CN116664582B - Road surface detection method and device based on neural vision network - Google Patents
Road surface detection method and device based on neural vision network Download PDFInfo
- Publication number
- CN116664582B CN116664582B CN202310960793.0A CN202310960793A CN116664582B CN 116664582 B CN116664582 B CN 116664582B CN 202310960793 A CN202310960793 A CN 202310960793A CN 116664582 B CN116664582 B CN 116664582B
- Authority
- CN
- China
- Prior art keywords
- model
- detection
- parameters
- detection model
- road surface
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 349
- 230000001537 neural effect Effects 0.000 title claims abstract description 47
- 230000007613 environmental effect Effects 0.000 claims abstract description 80
- 238000012360 testing method Methods 0.000 claims description 74
- 238000012549 training Methods 0.000 claims description 34
- 238000012423 maintenance Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 18
- 239000000463 material Substances 0.000 claims description 16
- 230000008439 repair process Effects 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 8
- 238000000034 method Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 5
- 238000003062 neural network model Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 239000004568 cement Substances 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- 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
- 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
- G06T2207/30168—Image quality inspection
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Abstract
The application discloses a road surface detection method and device based on a neural vision network, and relates to the technical field of neural networks; the road surface detection method based on the neural vision network comprises the following steps: acquiring a plurality of pavement images corresponding to a target pavement; determining environmental parameters corresponding to the plurality of road surface images respectively based on a first detection model; updating model parameters of a second detection model based on environment parameters respectively corresponding to the plurality of road surface images, model parameters of the first detection model and a preset model association relationship so as to determine an updated second detection model; and determining a breakage detection result corresponding to the target pavement based on the updated second detection model and the plurality of pavement images. It can realize accurate and effective road surface detection.
Description
Technical Field
The application relates to the technical field of neural networks, in particular to a road surface detection method and device based on a neural vision network.
Background
With the development of neural network technology, neural vision networks have also developed. The neural vision network belongs to one of the neural network models, and can realize the functions of object recognition, object classification and the like.
In the traffic field, in order to ensure stable and safe use of a road surface, maintenance of the road surface is required frequently, and the maintenance of the road surface can be realized based on detection of the road surface. The damage degree of the road surface can be detected according to the detection of the road surface, and if the damage degree of the road surface is higher, the road surface needs to be maintained. At present, the detection scheme of the road surface can be realized on the basis of a neural network model; however, the accuracy of the road surface detection result is difficult to be ensured.
Disclosure of Invention
The application aims to provide a road surface detection method and device based on a neural vision network, which can realize accurate and effective road surface detection.
In order to achieve the above object, an embodiment of the present application provides a road surface detection method based on a neural vision network, including: acquiring a plurality of pavement images corresponding to a target pavement; determining environmental parameters corresponding to the plurality of road surface images respectively based on a first detection model; the first detection model is a model based on a neural vision network; updating model parameters of a second detection model based on corresponding environment parameters of the plurality of pavement images, model parameters of the first detection model and a preset model parameter association relation so as to determine an updated second detection model; the second detection model corresponds to initial model parameters, and the preset model parameter association relationship is used for representing association relationships among the model parameters of the first detection model, the model parameters of the second detection model and environment parameters; and determining a breakage detection result corresponding to the target pavement based on the updated second detection model and the plurality of pavement images.
In one possible implementation manner, the acquiring a plurality of road surface images corresponding to the target road surface includes: determining road surface information of the target road surface; the road surface information includes: at least one of road surface material, road surface use frequency, and road surface maintenance frequency; determining an acquisition strategy of the plurality of road surface images based on the road surface information of the target road surface; the acquisition strategy comprises the following steps: at least one item of information of image acquisition period, image acquisition quantity and image quality; and acquiring a plurality of pavement images corresponding to the target pavement based on the acquisition strategy.
In one possible implementation manner, the determining the acquisition strategy of the plurality of road surface images based on the road surface information of the target road surface includes: determining the image acquisition period based on the road surface use frequency and a preset first association relation; the first association relationship is used for representing the association relationship between the road surface use frequency and the image acquisition period; determining the image acquisition quantity based on the road surface maintenance frequency, the road surface use frequency and a preset second association relation; the second association relationship is used for representing association relationships among the road maintenance frequency, the road use frequency and the image acquisition quantity; determining the image quality based on the pavement material and a preset third association relation; the third association relationship is used for representing the association relationship between the pavement material and the image quality.
In one possible implementation manner, the road surface detection method based on the neural vision network further comprises the following steps: acquiring a first training data set; the first training data set comprises a plurality of sample pavement images, and the plurality of sample pavement images respectively correspond to environmental parameter labels; training an initial first detection model according to the first training data set to obtain the first detection model; wherein initial model parameters of the initial first detection model are determined based on initial model parameters of the second detection model.
In one possible implementation manner, the road surface detection method based on the neural vision network further comprises the following steps: determining the model precision of the first detection model according to a first test data set; the first test data set comprises a plurality of test pavement images, and the plurality of test pavement images respectively correspond to environmental parameter labels; determining the model precision of the second detection model according to the second test data set; the second test data set comprises a plurality of test pavement images, and the plurality of test pavement images respectively correspond to breakage detection result labels; and determining the preset model parameter association relation according to the precision of the first detection model, the precision of the second detection model and the environment parameter labels and damage detection result labels respectively corresponding to the plurality of test pavement images.
In a possible implementation manner, the determining the preset association relationship of the model parameters according to the accuracy of the first detection model, the accuracy of the second detection model, and the environmental parameter labels and the damage detection result labels corresponding to the plurality of test pavement images respectively includes: if the precision of the first detection model is greater than or equal to that of the second detection model, determining a first influence value of the environmental parameter labels respectively corresponding to the plurality of test pavement images on the damage detection result labels respectively corresponding to the plurality of test pavement images; determining a first impact weight of an environmental parameter on a model parameter of the second detection model based on the first impact value; determining a second impact weight based on a precision difference between the precision of the first detection model and the precision of the second detection model; determining influence weight values of environment parameter labels corresponding to the plurality of test pavement images respectively based on the first influence weight and the second influence weight; and determining the preset model parameter association relation according to the influence weight value and the environment parameter labels respectively corresponding to the plurality of test pavement images.
In a possible implementation manner, the determining the preset association relationship of the model parameters according to the accuracy of the first detection model, the accuracy of the second detection model, and the environmental parameter labels and the damage detection result labels corresponding to the plurality of test pavement images respectively includes: if the precision of the first detection model is smaller than that of the second detection model, determining a target environment parameter label from environment parameter labels respectively corresponding to the plurality of test pavement images; the image quality of the test pavement image corresponding to the target environment parameter label is higher than the preset image quality; determining second influence values of the target environment parameter labels on damage detection result labels corresponding to the plurality of test pavement images respectively; determining a third influence weight of the target environmental parameter label on the model parameters of the second detection model based on the second influence value; determining a fourth impact weight based on a precision difference between the precision of the first detection model and the precision of the second detection model; determining an impact weight value of the target environmental parameter tag based on the third impact weight and the fourth impact weight; and determining the preset model parameter association relation according to the influence weight value and the target environment parameter label.
In one possible implementation manner, the preset association relationship of the model parameters includes a plurality of environmental parameters, where the environmental parameters respectively correspond to the influence weight values, and includes the association relationship among the model parameters; the updating the model parameters of the second detection model based on the environmental parameters respectively corresponding to the plurality of road surface images, the model parameters of the first detection model and the preset model parameter association relation comprises the following steps: determining target environmental parameters in the environmental parameters corresponding to the plurality of pavement images respectively; the target environment parameters have matched environment parameters in the preset model association relation; updating corresponding model parameters in the second detection model based on the influence weight value corresponding to the target environment parameter aiming at the same model parameters in the first detection model and the second detection model; and updating corresponding model parameters of the second detection model based on the association relation between the different model parameters based on the influence weight value corresponding to the target environment parameter aiming at the different model parameters in the first detection model and the second detection model.
In one possible embodiment, the damage detection result includes damage information, where the damage information includes at least one of damage area, damage degree and damage repair rate, and the road surface detection method based on the neural vision network further includes: generating a second training data set based on the plurality of road surface images, the environmental parameters respectively corresponding to the plurality of road surface images and the damage information; training the initial third detection model based on the second training data set to obtain a trained third detection model; model parameters of the initial third detection model are determined based on model parameters of the first detection model and model parameters of the updated second detection model.
The embodiment of the application also provides a road surface detection device based on the neural vision network, which comprises: the acquisition unit is used for acquiring a plurality of pavement images corresponding to the target pavement; a processing unit for: determining environmental parameters corresponding to the plurality of road surface images respectively based on a first detection model; the first detection model is a model based on a neural vision network; updating model parameters of a second detection model based on corresponding environment parameters of the plurality of pavement images, model parameters of the first detection model and a preset model parameter association relation so as to determine an updated second detection model; the second detection model corresponds to initial model parameters, and the preset model parameter association relationship is used for representing association relationships among the model parameters of the first detection model, the model parameters of the second detection model and environment parameters; and determining a breakage detection result corresponding to the target pavement based on the updated second detection model and the plurality of pavement images.
Compared with the prior art, the road surface detection method and device based on the neural vision network, provided by the embodiment of the application, have the advantages that after the road surface images of the road surface to be detected are obtained, the environment parameters corresponding to the road surface images are determined, and the parameters of the neural vision network for detecting the road surface damage are optimized based on the environment parameters. The road surface detection scheme not only utilizes the multi-layer network to realize road surface detection, but also correlates model parameters of the multi-layer network with each other and environmental parameters; therefore, the road surface detection scheme not only can realize effective detection of the road surface based on the neural vision network; the model parameters of the neural vision network can be optimized based on the relevance among the neural vision networks, so that more accurate pavement detection is realized. Thus, the pavement detection scheme can realize accurate and effective pavement detection.
Drawings
Fig. 1 is a schematic diagram of a road surface detection system according to an embodiment of the present application;
FIG. 2 is a flow chart of a road surface detection method based on a neural vision network, according to an embodiment of the present application;
fig. 3 is a schematic structural view of a road surface detection device based on a neural vision network according to an embodiment of the present application;
Fig. 4 is a schematic structural view of a terminal device according to an embodiment of the present application.
Description of the embodiments
The following detailed description of embodiments of the application is, therefore, to be taken in conjunction with the accompanying drawings, and it is to be understood that the scope of the application is not limited to the specific embodiments.
Throughout the specification and claims, unless explicitly stated otherwise, the term "comprise" or variations thereof such as "comprises" or "comprising", etc. will be understood to include the stated element or component without excluding other elements or components.
The technical scheme provided by the embodiment of the application can be applied to various application scenes in which pavement detection is required, and in the application scenes, the damage condition of the pavement is required to be detected, and the method is not limited to the pavement; that is, the technical scheme can be suitable for detection of various road surfaces.
In the traffic field, in order to ensure stable and safe use of a road surface, maintenance of the road surface is required frequently, and the maintenance of the road surface can be realized based on detection of the road surface. The damage degree of the road surface can be detected according to the detection of the road surface, and if the damage degree of the road surface is higher, the road surface needs to be maintained. At present, the detection scheme of the road surface can be realized on the basis of a neural network model; however, the accuracy of the road surface detection result is difficult to be ensured.
Based on the above, the embodiment of the application provides a pavement detection scheme, which not only utilizes a multi-layer network to realize pavement detection, but also correlates model parameters of the multi-layer network with each other and environmental parameters; therefore, the road surface detection scheme not only can realize effective detection of the road surface based on the neural vision network; the model parameters of the neural vision network can be optimized based on the relevance among the neural vision networks, so that more accurate pavement detection is realized. Thus, the pavement detection scheme can realize accurate and effective pavement detection.
Referring next to fig. 1, a schematic structural diagram of a pavement detection system includes: image acquisition equipment and terminal equipment; the image acquisition equipment is arranged at the position of the road surface to be detected, the terminal equipment can be used as a monitoring end of the road surface, and the image acquisition equipment and the terminal equipment are connected in a communication manner.
In some embodiments, the image acquisition device may be configured according to different road conditions, so that the image acquisition device is more suitable for a corresponding road, acquires a better image, and improves the accuracy of a final road detection result.
In some embodiments, the terminal device is used as a data processing end, which may be implemented in a server, a cloud end, or the like, which is not limited herein.
Referring next to fig. 2, a flowchart of a road surface detection method based on a neural vision network according to an embodiment of the present application includes:
step 201, a plurality of road surface images corresponding to a target road surface are acquired.
In some embodiments, step 201 comprises: determining road surface information of a target road surface; the road surface information includes: at least one of road surface material, road surface use frequency, and road surface maintenance frequency; determining a collection strategy of a plurality of road surface images based on road surface information of a target road surface; the acquisition strategy comprises the following steps: at least one item of information of image acquisition period, image acquisition quantity and image quality; based on the acquisition strategy, a plurality of pavement images corresponding to the target pavement are acquired.
In some embodiments, pavement material refers to constituent materials of pavement, such as: cement, concrete, etc.; the road surface use frequency may refer to a vehicle passing frequency, a pedestrian passing frequency, and the like; the road maintenance frequency may refer to the frequency with which the relevant personnel maintain the road.
It will be appreciated that the road surface information may be determined by counting the historical conditions of the target road surface and then stored in a corresponding database, and the information may be obtained directly when required.
In some embodiments, determining an acquisition strategy for a plurality of road surface images based on road surface information of a target road surface includes: determining an image acquisition period based on the road surface use frequency and a preset first association relation; the first association relationship is used for representing the association relationship between the road surface use frequency and the image acquisition period; determining the image acquisition quantity based on the road maintenance frequency, the road use frequency and a preset second association relation; the second association relationship is used for representing association relationships among the road surface maintenance frequency, the road surface use frequency and the image acquisition quantity; determining image quality based on pavement materials and a preset third association relation; the third association relationship is used for representing the association relationship between the pavement material and the image quality.
In some embodiments, the first association, the second association, and the third association may be configured according to different application scenarios.
In some embodiments, the higher the road surface frequency of use, the shorter the image acquisition period; the lower the road surface frequency of use, the longer the image acquisition period.
In some embodiments, if the frequency of road maintenance and the frequency of road use are both high, the number of image acquisitions may be small; if the road surface maintenance frequency is higher and the road surface use frequency is lower, the image acquisition quantity can be smaller; if the road surface maintenance frequency is low and the road surface use frequency is high, the image acquisition quantity can be more; if the road maintenance frequency and the road use frequency are both lower, the image acquisition number can be smaller.
In some embodiments, the respective image quality of the different road surface materials may be preset, such that the respective image quality is determined based on the current road surface material.
In other embodiments, the first association relationship, the second association relationship, and the third association relationship may also be implemented in other embodiments, which are not limited herein.
Further, after the acquisition strategies of the plurality of road surface images are determined, image acquisition is performed according to the acquisition strategies of the road surface images, so that the plurality of road surface images corresponding to the target road surface can be acquired.
In some embodiments, a plurality of image capturing devices may be configured, and different image capturing devices respectively correspond to different image capturing quality, image capturing number, image capturing period, and the like, so that an image capturing policy may be executed using these image capturing devices.
Step 202, determining environmental parameters corresponding to the plurality of road surface images respectively based on the first detection model. The first detection model is a model based on a neural vision network.
In some embodiments, the environmental parameters corresponding to the plurality of road surface images respectively may be understood as environmental parameters of the road surface in the road surface image, for example: illumination intensity, surrounding occlusions, weather, etc., there may be information about these parameters that affect the detection of the road surface.
In some embodiments, the first detection model may be trained in advance so that the first detection model may enable detection of these environmental parameters.
As an alternative embodiment, the pavement detection method further includes: acquiring a first training data set; the first training data set comprises a plurality of sample pavement images, and the plurality of sample pavement images are respectively corresponding to environmental parameter labels; training an initial first detection model according to the first training data set to obtain a first detection model; wherein initial model parameters of the initial first detection model are determined based on initial model parameters of the second detection model.
In some embodiments, the environmental parameter labels corresponding to the plurality of sample road surface images are manually labeled environmental parameter labels.
In some embodiments, some implementations may be employed to improve model accuracy during training of the initial first detection model. For example, a training number is preset, and model training is performed based on the training number; for another example, a test data set is preset, and a trained model is optimized based on the test accuracy obtained by the test data set.
In some embodiments, the first detection model and the second detection model are the same type of model, i.e., are both vision neural network models. For both the first detection model and the second detection model, there are corresponding model parameters, based on which the first detection model and the second detection model can be associated first.
In some embodiments, the first detection model and the second detection model may preset a common model parameter, where the model parameter is used to correlate the first detection model and the second detection model, and the model parameter may be a key parameter in a model algorithm corresponding to the detection model.
In some embodiments, the initial model parameters of the second detection model may be pre-tested parameters of higher accuracy; accordingly, the initial model parameters of the first detection model may be parameters that are slightly less accurate than the model parameters of the first detection model. That is, the initial parameter accuracy of the first detection model is lower than the initial parameter accuracy of the second detection model.
Therefore, when the initial model parameters of the first detection model are determined, only the accuracy lower than that of the initial model parameters of the second detection model needs to be ensured.
Thus, for the first detection model, after training, the environmental parameters respectively corresponding to the plurality of road surface images can be determined based on the plurality of road surface images.
Step 203, updating the model parameters of the second detection model based on the environmental parameters, the model parameters of the first detection model and the preset model parameter association relation corresponding to the plurality of road surface images respectively, so as to determine an updated second detection model.
In some embodiments, the second detection model may be trained in advance with reference to the training pattern of the first detection model, such that the trained second detection model may determine the road surface detection result based on the road surface image.
Further, in step 203, the model parameters of the second detection model are optimized based on the environmental parameters corresponding to the plurality of road surface images, and then the road surface detection is implemented by using the optimized second detection model (i.e., the updated second detection model).
The second detection model corresponds to initial model parameters, and a preset model parameter association relationship is used for representing association relationships among model parameters of the first detection model, model parameters of the second detection model and environment parameters.
As an optional embodiment, the road surface detection method based on the neural vision network further includes: determining the model precision of the first detection model according to the first test data set; the first test data set comprises a plurality of test pavement images, and the plurality of test pavement images respectively correspond to environmental parameter labels; determining the model accuracy of the second detection model according to the second test data set; the second test data set comprises a plurality of test pavement images, and the plurality of test pavement images respectively correspond to breakage detection result labels; and determining a preset model parameter association relation according to the precision of the first detection model, the precision of the second detection model and the environment parameter labels and damage detection result labels respectively corresponding to the plurality of test pavement images.
In some embodiments, determining the preset association relationship of the model parameters according to the accuracy of the first detection model, the accuracy of the second detection model and the environmental parameter labels and the damage detection result labels respectively corresponding to the plurality of test pavement images includes: if the precision of the first detection model is greater than or equal to that of the second detection model, determining a first influence value of the environmental parameter labels respectively corresponding to the plurality of test pavement images on the damage detection result labels respectively corresponding to the plurality of test pavement images; determining a first impact weight of the environmental parameter on the model parameter of the second detection model based on the first impact value; determining a second impact weight based on a precision difference between the precision of the first detection model and the precision of the second detection model; determining influence weight values of environment parameter labels corresponding to the plurality of test pavement images respectively based on the first influence weight and the second influence weight; and determining a preset model parameter association relation according to the influence weight value and environment parameter labels respectively corresponding to the plurality of test pavement images.
In some embodiments, if the environmental parameter corresponding to the road surface image has a higher influence value on the damage detection result, the first influence value is larger; otherwise, the smaller the first impact value. Further, the first influence value and the first influence weight are in a proportional relation, and the first influence weight can be correspondingly determined.
In some embodiments, the greater the accuracy difference, the less the second impact weight; otherwise, the greater the second impact weight.
In some embodiments, the weight impact value is an average of the first impact weight and the second impact weight; or other embodiments.
In some embodiments, after the determination of the impact weight value, the preset association relationship of the model parameters includes: a plurality of environmental parameters, an association relationship between the plurality of environmental parameters; and an association relationship between the plurality of model parameters. The association relation among the plurality of model parameters is determined according to the actual conditions of the first detection model and the second detection model.
That is, the above-described embodiment describes a procedure of determining an environmental parameter and an influence weight value corresponding to the environmental parameter.
As an optional implementation manner, determining a preset model parameter association relationship according to the accuracy of the first detection model, the accuracy of the second detection model and the environmental parameter labels and the damage detection result labels respectively corresponding to the plurality of test pavement images includes: if the accuracy of the first detection model is smaller than that of the second detection model, determining a target environment parameter label from environment parameter labels respectively corresponding to a plurality of test pavement images; the image quality of the test pavement image corresponding to the target environment parameter label is higher than the preset image quality; determining second influence values of the target environment parameter labels on damage detection result labels corresponding to the plurality of test pavement images respectively; determining a third influence weight of the target environment parameter label on the model parameters of the second detection model based on the second influence value; determining a fourth impact weight based on a precision difference between the precision of the first detection model and the precision of the second detection model; determining an influence weight value of the target environment parameter label based on the third influence weight and the fourth influence weight; and determining a preset model parameter association relation according to the influence weight value and the target environment parameter label.
In some embodiments, the image quality of a plurality of test pavement images is determined, and then the target environmental parameter label is screened out according to the image quality. The preset image quality may be configured according to different application scenarios, which is not limited herein.
In some embodiments, the determining implementation of the second influence value, the third influence weight and the fourth influence weight may refer to the determining implementation of the first influence value, the first influence weight and the second influence weight, which are described above, but the foregoing description is for all environment parameter tags, and only the target environment parameter tag is needed here.
Further, based on the third impact weight and the fourth impact weight, an average or other arithmetic integration method may be performed to determine a final impact weight value.
Furthermore, the preset association relationship of the model parameters includes a plurality of environment parameters, and the plurality of environment parameters respectively correspond to the influence weight values and include the association relationship among the plurality of model parameters.
Then, as an alternative embodiment, step 203 includes: determining target environmental parameters in the environmental parameters corresponding to the pavement images respectively; in a preset model association relationship, the matched environment parameters exist in the target environment parameters; updating corresponding model parameters in the second detection model based on the influence weight value corresponding to the target environment parameter aiming at the same model parameters in the first detection model and the second detection model; and updating corresponding model parameters of the second detection model based on the association relation between different model parameters according to the influence weight value corresponding to the target environment parameter aiming at the different model parameters in the first detection model and the second detection model.
In some embodiments, if the impact weight value corresponding to the target environmental parameter is greater than the preset weight value, the accuracy of the corresponding model parameter in the second detection model is improved, that is, the model parameter with higher accuracy is adopted. Otherwise, the accuracy of the corresponding model parameters in the second detection model can be reserved.
In some embodiments, if the impact weight value corresponding to the target environmental parameter is greater than the preset weight value, and a correlation exists between the different model parameters; or the influence weight value corresponding to the target environment parameter is smaller than or equal to the preset weight value, and the different model parameters have a mutual influence relationship; or the influence weight value corresponding to the target environment parameter is smaller than or equal to the preset weight value, and the different model parameters have a mutual influence relationship; and then, improving the precision of the corresponding model parameters in the second detection model, namely adopting the model parameters with higher precision.
If the influence weight value corresponding to the target environment parameter is larger than the preset weight value, and no mutual influence relation exists between the different model parameters; then the accuracy of the corresponding model parameters in the second detection model may be preserved.
Thus, by step 203, the model accuracy of the second detection model is correspondingly changed along with the detection result of the first detection model, so that the second detection model is more suitable for detecting multiple pavement images.
Further, in step 204, a breakage detection result corresponding to the target road surface is determined based on the updated second detection model and the plurality of road surface images.
In some embodiments, the plurality of road surface images are input into an updated second detection model, which then outputs a corresponding breakage detection result.
In some embodiments, the damage detection result includes damage information, where the damage information includes at least one of damage area, damage degree, and damage repair rate, and the road surface detection method based on the neural vision network further includes: generating a second training data set based on the plurality of road surface images and the environment parameters and the damage information respectively corresponding to the plurality of road surface images; training the initial third detection model based on the second training data set to obtain a trained third detection model; model parameters of the initial third detection model are determined based on model parameters of the first detection model and model parameters of the updated second detection model.
In some embodiments, the accuracy of the model parameters of the initial third detection model may lie within an accuracy range defined by the accuracy of the model parameters of the first detection model and the accuracy of the model parameters of the second detection model.
In some embodiments, the second training data set is used to train the initial third detection model, the trained third detection model can directly determine corresponding environmental parameters and damage information based on the plurality of road surface images, and the accuracy of the output recognition result is higher than that of the detection results obtained by the first detection model and the second detection model respectively.
It can be seen from the description of the embodiment of the present application that after the road surface images of the road surface to be detected are obtained, the environment parameters corresponding to the road surface images are determined first, and the parameters of the neural vision network for detecting the road surface breakage are optimized based on the environment parameters. The road surface detection scheme not only utilizes the multi-layer network to realize road surface detection, but also correlates model parameters of the multi-layer network with each other and environmental parameters; therefore, the road surface detection scheme not only can realize effective detection of the road surface based on the neural vision network; the model parameters of the neural vision network can be optimized based on the relevance among the neural vision networks, so that more accurate pavement detection is realized. Thus, the pavement detection scheme can realize accurate and effective pavement detection.
Referring next to fig. 3, a road surface detection device based on a neural vision network according to an embodiment of the present application includes:
an acquiring unit 301, configured to acquire a plurality of road surface images corresponding to a target road surface; a processing unit 302, configured to: determining environmental parameters corresponding to the plurality of road surface images respectively based on a first detection model; the first detection model is a model based on a neural vision network; updating model parameters of a second detection model based on corresponding environment parameters of the plurality of pavement images, model parameters of the first detection model and a preset model parameter association relation so as to determine an updated second detection model; the second detection model corresponds to initial model parameters, and the preset model parameter association relationship is used for representing association relationships among the model parameters of the first detection model, the model parameters of the second detection model and environment parameters; and determining a breakage detection result corresponding to the target pavement based on the updated second detection model and the plurality of pavement images.
In some embodiments, the acquisition unit 301 is further configured to: determining road surface information of the target road surface; the road surface information includes: at least one of road surface material, road surface use frequency, and road surface maintenance frequency; determining an acquisition strategy of the plurality of road surface images based on the road surface information of the target road surface; the acquisition strategy comprises the following steps: at least one item of information of image acquisition period, image acquisition quantity and image quality; and acquiring a plurality of pavement images corresponding to the target pavement based on the acquisition strategy.
In some embodiments, the processing unit 302 is further to: determining the image acquisition period based on the road surface use frequency and a preset first association relation; the first association relationship is used for representing the association relationship between the road surface use frequency and the image acquisition period; determining the image acquisition quantity based on the road surface maintenance frequency, the road surface use frequency and a preset second association relation; the second association relationship is used for representing association relationships among the road maintenance frequency, the road use frequency and the image acquisition quantity; determining the image quality based on the pavement material and a preset third association relation; the third association relationship is used for representing the association relationship between the pavement material and the image quality.
In some embodiments, the obtaining unit 301 is further configured to obtain a first training data set; the first training data set comprises a plurality of sample pavement images, and the plurality of sample pavement images respectively correspond to environmental parameter labels; the processing unit 302 is further configured to train an initial first detection model according to the first training data set, so as to obtain the first detection model; wherein initial model parameters of the initial first detection model are determined based on initial model parameters of the second detection model.
In some embodiments, the processing unit 302 is further to: determining the model precision of the first detection model according to a first test data set; the first test data set comprises a plurality of test pavement images, and the plurality of test pavement images respectively correspond to environmental parameter labels; determining the model precision of the second detection model according to the second test data set; the second test data set comprises a plurality of test pavement images, and the plurality of test pavement images respectively correspond to breakage detection result labels; and determining the preset model parameter association relation according to the precision of the first detection model, the precision of the second detection model and the environment parameter labels and damage detection result labels respectively corresponding to the plurality of test pavement images.
In some embodiments, the processing unit 302 is further to: if the precision of the first detection model is greater than or equal to that of the second detection model, determining a first influence value of the environmental parameter labels respectively corresponding to the plurality of test pavement images on the damage detection result labels respectively corresponding to the plurality of test pavement images; determining a first impact weight of an environmental parameter on a model parameter of the second detection model based on the first impact value; determining a second impact weight based on a precision difference between the precision of the first detection model and the precision of the second detection model; determining influence weight values of environment parameter labels corresponding to the plurality of test pavement images respectively based on the first influence weight and the second influence weight; and determining the preset model parameter association relation according to the influence weight value and the environment parameter labels respectively corresponding to the plurality of test pavement images.
In some embodiments, the processing unit 302 is further to: if the precision of the first detection model is smaller than that of the second detection model, determining a target environment parameter label from environment parameter labels respectively corresponding to the plurality of test pavement images; the image quality of the test pavement image corresponding to the target environment parameter label is higher than the preset image quality; determining second influence values of the target environment parameter labels on damage detection result labels corresponding to the plurality of test pavement images respectively; determining a third influence weight of the target environmental parameter label on the model parameters of the second detection model based on the second influence value; determining a fourth impact weight based on a precision difference between the precision of the first detection model and the precision of the second detection model; determining an impact weight value of the target environmental parameter tag based on the third impact weight and the fourth impact weight; and determining the preset model parameter association relation according to the influence weight value and the target environment parameter label.
In some embodiments, the processing unit 302 is further to: determining target environmental parameters in the environmental parameters corresponding to the plurality of pavement images respectively; the target environment parameters have matched environment parameters in the preset model association relation; updating corresponding model parameters in the second detection model based on the influence weight value corresponding to the target environment parameter aiming at the same model parameters in the first detection model and the second detection model; and updating corresponding model parameters of the second detection model based on the association relation between the different model parameters based on the influence weight value corresponding to the target environment parameter aiming at the different model parameters in the first detection model and the second detection model.
In some embodiments, the processing unit 302 is further configured to: generating a second training data set based on the plurality of road surface images, the environmental parameters respectively corresponding to the plurality of road surface images and the damage information; training the initial third detection model based on the second training data set to obtain a trained third detection model; model parameters of the initial third detection model are determined based on model parameters of the first detection model and model parameters of the updated second detection model.
As shown in fig. 4, the embodiment of the present application further provides a terminal device, which includes a processor 401 and a memory 402, where the processor 401 and the memory 402 are communicatively connected, and the terminal device may be used as an execution body of the aforementioned road surface detection method.
The processor 401 and the memory 402 are directly or indirectly electrically connected to each other to realize data transmission or interaction. For example, electrical connections may be made between these elements through one or more communication buses or signal buses. The aforementioned road surface detection methods each include at least one software functional module that may be stored in the memory 402 in the form of software or firmware (firmware).
The processor 401 may be an integrated circuit chip having signal processing capabilities. The processor 401 may be a general-purpose processor including a CPU (Central Processing Unit ), NP (Network Processor, network processor), etc.; but may be a digital signal processor, an application specific integrated circuit, an off-the-shelf programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component. Which may implement or perform the disclosed methods, steps, and logic blocks in embodiments of the application. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory 402 may store various software programs and modules, such as program instructions/modules corresponding to the image processing methods and apparatuses provided in the embodiments of the present application. The processor 401 executes various functional applications and data processing, i.e., implements the methods of embodiments of the present application, by running software programs and modules stored in the memory 402.
Memory 402 may include, but is not limited to, RAM (Random Access Memory ), ROM (Read Only Memory), PROM (Programmable Read-Only Memory, programmable Read Only Memory), EPROM (Erasable Programmable Read-Only Memory, erasable Read Only Memory), EEPROM (Electric Erasable Programmable Read-Only Memory, electrically erasable Read Only Memory), and the like.
It will be appreciated that the configuration shown in fig. 4 is merely illustrative, and that the terminal device may also include more or fewer components than shown in fig. 4, or have a different configuration than shown in fig. 4.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing descriptions of specific exemplary embodiments of the present application are presented for purposes of illustration and description. It is not intended to limit the application to the precise form disclosed, and obviously many modifications and variations are possible in light of the above teaching. The exemplary embodiments were chosen and described in order to explain the specific principles of the application and its practical application to thereby enable one skilled in the art to make and utilize the application in various exemplary embodiments and with various modifications as are suited to the particular use contemplated. It is intended that the scope of the application be defined by the claims and their equivalents.
Claims (10)
1. The road surface detection method based on the neural vision network is characterized by comprising the following steps of:
Acquiring a plurality of pavement images corresponding to a target pavement;
determining environmental parameters corresponding to the plurality of road surface images respectively based on a first detection model; the first detection model is a model based on a neural vision network, the environmental parameter is a parameter corresponding to the environment where the target pavement is located, and the environmental parameter has an influence on a pavement detection result;
updating model parameters of a second detection model based on corresponding environment parameters of the plurality of pavement images, model parameters of the first detection model and a preset model parameter association relation so as to determine an updated second detection model; the second detection model corresponds to initial model parameters, and the preset model parameter association relationship is used for representing association relationships among the model parameters of the first detection model, the model parameters of the second detection model and environment parameters; the preset model parameter association relationship comprises a plurality of environment parameters, wherein the environment parameters respectively correspond to the influence weight values and comprise association relationships among the model parameters;
and determining a breakage detection result corresponding to the target pavement based on the updated second detection model and the plurality of pavement images.
2. The road surface detection method based on the neural vision network according to claim 1, wherein the obtaining a plurality of road surface images corresponding to the target road surface includes:
determining road surface information of the target road surface; the road surface information includes: at least one of road surface material, road surface use frequency, and road surface maintenance frequency;
determining an acquisition strategy of the plurality of road surface images based on the road surface information of the target road surface; the acquisition strategy comprises the following steps: at least one item of information of image acquisition period, image acquisition quantity and image quality;
and acquiring a plurality of pavement images corresponding to the target pavement based on the acquisition strategy.
3. The road surface detection method based on the neural vision network according to claim 2, wherein the determining the acquisition strategy of the plurality of road surface images based on the road surface information of the target road surface includes:
determining the image acquisition period based on the road surface use frequency and a preset first association relation; the first association relationship is used for representing the association relationship between the road surface use frequency and the image acquisition period;
determining the image acquisition quantity based on the road surface maintenance frequency, the road surface use frequency and a preset second association relation; the second association relationship is used for representing association relationships among the road maintenance frequency, the road use frequency and the image acquisition quantity;
Determining the image quality based on the pavement material and a preset third association relation; the third association relationship is used for representing the association relationship between the pavement material and the image quality.
4. The neural vision network-based pavement detection method of claim 1, further comprising:
acquiring a first training data set; the first training data set comprises a plurality of sample pavement images, and the plurality of sample pavement images respectively correspond to environmental parameter labels;
training an initial first detection model according to the first training data set to obtain the first detection model; wherein the initial model parameter accuracy of the initial first detection model is determined based on the initial model parameter accuracy of the second detection model.
5. The neural vision network-based pavement detection method of claim 1, further comprising:
determining the model precision of the first detection model according to a first test data set; the first test data set comprises a plurality of test pavement images, the plurality of test pavement images respectively correspond to environment parameter labels, and the model precision of the first detection model is used for representing the precision of model parameters of the first detection model;
Determining the model precision of the second detection model according to the second test data set; the second test data set comprises a plurality of test pavement images, the plurality of test pavement images respectively correspond to damage detection result labels, and the model precision of the second detection model is used for representing the precision of model parameters of the second detection model;
and determining the environmental parameters and the impact weight values corresponding to the environmental parameters according to the precision of the first detection model, the precision of the second detection model, the environmental parameter labels corresponding to the plurality of test pavement images and the damage detection result labels.
6. The neural vision network-based pavement detection method of claim 5, wherein determining the plurality of environmental parameters and the impact weight values corresponding to the plurality of environmental parameters according to the accuracy of the first detection model, the accuracy of the second detection model, and the environmental parameter labels and the breakage detection result labels corresponding to the plurality of test pavement images, respectively, comprises:
if the precision of the first detection model is greater than or equal to that of the second detection model, determining a first influence value of the environmental parameter labels respectively corresponding to the plurality of test pavement images on the damage detection result labels respectively corresponding to the plurality of test pavement images;
Determining a first impact weight of an environmental parameter on a model parameter of the second detection model based on the first impact value;
determining a second impact weight based on a precision difference between the precision of the first detection model and the precision of the second detection model;
determining influence weight values of environment parameter labels corresponding to the plurality of test pavement images respectively based on the first influence weight and the second influence weight;
and determining the plurality of environment parameters and the influence weight values corresponding to the plurality of environment parameters according to the influence weight values and the environment parameter labels corresponding to the plurality of test pavement images respectively.
7. The neural vision network-based pavement detection method of claim 5, wherein determining the plurality of environmental parameters and the impact weight values corresponding to the plurality of environmental parameters according to the accuracy of the first detection model, the accuracy of the second detection model, and the environmental parameter labels and the breakage detection result labels corresponding to the plurality of test pavement images, respectively, comprises:
if the precision of the first detection model is smaller than that of the second detection model, determining a target environment parameter label from environment parameter labels respectively corresponding to the plurality of test pavement images; the image quality of the test pavement image corresponding to the target environment parameter label is higher than the preset image quality;
Determining second influence values of the target environment parameter labels on damage detection result labels corresponding to the plurality of test pavement images respectively;
determining a third influence weight of the target environmental parameter label on the model parameters of the second detection model based on the second influence value; wherein the second impact value is proportional to the third impact weight;
determining a fourth impact weight based on a precision difference between the precision of the first detection model and the precision of the second detection model; wherein the precision difference is inversely proportional to the fourth impact weight;
determining an impact weight value of the target environmental parameter based on the third impact weight and the fourth impact weight;
and determining the influence weight values respectively corresponding to the environment parameters and the environment parameters according to the influence weight values and the target environment parameter labels.
8. The road surface detection method based on the neural vision network according to claim 1, wherein updating the model parameters of the second detection model based on the environmental parameters, the model parameters of the first detection model, and the preset model parameter association relation respectively corresponding to the plurality of road surface images comprises:
Determining target environmental parameters in the environmental parameters corresponding to the plurality of pavement images respectively; the target environment parameters have matched environment parameters in the preset model association relation;
updating corresponding model parameters in the second detection model based on the influence weight value corresponding to the target environment parameter aiming at the same model parameters in the first detection model and the second detection model;
and updating corresponding model parameters of the second detection model based on the association relation between the different model parameters based on the influence weight value corresponding to the target environment parameter aiming at the different model parameters in the first detection model and the second detection model.
9. The neural vision network-based pavement detection method of claim 1, wherein the breakage detection result includes breakage information including at least one of a breakage area, a breakage degree, and a breakage repair rate, and further comprising:
generating a second training data set based on the plurality of road surface images, the environmental parameters respectively corresponding to the plurality of road surface images and the damage information;
Training the initial third detection model based on the second training data set to obtain a trained third detection model; model parameter accuracy of the initial third detection model is determined based on model parameters of the first detection model and model parameter accuracy of the updated second detection model.
10. Road surface detection device based on neural vision network, characterized by comprising:
the acquisition unit is used for acquiring a plurality of pavement images corresponding to the target pavement;
a processing unit for:
determining environmental parameters corresponding to the plurality of road surface images respectively based on a first detection model; the first detection model is a model based on a neural vision network, the environmental parameter is a parameter corresponding to the environment where the target pavement is located, and the environmental parameter has an influence on a pavement detection result;
updating model parameters of a second detection model based on corresponding environment parameters of the plurality of pavement images, model parameters of the first detection model and a preset model parameter association relation so as to determine an updated second detection model; the second detection model corresponds to initial model parameters, and the preset model parameter association relationship is used for representing association relationships among the model parameters of the first detection model, the model parameters of the second detection model and environment parameters; the preset model parameter association relationship comprises a plurality of environment parameters, wherein the environment parameters respectively correspond to the influence weight values and comprise association relationships among the model parameters;
And determining a breakage detection result corresponding to the target pavement based on the updated second detection model and the plurality of pavement images.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310960793.0A CN116664582B (en) | 2023-08-02 | 2023-08-02 | Road surface detection method and device based on neural vision network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310960793.0A CN116664582B (en) | 2023-08-02 | 2023-08-02 | Road surface detection method and device based on neural vision network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116664582A CN116664582A (en) | 2023-08-29 |
| CN116664582B true CN116664582B (en) | 2023-10-27 |
Family
ID=87724661
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310960793.0A Active CN116664582B (en) | 2023-08-02 | 2023-08-02 | Road surface detection method and device based on neural vision network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116664582B (en) |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009025198A (en) * | 2007-07-20 | 2009-02-05 | Denso It Laboratory Inc | Apparatus and method for detecting road surface condition |
| CN105320942A (en) * | 2015-10-21 | 2016-02-10 | 东南大学 | Method for detecting pavement damage based on joint detector |
| CN106368732A (en) * | 2016-12-01 | 2017-02-01 | 中铁十二局集团有限公司 | Pressure releasing and drainage system for double-track tunnel base of high speed railway |
| CN109146849A (en) * | 2018-07-26 | 2019-01-04 | 昆明理工大学 | A kind of road surface crack detection method based on convolutional neural networks and image recognition |
| CN109685246A (en) * | 2018-11-13 | 2019-04-26 | 平安科技（深圳）有限公司 | Environmental data predictor method, device and storage medium, server |
| DE102018008788A1 (en) * | 2018-11-08 | 2019-05-02 | Daimler Ag | Method for determining a road surface roughness |
| WO2019085793A1 (en) * | 2017-11-01 | 2019-05-09 | 腾讯科技（深圳）有限公司 | Image classification method, computer device and computer readable storage medium |
| CN110084844A (en) * | 2019-04-25 | 2019-08-02 | 中国民航大学 | A kind of airfield pavement crack detection method based on depth camera |
| CN111242017A (en) * | 2020-01-10 | 2020-06-05 | 北京邮电大学 | Multi-marking-line pavement crack identification method, device, equipment and storage medium |
| CN112308912A (en) * | 2020-11-03 | 2021-02-02 | 长安大学 | System, device and method for acquiring road surface disease homologous multi-characteristic image |
| CN112581441A (en) * | 2020-12-10 | 2021-03-30 | 北京工业大学 | Pavement crack detection method and device, electronic equipment and storage medium |
| CN114140467A (en) * | 2022-02-08 | 2022-03-04 | 中大检测(湖南)股份有限公司 | Pavement crack monitoring method and system based on image recognition and storage medium |
| CN115470566A (en) * | 2022-11-07 | 2022-12-13 | 南京惠派智慧后勤服务有限公司 | Intelligent building energy consumption control method and system based on BIM |
| WO2023109221A1 (en) * | 2021-12-14 | 2023-06-22 | 北京地平线信息技术有限公司 | Method and apparatus for determining homography matrix, medium, device, and program product |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015020687A1 (en) * | 2013-08-08 | 2015-02-12 | Iteris, Inc. | Pavement condition analysis from modeling impact of traffic characteristics, weather data and road conditions |
| CN110945449B (en) * | 2018-11-15 | 2023-10-17 | 灵动科技（北京）有限公司 | Real-time supervision type machine learning system and method for field environment |
-
2023
- 2023-08-02 CN CN202310960793.0A patent/CN116664582B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009025198A (en) * | 2007-07-20 | 2009-02-05 | Denso It Laboratory Inc | Apparatus and method for detecting road surface condition |
| CN105320942A (en) * | 2015-10-21 | 2016-02-10 | 东南大学 | Method for detecting pavement damage based on joint detector |
| CN106368732A (en) * | 2016-12-01 | 2017-02-01 | 中铁十二局集团有限公司 | Pressure releasing and drainage system for double-track tunnel base of high speed railway |
| WO2019085793A1 (en) * | 2017-11-01 | 2019-05-09 | 腾讯科技（深圳）有限公司 | Image classification method, computer device and computer readable storage medium |
| CN109146849A (en) * | 2018-07-26 | 2019-01-04 | 昆明理工大学 | A kind of road surface crack detection method based on convolutional neural networks and image recognition |
| DE102018008788A1 (en) * | 2018-11-08 | 2019-05-02 | Daimler Ag | Method for determining a road surface roughness |
| CN109685246A (en) * | 2018-11-13 | 2019-04-26 | 平安科技（深圳）有限公司 | Environmental data predictor method, device and storage medium, server |
| CN110084844A (en) * | 2019-04-25 | 2019-08-02 | 中国民航大学 | A kind of airfield pavement crack detection method based on depth camera |
| CN111242017A (en) * | 2020-01-10 | 2020-06-05 | 北京邮电大学 | Multi-marking-line pavement crack identification method, device, equipment and storage medium |
| CN112308912A (en) * | 2020-11-03 | 2021-02-02 | 长安大学 | System, device and method for acquiring road surface disease homologous multi-characteristic image |
| CN112581441A (en) * | 2020-12-10 | 2021-03-30 | 北京工业大学 | Pavement crack detection method and device, electronic equipment and storage medium |
| WO2023109221A1 (en) * | 2021-12-14 | 2023-06-22 | 北京地平线信息技术有限公司 | Method and apparatus for determining homography matrix, medium, device, and program product |
| CN114140467A (en) * | 2022-02-08 | 2022-03-04 | 中大检测(湖南)股份有限公司 | Pavement crack monitoring method and system based on image recognition and storage medium |
| CN115470566A (en) * | 2022-11-07 | 2022-12-13 | 南京惠派智慧后勤服务有限公司 | Intelligent building energy consumption control method and system based on BIM |
Non-Patent Citations (4)
| Title |
|---|
| Analysis of Thermal Environment and Its Influencing Factors in Deep Stope of Metal Mine;Shuang Y 等;Geofluids;1-3 * |
| Comparative analysis of the influence of various materials on the state of the roadside environment during the road repair;Andrii B 等;Environmental science and pollution research international;第30卷(第6期);15523-15530 * |
| 农村路面多类型病害检测方法研究;朱洪波 等;测绘科学;第47卷(第09期);170-180 * |
| 基于典型裂纹的沥青路面损坏状况视觉评估研究;边琛;中国优秀硕士学位论文全文数据库 (工程科技Ⅱ辑)(第（2022）05期);C034-92 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116664582A (en) | 2023-08-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9365217B2 (en) | Mobile pothole detection system and method | |
| Akagic et al. | Pothole detection: An efficient vision based method using rgb color space image segmentation | |
| CN110533950A (en) | Detection method, device, electronic equipment and the storage medium of parking stall behaviour in service | |
| CN112929613B (en) | Inspection method and system for equipment operation and maintenance based on image recognition | |
| Duthon et al. | Methodology used to evaluate computer vision algorithms in adverse weather conditions | |
| CN111967368A (en) | Traffic light identification method and device | |
| CN108573244B (en) | Vehicle detection method, device and system | |
| CN111582255A (en) | Vehicle overrun detection method and device, computer equipment and storage medium | |
| US11120308B2 (en) | Vehicle damage detection method based on image analysis, electronic device and storage medium | |
| CN112487884A (en) | Traffic violation behavior detection method and device and computer readable storage medium | |
| CN113505638B (en) | Method and device for monitoring traffic flow and computer readable storage medium | |
| CN103913150A (en) | Consistency detection method for electron components of intelligent ammeter | |
| Deepa et al. | Pothole Detection using Roboflow Convolutional Neural Networks | |
| CN116664582B (en) | Road surface detection method and device based on neural vision network | |
| CN116563841B (en) | Detection method and detection device for power distribution network equipment identification plate and electronic equipment | |
| CN113313110A (en) | License plate type recognition model construction and license plate type recognition method | |
| CN117409381A (en) | Expressway toll station congestion detection model and method based on scene image segmentation | |
| CN112183382A (en) | Unmanned traffic light detection and classification method and device | |
| Khan | Vehicle and pedestrian detection using YOLOv3 and YOLOv4 for self-driving cars | |
| KR20200123324A (en) | A method for pig segmentation using connected component analysis and yolo algorithm | |
| CN112729366B (en) | Test evaluation method and device for weather simulation equipment for automatic driving field test | |
| Song et al. | Real-time visibility distance evaluation based on monocular and dark channel prior | |
| CN113989774A (en) | Traffic light detection method and device, vehicle and readable storage medium | |
| CN113085861A (en) | Control method and device for automatic driving vehicle and automatic driving vehicle | |
| CN112861701A (en) | Illegal parking identification method and device, electronic equipment and computer readable medium |
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 |