diseases by intelligent cruise via unmanned aerial vehicle, unmanned aerial vehicle and detecting system therefor |
-
2022
- 2022-08-08 CN CN202210941820.5A patent/CN114998713B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20220237928A1 (en) * | 2020-11-02 | 2022-07-28 | BeSTDR Infrastructure Hospital(Pingyu) | Method for detecting road diseases by intelligent cruise via unmanned aerial vehicle, unmanned aerial vehicle and detecting system therefor |
| CN112330664A (en) * | 2020-11-25 | 2021-02-05 | 腾讯科技（深圳）有限公司 | Pavement disease detection method and device, electronic equipment and storage medium |
| CN113706472A (en) * | 2021-07-30 | 2021-11-26 | 中国公路工程咨询集团有限公司 | Method, device and equipment for detecting road surface diseases and storage medium |
| CN114120086A (en) * | 2021-10-29 | 2022-03-01 | 北京百度网讯科技有限公司 | Pavement disease recognition method, image processing model training method, device and electronic equipment |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116630899A (en) * | 2023-07-21 | 2023-08-22 | 四川公路工程咨询监理有限公司 | Highway side slope disease monitoring and early warning system |
| CN116630899B (en) * | 2023-07-21 | 2023-10-20 | 四川公路工程咨询监理有限公司 | Highway side slope disease monitoring and early warning system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114998713B (en) | 2022-11-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109584248B (en) | Infrared target instance segmentation method based on feature fusion and dense connection network | |
| Kheradmandi et al. | A critical review and comparative study on image segmentation-based techniques for pavement crack detection | |
| CN108921799B (en) | Remote sensing image thin cloud removing method based on multi-scale collaborative learning convolutional neural network | |
| CN108122236B (en) | Iterative fundus image blood vessel segmentation method based on distance modulation loss | |
| CN108197326B (en) | Vehicle retrieval method and device, electronic equipment and storage medium | |
| CN109583345B (en) | Road recognition method, device, computer device and computer readable storage medium | |
| US11651477B2 (en) | Generating an image mask for a digital image by utilizing a multi-branch masking pipeline with neural networks | |
| CN112232349A (en) | Model training method, image segmentation method and device | |
| US11393100B2 (en) | Automatically generating a trimap segmentation for a digital image by utilizing a trimap generation neural network | |
| CN111462140B (en) | Real-time image instance segmentation method based on block stitching | |
| CN116797787B (en) | Remote sensing image semantic segmentation method based on cross-modal fusion and graph neural network | |
| CN113435407B (en) | Small target identification method and device for power transmission system | |
| Xu et al. | HA U-Net: Improved model for building extraction from high resolution remote sensing imagery | |
| CN114998713B (en) | Pavement disease identification method, device and system, electronic equipment and storage medium | |
| CN113269224A (en) | Scene image classification method, system and storage medium | |
| CN117197763A (en) | Road crack detection method and system based on cross attention guide feature alignment network | |
| CN116597270A (en) | Road damage target detection method based on attention mechanism integrated learning network | |
| CN115239644A (en) | Concrete defect identification method and device, computer equipment and storage medium | |
| CN113505702A (en) | Pavement disease identification method and system based on double neural network optimization | |
| CN113962878B (en) | Low-visibility image defogging model method | |
| Khoshboresh-Masouleh et al. | A deep multi-modal learning method and a new RGB-depth data set for building roof extraction | |
| Babu et al. | An efficient image dahazing using Googlenet based convolution neural networks | |
| CN114596503A (en) | Road extraction method based on remote sensing satellite image | |
| CN116977750B (en) | Construction method and classification method of land covering scene classification model | |
| CN116883650A (en) | Image-level weak supervision semantic segmentation method based on attention and local stitching |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |