CN114943693B - Jetson Nano bridge crack detection method and system - Google Patents
Jetson Nano bridge crack detection method and system Download PDFInfo
- Publication number
- CN114943693B CN114943693B CN202210497871.3A CN202210497871A CN114943693B CN 114943693 B CN114943693 B CN 114943693B CN 202210497871 A CN202210497871 A CN 202210497871A CN 114943693 B CN114943693 B CN 114943693B
- Authority
- CN
- China
- Prior art keywords
- bridge
- picture
- detected
- shooting
- data
- 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
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
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Signal Processing (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a jetson nano bridge crack detection method and system, wherein the method comprises the following steps: configuring a neural network model for the JetsonNano arranged at the bridge detection workstation; receiving analysis data generated based on a neural network model and a surface picture of a bridge to be detected of Jetson Nano; and determining a bridge crack detection result based on the analysis data. The Jetson nano bridge crack detection method based on the invention realizes intelligent monitoring of bridge cracks and improves reliability and certainty of acquired data.
Description
Technical Field
The invention relates to the technical field of bridge detection, in particular to a Jetson Nano bridge crack detection method and system.
Background
The bridge plays a role in national economy development. After the bridge is built, the surface of the bridge is influenced by environmental factors, natural conditions, loading actions and other factors for a long time, cracks can appear on the surface of the bridge, the cracks not only lead to the protection failure of the concrete layer to the internal reinforcing steel bars, but also can lead to the falling of the concrete, and serious cracks are more precursors of the collapse of the bridge, so that the cracks are one of main evaluation indexes of the health condition of the bridge.
The existing detection method still adopts a manual detection mode, and has a plurality of defects:
(1) The cost is high: the manual detection mode requires a great deal of manpower and equipment such as a bridge inspection vehicle, and is long in time consumption and high in cost;
(2) The real-time performance is poor: the manual detection mode is carried out regularly, so that problems cannot be found in time;
(3) The informatization degree is low: the bridge crack file cannot be established, so that the bridge is inconvenient to manage and maintain, and decision support information cannot be provided for management departments.
Disclosure of Invention
The invention aims to provide a Jetson Nano bridge crack detection method, which is used for realizing intelligent monitoring of bridge cracks and improving reliability and certainty of acquired data.
The jetson nano bridge crack detection method provided by the embodiment of the invention comprises the following steps:
configuring a neural network model for the JetsonNano arranged at the bridge detection workstation;
receiving analysis data generated based on a neural network model and a surface picture of a bridge to be detected of Jetson Nano;
and determining a bridge crack detection result based on the analysis data.
Preferably, the jetson nano analysis data generated based on the neural network model and the surface picture of the bridge to be tested performs the following operations:
acquiring a surface picture of at least one bridge to be detected and a position of the surface picture corresponding to the bridge to be detected;
numbering the surface pictures based on the positions;
inputting each surface picture into a neural network model, and obtaining output data corresponding to each surface picture;
and correlating the numbers of the surface pictures with the output data to form analysis data.
Preferably, determining the bridge crack detection result based on the analysis data includes:
analyzing the analysis data to construct an analysis feature set;
acquiring a preset detection library;
matching the analysis feature set with each standard feature set in the detection library one by one;
when the analysis feature set is matched with the standard feature set, obtaining a bridge crack detection result correspondingly associated with the standard feature set in the detection library;
analyzing the analysis data to construct an analysis feature set, including:
sampling analysis data according to a preset analysis data sampling rule, and obtaining N groups of sampling data altogether; n is greater than or equal to 1;
the output data in the sampling data of each group are arranged in sequence according to the serial numbers to form one row of data;
arranging each row of data from top to bottom according to the time sequence of each group of sampling data to form an analysis feature set; wherein the sample data closest to the current time is located in the first row.
Preferably, obtaining a surface picture of a bridge to be detected includes:
shooting first pictures of the bridge to be detected through a plurality of cameras respectively shooting a plurality of preset first positions of the bridge to be detected;
shooting a second picture of the bridge deck of the bridge to be detected, which is used for the vehicle to run, by a bridge detection vehicle;
shooting a third picture of the outer side of the bridge to be detected by unmanned aerial vehicle shooting equipment;
obtaining the position of the surface picture corresponding to the bridge to be detected comprises the following steps:
acquiring a preset three-dimensional space containing a bridge to be detected;
acquiring setting parameters and first shooting parameters of a camera;
mapping the camera to a three-dimensional space based on the setting parameters;
determining the position of a bridge to be detected corresponding to a first picture corresponding to a camera based on the first shooting parameter;
acquiring first positioning information and second shooting parameters of the bridge detection vehicle in shooting a second picture;
mapping the bridge inspection vehicle to a three-dimensional space based on the first positioning information;
determining the position of the bridge to be detected corresponding to the second picture based on the second shooting parameters;
acquiring second positioning information of the unmanned aerial vehicle shooting equipment when the unmanned aerial vehicle shooting equipment shoots a third picture;
mapping the unmanned aerial vehicle shooting equipment to a three-dimensional space based on the second positioning information;
and determining the position of the bridge to be detected corresponding to the third picture based on the third shooting parameter.
Preferably, obtaining at least one surface image of the bridge to be detected and a position of the surface image corresponding to the bridge to be detected includes:
carrying out region extraction on a preset region of the surface picture;
acquiring a first region picture;
acquiring a preset position determination library;
matching the first region picture with each standard picture in the position determining library, and obtaining position information corresponding to the standard picture matched with the first region picture;
and analyzing the position information and determining the position of the bridge to be detected corresponding to the surface picture.
Preferably, numbering the surface pictures based on position includes:
acquiring a preset three-dimensional space containing a bridge to be detected;
determining coordinate values in a three-dimensional space corresponding to the surface picture based on the position of the bridge to be detected corresponding to the surface picture;
and arranging the coordinate values in sequence to form the number of the surface picture.
The invention provides a crack detection system based on a Jetson Nano bridge, which comprises the following components:
the configuration module is used for configuring a neural network model for the Jetson Nano arranged at the bridge detection workstation;
the analysis data receiving module is used for receiving analysis data generated based on a neural network model of the Jetson Nano and a surface picture of the bridge to be detected;
and the detection module is used for determining a bridge crack detection result based on the analysis data.
Preferably, the jetson nano analysis data generated based on the neural network model and the surface picture of the bridge to be tested performs the following operations:
acquiring a surface picture of at least one bridge to be detected and a position of the surface picture corresponding to the bridge to be detected;
numbering the surface pictures based on the positions;
inputting each surface picture into a neural network model, and obtaining output data corresponding to each surface picture;
and correlating the numbers of the surface pictures with the output data to form analysis data.
Preferably, the detection module determines a bridge crack detection result based on the analysis data, and performs the following operations:
analyzing the analysis data to construct an analysis feature set;
acquiring a preset detection library;
matching the analysis feature set with each standard feature set in the detection library one by one;
when the analysis feature set is matched with the standard feature set, obtaining a bridge crack detection result correspondingly associated with the standard feature set in the detection library;
analyzing the analysis data to construct an analysis feature set, including:
sampling analysis data according to a preset analysis data sampling rule, and obtaining N groups of sampling data altogether; n is greater than or equal to 1;
the output data in the sampling data of each group are arranged in sequence according to the serial numbers to form one row of data;
arranging each row of data from top to bottom according to the time sequence of each group of sampling data to form an analysis feature set; wherein the sample data closest to the current time is located in the first row.
Preferably, jetson nano obtains a surface picture of a bridge to be detected, including:
shooting first pictures of the bridge to be detected through a plurality of cameras respectively shooting a plurality of preset first positions of the bridge to be detected;
shooting a second picture of the bridge deck of the bridge to be detected, which is used for the vehicle to run, by a bridge detection vehicle;
a third picture of the outer side of the bridge to be tested is shot through the unmanned aerial vehicle or the handheld camera;
jetson Nano obtains the position of the surface picture corresponding to the bridge to be tested, including:
acquiring a preset three-dimensional space containing a bridge to be detected;
acquiring setting parameters and first shooting parameters of a camera;
mapping the camera to a three-dimensional space based on the setting parameters;
determining the position of a bridge to be detected corresponding to a first picture corresponding to a camera based on the first shooting parameter;
acquiring first positioning information and second shooting parameters of the bridge detection vehicle in shooting a second picture;
mapping the bridge inspection vehicle to a three-dimensional space based on the first positioning information;
determining the position of the bridge to be detected corresponding to the second picture based on the second shooting parameters;
acquiring second positioning information of the unmanned aerial vehicle shooting equipment when the unmanned aerial vehicle shooting equipment shoots a third picture;
mapping the unmanned aerial vehicle shooting equipment to a three-dimensional space based on the second positioning information;
and determining the position of the bridge to be detected corresponding to the third picture based on the third shooting parameter.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:
FIG. 1 is a schematic diagram of a crack detection method based on Jetson Nano bridge in an embodiment of the invention;
FIG. 2 is a schematic diagram of a model construction in an embodiment of the invention;
FIG. 3 is a schematic diagram of a model configuration according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a jetson nano-based bridge crack detection method in an embodiment of the invention.
Detailed Description
The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.
The embodiment of the invention provides a Jetson Nano bridge crack detection method, which is shown in figure 1 and comprises the following steps:
step S1: configuring a neural network model for the JetsonNano arranged at the bridge detection workstation;
step S2: receiving analysis data generated based on a neural network model and a surface picture of a bridge to be detected of Jetson Nano;
step S3: and determining a bridge crack detection result based on the analysis data.
The working principle and the beneficial effects of the technical scheme are as follows:
jetson Nano is an edge computing terminal, and is arranged near a bridge to be detected to realize on-site data analysis, and a platform side can acquire a final detection result only by analyzing analysis data; the yolov5 target detection algorithm was carried using jetson nano artificial intelligence edge computing device of inflight. And (3) performing visual measurement and quantitative evaluation of the damage of the bridge surface according to the detection result after the bridge crack is detected by using a yolov5 algorithm. As in fig. 2, tensort is an SDK developed by inflight for high performance deep learning reasoning, after training the neural network, the tensort can compress, optimize and run-time deploy the network without the overhead of a framework. The TensorRT improves the delay, throughput and efficiency of the network by optimizing selection of combinations layers and kernel, and performing normalization and conversion to an optimal matrix math method according to specified accuracy. The model is made as lightweight as possible and accelerated as possible at the terminal. The py file trained with Pytorch is converted to a file of TensorRT, which is then deployed on Jetson Nano, injeida. A speed of 20 frames per second can be achieved. Jetson Nano is an artificial intelligence computing module introduced by NVIDIA, has small appearance, can process a plurality of sensors in parallel, can run a plurality of modern neural networks on each sensor, supports a plurality of common artificial intelligence frameworks, and is suitable for edge computing deployment. The operation steps of the invention can be simply summarized as training the own YOLOv5 model on the host, converting into a TensorRT model, deploying the TensorRT model on the Jetson Nano, and operating by deep stream. Hardware environment: RTX 2080TI host Jetson Nano 4G B01; software environment: jetson Nano, ubuntu 18.04, jetpack 4.5.1, deep stream 5.1; and (3) a host computer: ubuntu 18.04, CUDA 10.2, yolov 5.0; the specific flow chart is shown in fig. 3; training model (on platform): the model used in the present invention is yolov5. The single-stage yolov5 detection algorithm is adopted, a complex manual process of extracting the bridge surface defect characteristics is omitted, and the deep semantic characteristics of the bridge surface defects are automatically extracted by using the deep convolutional neural network, so that the detection algorithm has higher adaptability and robustness. Meanwhile, the detection problem is solved as a regression problem, the bridge gap detection and positioning can be completed by only once passing through a network, and the detection speed is considered on the basis of ensuring high detection precision; preparation environment: preparing more than python3.8 environment, creating a virtual environment by using conda, and installing the dependence in yolov5/requirements. Txt under yolov5 project; preparing a data set: and providing a corresponding labeling normal form according to the existing bridge surface damage standard, and manually labeling according to the labeling normal form. And labeling by labellmg to obtain a labeling file of each category rectangular position of each picture. For images which appear in the dataset with a relatively low contrast and are not easily marked by the naked eye. By using the image enhancement technology, the image contrast is improved, and the bridge crack details are enhanced. Finally, the number of images, the number of marking frames and the number of small targets in each category are obtained. Converting the voc format into a yolo format data set by using a label-marked voc format data set (the voc format data set can be directly marked by using a label image or by using a label image) to generate an images folder (all pictures are stored), a labels folder (marked labels are stored), a test. Txt (test set), a train. Txt (training set) and a val. Txt (verification set); creating a configuration file: creating a dataset configuration file dataset. Model configuration file yaml is created, and a model to be trained is copied in yolov5/models under yolov5 item to be modified. Training: modifying a parameter path data set configuration file path, a model configuration file path, a pre-training weight file path, a CUDA device or CPU, a picture folder path or camera to be identified, a weight path and a display identification result according to actual conditions. Transfer TensorRT: the tensorrtx/yolov5/gen_wts.py was copied to the yolov5 project root directory to execute the command generation wts file. The environment goes to Jetson Nano: a tensorrtx project was also cloned on nano and the generated wts was placed under tensorrtx/yolov 5/and tensorrtx/yolov 5/yolayer.h was modified. Deployment with deep stream (on Nano): the project was cloned. YOLOv 5-based bridge crack target detection algorithm. Is a single-stage general target detection algorithm based on deep learning and a neural network. The object detection problem is converted into a regression problem. The model is light, and the reasoning speed is high, so that the model is optimized end to end. The accuracy and the speed are continuously improved through the gradual iterative optimization of the backbone network, the loss function, the multi-scale fusion, the detection head, the data enhancement, the anchor and other algorithm skills of the v1-v4 algorithm. Jetson has the advantage of having a TensorRT addition. TensorRT can improve the network reasoning performance several times. When using neural networks for reasoning, no back propagation is required, here reducing the use of large amounts of temporary storage. The reasoning framework can fuse part of layers, so that IO performance is improved. In addition, the input size of the network is generally fixed, the network is frozen, and the video memory can be distributed more reasonably. In addition to layer fusion, tensor fusion is performed by TensorRT, and methods such as kernel automatic adjustment and the like are automatically optimized, and FP32 and FP16 can be mixed and used on a Jetson platform. The real-time yolov5 in the current detection algorithm is selected and the Jetson Nano artificial intelligent edge computing equipment is used. Compared with the prior edge computing device, the SOC system-on-chip is built in.
In addition, edge computing is on the side close to the object or data source, and an open platform integrating network, computing, storage and application core capabilities is adopted to provide nearest-end services nearby. The application program is initiated at the edge side, faster network service response is generated, and the basic requirements of real-time service and application intelligence are met. The cloud computing can still access the historical data of the edge computing.
In one embodiment, jetson nano analysis data generated based on a neural network model and a surface picture of a bridge to be tested performs the following operations:
acquiring a surface picture of at least one bridge to be detected and a position of the surface picture corresponding to the bridge to be detected;
numbering the surface pictures based on the positions;
inputting each surface picture into a neural network model, and obtaining output data corresponding to each surface picture;
and correlating the numbers of the surface pictures with the output data to form analysis data.
The working principle and the beneficial effects of the technical scheme are as follows:
the surface pictures of the bridge to be detected are mainly pictures of positions such as road surface pictures, side surfaces and bottom surfaces of the travelling crane; the pictures can adopt extraction key positions; the key position is determined empirically according to the structure of the bridge and historical crack detection data; the surface picture corresponds to the position of the bridge to be detected, namely the surface picture is the position of the surface on the bridge to be detected, which is shot; jetson nano inputs the surface picture shot by the camera equipment into a neural network model configured by a platform to acquire output data; the output data includes: the number of cracks in unit area, the maximum crack depth, the area of the maximum crack, the area occupation ratio of the cracks in unit area and the like; numbering the surface pictures according to the positions corresponding to the surface pictures, distinguishing the surface pictures, correlating the numbers with output data, and constructing analysis data; the method is convenient for determining the position of the bridge corresponding to the analysis data, and is convenient for a user to trace back.
In one embodiment, determining bridge crack detection results based on the analysis data includes:
analyzing the analysis data to construct an analysis feature set;
acquiring a preset detection library;
matching the analysis feature set with each standard feature set in the detection library one by one;
when the analysis feature set is matched with the standard feature set, obtaining a bridge crack detection result correspondingly associated with the standard feature set in the detection library;
analyzing the analysis data to construct an analysis feature set, including:
sampling analysis data according to a preset analysis data sampling rule, and obtaining N groups of sampling data altogether; n is greater than or equal to 1;
the output data in the sampling data of each group are arranged in sequence according to the serial numbers to form one row of data;
arranging each row of data from top to bottom according to the time sequence of each group of sampling data to form an analysis feature set; wherein the sample data closest to the current time is located in the first row.
The working principle and the beneficial effects of the technical scheme are as follows:
the platform carries out detection analysis of the system through a detection library; the detection analysis mainly determines the influence of cracks on the service life of the bridge, and a user can conveniently deal with the influence according to the detection result. Before matching the analysis feature set and the standard feature set, carrying out normalization processing on the numerical values in the analysis feature set; when the analysis feature set is matched with the standard feature set, a mode of calculating similarity can be adopted, and a similarity calculation formula is as follows: x ij data of an ith row and a jth column in the feature set are analyzed; y is ij For the purpose of markingData of an ith row and a jth column in the quasi feature set; n is the number of lines of the analysis feature set or the standard feature set; m is the number of columns of the analysis feature set or the standard feature set; x is similarity; when the similarity is the largest in the detection library, determining that the two are matched; the detection library is constructed after analysis and summarization based on a large amount of data in advance.
In one embodiment, obtaining a surface picture of a bridge to be measured includes:
shooting first pictures of the bridge to be detected through a plurality of cameras respectively shooting a plurality of preset first positions of the bridge to be detected;
shooting a second picture of the bridge deck of the bridge to be detected, which is used for the vehicle to run, by a bridge detection vehicle;
shooting a third picture of the outer side of the bridge to be detected by unmanned aerial vehicle shooting equipment;
the working principle and the beneficial effects of the technical scheme are as follows:
the method comprises the steps that three ways are adopted for obtaining the surface pictures of bridge detection, firstly, the bridge is shot through a camera arranged at a fixed point, and the bridge can be arranged at any position through a bracket; the second type is a bridge inspection vehicle; the bridge detection vehicle is a tool capable of running on the running surface of a bridge and is provided with a camera capable of shooting; thirdly, shooting through an unmanned aerial vehicle; of course, before shooting, all need to be connected to an edge computing terminal; the camera arranged at the fixed point can be connected to the edge computing terminal through a cable; bridge detects car and unmanned aerial vehicle and is connected to edge computing terminal through bluetooth.
In order to achieve the determination of the position of the surface picture, in an embodiment, obtaining the position of the surface picture corresponding to the bridge to be tested includes:
acquiring a preset three-dimensional space containing a bridge to be detected;
acquiring setting parameters and first shooting parameters of a camera; the setting parameters comprise: relative positional relation with the bridge, camera setting angle etc., first shooting parameters: including shooting direction, focal length, etc.;
mapping the camera to a three-dimensional space based on the setting parameters;
determining the position of a bridge to be detected corresponding to a first picture corresponding to a camera based on the first shooting parameter;
acquiring first positioning information and second shooting parameters of the bridge detection vehicle in shooting a second picture;
mapping the bridge inspection vehicle to a three-dimensional space based on the first positioning information;
determining the position of the bridge to be detected corresponding to the second picture based on the second shooting parameters;
acquiring second positioning information of the unmanned aerial vehicle shooting equipment when the unmanned aerial vehicle shooting equipment shoots a third picture;
mapping the unmanned aerial vehicle shooting equipment to a three-dimensional space based on the second positioning information;
and determining the position of the bridge to be detected corresponding to the third picture based on the third shooting parameter.
In one embodiment, obtaining at least one surface image of the bridge to be measured and a position of the surface image corresponding to the bridge to be measured includes:
carrying out region extraction on a preset region of the surface picture;
acquiring a first region picture;
acquiring a preset position determination library;
matching the first region picture with each standard picture in the position determining library, and obtaining position information corresponding to the standard picture matched with the first region picture;
and analyzing the position information and determining the position of the bridge to be detected corresponding to the surface picture.
The working principle and the beneficial effects of the technical scheme are as follows:
the first area is the middle part or other positions of the surface picture, and the preset area of the surface picture is extracted, namely the preset area is extracted, and the area is extracted from the surface picture; the position determination is carried out through the position determination library, so that when the surface picture is independently extracted, the position can be determined from the picture; binding the position with the picture to ensure the accuracy of position determination; the specific application is that the position calibration is realized by attaching a two-dimensional code or a mark pattern to the bridge position corresponding to the surface picture; of course, when shooting, shooting parameters of the camera can be corrected through the calibrated marks, so that accurate picture acquisition is realized; in the position determining library, the position information is associated with the standard picture; the location determination library is also built in advance.
In one embodiment, numbering the surface pictures based on location includes:
acquiring a preset three-dimensional space containing a bridge to be detected;
determining coordinate values in a three-dimensional space corresponding to the surface picture based on the position of the bridge to be detected corresponding to the surface picture;
and arranging the coordinate values in sequence to form the number of the surface picture.
The working principle and the beneficial effects of the technical scheme are as follows:
by associating the coordinate values with the numbers of the surface pictures, a user can conveniently position the abnormal positions of the bridge on site according to the abnormal pictures.
The invention provides a crack detection system based on a Jetson Nano bridge, as shown in figure 4, comprising:
the configuration module 1 is used for configuring a neural network model for the Jetson Nano arranged at the bridge detection workstation;
the analysis data receiving module 2 is used for receiving analysis data generated based on a neural network model of the Jetson Nano and a surface picture of the bridge to be detected;
and the detection module 3 is used for determining a bridge crack detection result based on the analysis data.
In one embodiment, jetson nano analysis data generated based on a neural network model and a surface picture of a bridge to be tested performs the following operations:
acquiring a surface picture of at least one bridge to be detected and a position of the surface picture corresponding to the bridge to be detected;
numbering the surface pictures based on the positions;
inputting each surface picture into a neural network model, and obtaining output data corresponding to each surface picture;
and correlating the numbers of the surface pictures with the output data to form analysis data.
In one embodiment, the detection module determines a bridge crack detection result based on the analysis data, performing the following:
analyzing the analysis data to construct an analysis feature set;
acquiring a preset detection library;
matching the analysis feature set with each standard feature set in the detection library one by one;
when the analysis feature set is matched with the standard feature set, obtaining a bridge crack detection result correspondingly associated with the standard feature set in the detection library;
analyzing the analysis data to construct an analysis feature set, including:
sampling analysis data according to a preset analysis data sampling rule, and obtaining N groups of sampling data altogether; n is greater than or equal to 1;
the output data in the sampling data of each group are arranged in sequence according to the serial numbers to form one row of data;
arranging each row of data from top to bottom according to the time sequence of each group of sampling data to form an analysis feature set; wherein the sample data closest to the current time is located in the first row.
In one embodiment, jetson nano acquires a surface picture of a bridge to be measured, including:
shooting first pictures of the bridge to be detected through a plurality of cameras respectively shooting a plurality of preset first positions of the bridge to be detected;
shooting a second picture of the bridge deck of the bridge to be detected, which is used for the vehicle to run, by a bridge detection vehicle;
shooting a third picture of the outer side of the bridge to be detected by unmanned aerial vehicle shooting equipment;
jetson Nano obtains the position of the surface picture corresponding to the bridge to be tested, including:
acquiring a preset three-dimensional space containing a bridge to be detected;
acquiring setting parameters and first shooting parameters of a camera;
mapping the camera to a three-dimensional space based on the setting parameters;
determining the position of a bridge to be detected corresponding to a first picture corresponding to a camera based on the first shooting parameter;
acquiring first positioning information and second shooting parameters of the bridge detection vehicle in shooting a second picture;
mapping the bridge inspection vehicle to a three-dimensional space based on the first positioning information;
determining the position of the bridge to be detected corresponding to the second picture based on the second shooting parameters;
acquiring second positioning information of the unmanned aerial vehicle shooting equipment when the unmanned aerial vehicle shooting equipment shoots a third picture;
mapping the unmanned aerial vehicle shooting equipment to a three-dimensional space based on the second positioning information;
and determining the position of the bridge to be detected corresponding to the third picture based on the third shooting parameter.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (6)
1. The jetson nano bridge crack detection method is characterized by comprising the following steps of:
configuring a neural network model for the JetsonNano arranged at the bridge detection workstation;
receiving analysis data generated by the Jetson Nano based on the neural network model and the surface picture of the bridge to be detected;
determining a bridge crack detection result based on the analysis data;
the jetson nano analysis data generated based on the neural network model and the surface picture of the bridge to be detected execute the following operations:
acquiring the surface picture of at least one bridge to be detected and the position of the surface picture corresponding to the bridge to be detected;
numbering the surface pictures based on the positions;
inputting each surface picture into the neural network model, and obtaining output data corresponding to each surface picture;
correlating the number of each surface picture with the output data to form the analysis data;
the obtaining the surface picture of the bridge to be detected includes:
shooting first pictures of the bridge to be detected through a plurality of cameras respectively shooting a plurality of preset first positions of the bridge to be detected;
shooting a second picture of the bridge deck of the bridge to be detected, which is used for the vehicle to run, by a bridge detection vehicle;
shooting a third picture of the outer side of the bridge to be detected through unmanned aerial vehicle shooting equipment;
the obtaining the position of the surface picture corresponding to the bridge to be detected comprises the following steps:
acquiring a preset three-dimensional space containing the bridge to be detected;
acquiring setting parameters and first shooting parameters of the camera;
mapping the camera to the three-dimensional space based on the setting parameters;
determining the position of the bridge to be detected corresponding to the first picture corresponding to the camera based on the first shooting parameter;
acquiring first positioning information and second shooting parameters of the bridge detection vehicle in shooting the second picture;
mapping the bridge inspection vehicle to the three-dimensional space based on the first positioning information;
determining the position of the bridge to be detected corresponding to the second picture based on the second shooting parameters;
acquiring second positioning information of the unmanned aerial vehicle shooting equipment when the unmanned aerial vehicle shooting equipment shoots the third picture;
mapping the unmanned aerial vehicle shooting device to the three-dimensional space based on the second positioning information;
and determining the position of the bridge to be detected corresponding to the third picture based on a third shooting parameter.
2. The jetson nano bridge crack detection method of claim 1, wherein the determining a bridge crack detection result based on the analysis data comprises:
analyzing the analysis data to construct an analysis feature set;
acquiring a preset detection library;
matching the analysis feature set with each standard feature set in the detection library one by one;
when the analysis feature set is matched with the standard feature set, obtaining the bridge crack detection result correspondingly associated with the standard feature set in the detection library;
wherein analyzing the analysis data to construct the analysis feature set includes:
sampling the analysis data according to a preset analysis data sampling rule, and obtaining N groups of sampling data altogether; n is greater than or equal to 1;
arranging the output data in the sampling data of each group according to the sequence of the numbers to form one row of data;
arranging each row of data from top to bottom according to the time sequence of each group of sampling data to form the analysis feature set; wherein the sample data closest to the current time is located in the first row.
3. The jetson nano bridge crack detection method according to claim 1, wherein the obtaining the surface picture of at least one bridge to be detected and the position of the surface picture corresponding to the bridge to be detected includes:
carrying out region extraction on a preset region of the surface picture;
acquiring a first region picture;
acquiring a preset position determination library;
matching the first region picture with each standard picture in the position determination library, and obtaining the position information corresponding to the standard picture matched with the first region picture;
and analyzing the position information and determining the position of the bridge to be detected corresponding to the surface picture.
4. The jetson nano bridge crack detection method of claim 1, wherein the numbering the surface pictures based on the positions comprises:
acquiring a preset three-dimensional space containing the bridge to be detected;
determining coordinate values in the three-dimensional space corresponding to the surface picture based on the position of the bridge to be detected corresponding to the surface picture;
and arranging the coordinate values in sequence to form the serial numbers of the surface pictures.
5. Jetson Nano bridge crack detection system, which is characterized in that:
the configuration module is used for configuring a neural network model for the Jetson Nano arranged at the bridge detection workstation;
the analysis data receiving module is used for receiving analysis data generated by the Jetson Nano based on the neural network model and the surface picture of the bridge to be detected;
the detection module is used for determining a bridge crack detection result based on the analysis data;
the jetson nano analysis data generated based on the neural network model and the surface picture of the bridge to be detected execute the following operations:
acquiring the surface picture of at least one bridge to be detected and the position of the surface picture corresponding to the bridge to be detected;
numbering the surface pictures based on the positions;
inputting each surface picture into the neural network model, and obtaining output data corresponding to each surface picture;
correlating the number of each surface picture with the output data to form the analysis data;
the jetson nano obtaining the surface picture of the bridge to be detected comprises the following steps:
shooting first pictures of the bridge to be detected through a plurality of cameras respectively shooting a plurality of preset first positions of the bridge to be detected;
shooting a second picture of the bridge deck of the bridge to be detected, which is used for the vehicle to run, by a bridge detection vehicle;
shooting a third picture of the outer side of the bridge to be detected through unmanned aerial vehicle shooting equipment;
the jetson nano obtaining the position of the surface picture corresponding to the bridge to be detected comprises the following steps:
acquiring a preset three-dimensional space containing the bridge to be detected;
acquiring setting parameters and first shooting parameters of the camera;
mapping the camera to the three-dimensional space based on the setting parameters;
determining the position of the bridge to be detected corresponding to the first picture corresponding to the camera based on the first shooting parameter;
acquiring first positioning information and second shooting parameters of the bridge detection vehicle in shooting the second picture;
mapping the bridge inspection vehicle to the three-dimensional space based on the first positioning information;
determining the position of the bridge to be detected corresponding to the second picture based on the second shooting parameters;
acquiring second positioning information of the unmanned aerial vehicle shooting equipment when the unmanned aerial vehicle shooting equipment shoots the third picture;
mapping the unmanned aerial vehicle shooting device to the three-dimensional space based on the second positioning information;
and determining the position of the bridge to be detected corresponding to the third picture based on a third shooting parameter.
6. The jetson nano bridge crack detection system of claim 5, wherein the detection module determines a bridge crack detection result based on the analysis data, performing the following operations:
analyzing the analysis data to construct an analysis feature set;
acquiring a preset detection library;
matching the analysis feature set with each standard feature set in the detection library one by one;
when the analysis feature set is matched with the standard feature set, obtaining the bridge crack detection result correspondingly associated with the standard feature set in the detection library;
wherein analyzing the analysis data to construct the analysis feature set includes:
sampling the analysis data according to a preset analysis data sampling rule, and obtaining N groups of sampling data altogether; n is greater than or equal to 1;
arranging the output data in the sampling data of each group according to the sequence of the numbers to form one row of data;
arranging each row of data from top to bottom according to the time sequence of each group of sampling data to form the analysis feature set; wherein the sample data closest to the current time is located in the first row.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210497871.3A CN114943693B (en) | 2022-05-09 | 2022-05-09 | Jetson Nano bridge crack detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210497871.3A CN114943693B (en) | 2022-05-09 | 2022-05-09 | Jetson Nano bridge crack detection method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114943693A CN114943693A (en) | 2022-08-26 |
| CN114943693B true CN114943693B (en) | 2023-07-25 |
Family
ID=82907732
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210497871.3A Active CN114943693B (en) | 2022-05-09 | 2022-05-09 | Jetson Nano bridge crack detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114943693B (en) |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106645205A (en) * | 2017-02-24 | 2017-05-10 | 武汉大学 | Unmanned aerial vehicle bridge bottom surface crack detection method and system |
| KR102026449B1 (en) * | 2018-01-12 | 2019-09-27 | 인하대학교 산학협력단 | Simulation Data Preprocessing Technique for Development of Damage Detecting Method for Bridges Based on Convolutional Neural Network |
| CN109300126B (en) * | 2018-09-21 | 2022-01-07 | 重庆建工集团股份有限公司 | High-precision intelligent detection method for bridge diseases based on spatial positions |
| CN111402227B (en) * | 2020-03-13 | 2022-09-23 | 河海大学常州校区 | Bridge crack detection method |
| CN114049560A (en) * | 2021-11-25 | 2022-02-15 | 江苏科技大学 | Road surface multi-feature disease detection method and device based on combination of multiple neural networks |
| CN114419565A (en) * | 2021-12-24 | 2022-04-29 | 天津市水木兄弟科技有限公司 | Special vehicle operation collision early warning method and system based on YOLOv4 |
-
2022
- 2022-05-09 CN CN202210497871.3A patent/CN114943693B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN114943693A (en) | 2022-08-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ukhwah et al. | Asphalt pavement pothole detection using deep learning method based on YOLO neural network | |
| US20210319561A1 (en) | Image segmentation method and system for pavement disease based on deep learning | |
| US20200294222A1 (en) | Method and apparatus for outputting information | |
| Zou et al. | CNN-based statistics and location estimation of missing components in routine inspection of historic buildings | |
| CN110837839B (en) | High-precision unmanned aerial vehicle orthographic image manufacturing and data acquisition method | |
| CN109858414A (en) | A kind of invoice piecemeal detection method | |
| CN108711148A (en) | A kind of wheel tyre defect intelligent detecting method based on deep learning | |
| CN116363125B (en) | Deep learning-based battery module appearance defect detection method and system | |
| KR102308456B1 (en) | Tree species detection system based on LiDAR and RGB camera and Detection method of the same | |
| CN115409069A (en) | Village and town building identification method, classification method, device, electronic equipment and medium | |
| CN111598942A (en) | Method and system for automatically positioning electric power facility instrument | |
| CN111652835A (en) | Method for detecting insulator loss of power transmission line based on deep learning and clustering | |
| CN112446852A (en) | Tunnel imaging plane display method and intelligent defect identification system | |
| CN112613569A (en) | Image recognition method, and training method and device of image classification model | |
| CN116258980A (en) | Unmanned aerial vehicle distributed photovoltaic power station inspection method based on vision | |
| CN110688955A (en) | Building construction target detection method based on YOLO neural network | |
| CN115372877B (en) | Lightning arrester leakage ammeter inspection method of transformer substation based on unmanned aerial vehicle | |
| CN113674216A (en) | Subway tunnel disease detection method based on deep learning | |
| CN110969610A (en) | Power equipment infrared chart identification method and system based on deep learning | |
| CN116681962A (en) | Power equipment thermal image detection method and system based on improved YOLOv5 | |
| CN116543303A (en) | Bridge plate type rubber support disease identification and detection method based on deep learning | |
| CN115171045A (en) | YOLO-based power grid operation field violation identification method and terminal | |
| CN113902793A (en) | End-to-end building height prediction method and system based on single vision remote sensing image and electronic equipment | |
| CN116721104B (en) | Live three-dimensional model defect detection method and device, electronic equipment and storage medium | |
| CN114943693B (en) | Jetson Nano bridge crack detection method and system |
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 |