CN110706211A - Convolutional neural network-based real-time detection method for railway roadbed disease radar map - Google Patents
Convolutional neural network-based real-time detection method for railway roadbed disease radar map Download PDFInfo
- Publication number
- CN110706211A CN110706211A CN201910883896.5A CN201910883896A CN110706211A CN 110706211 A CN110706211 A CN 110706211A CN 201910883896 A CN201910883896 A CN 201910883896A CN 110706211 A CN110706211 A CN 110706211A
- Authority
- CN
- China
- Prior art keywords
- disease
- detection
- neural network
- convolutional neural
- railway roadbed
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 201000010099 disease Diseases 0.000 title claims abstract description 64
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 64
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 25
- 238000011897 real-time detection Methods 0.000 title claims abstract description 9
- 238000001514 detection method Methods 0.000 claims abstract description 46
- 238000012549 training Methods 0.000 claims abstract description 13
- 238000012360 testing method Methods 0.000 claims abstract description 12
- 230000008569 process Effects 0.000 claims abstract description 6
- 238000011478 gradient descent method Methods 0.000 claims abstract description 5
- 238000004364 calculation method Methods 0.000 claims abstract description 3
- 210000004027 cell Anatomy 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 5
- 230000000694 effects Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 239000013598 vector Substances 0.000 claims description 3
- 210000002287 horizontal cell Anatomy 0.000 claims description 2
- 230000001629 suppression Effects 0.000 claims description 2
- 230000009466 transformation Effects 0.000 claims description 2
- 238000003384 imaging method Methods 0.000 claims 1
- 238000004904 shortening Methods 0.000 abstract 1
- 238000013528 artificial neural network Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 230000000149 penetrating effect Effects 0.000 description 2
- 238000005086 pumping Methods 0.000 description 2
- 238000007792 addition Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 239000002002 slurry Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
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
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
-
- G06Q50/40—
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
- G06T2207/10044—Radar image
Abstract
The invention discloses a convolutional neural network-based real-time detection method for a railway roadbed disease radar map, wherein the method comprises the following steps: and marking the roadbed disease radar image to divide the roadbed disease radar image into a training set and a test set, expanding the training set, sending the training set into a convolutional neural network, outputting a disease type, a position coordinate where the disease is located and a disease confidence coefficient, obtaining a railway roadbed disease detection model through iterative calculation of a gradient descent method, and adopting average mean precision and frames per second as indexes for evaluating the quality of the model. The method fully utilizes the multi-scale prediction network, and the whole process has no step of generating a candidate region, thereby greatly shortening the detection time while ensuring the precision and realizing real-time detection.
Description
Technical Field
The invention relates to the technical field of railway roadbed disease detection and radar image intelligent identification, in particular to a railway roadbed disease radar map real-time detection method based on a convolutional neural network.
Background
The railway subgrade disease problem begins to appear due to the increase of the operation mileage and the time while the railway is told to be developed at the present stage in China. The railway roadbed diseases are major safety problems in the field of railway transportation, and bring hidden dangers to railway safe operation. How to rapidly and accurately identify potential roadbed disease risk sources from massive railway roadbed detection data, ensure railway transportation safety and become a technical problem which needs to be solved urgently.
The vehicle-mounted geological radar detection technology is rapidly popularized and applied in the field of railway roadbed detection due to the characteristics of no damage, high efficiency, high precision, visual result and the like. Compared with the traditional method, the geological radar can quickly find out the roadbed situation of the whole line, and is a main technical means for daily detection of the railway roadbed. The railway roadbed geological radar detection data belong to mass data, about 23G of geological radar data are generated every 30 kilometers of detection on average, however, at present, the railway roadbed disease identification still depends on manual interpretation of radar images for identification, the cost is high, the efficiency is low, and the requirements on knowledge and experience of operators are high. The existing detection technology mainly focuses on the combination of artificial design features and traditional machine learning methods such as a support vector machine and a shallow neural network, and the methods are time-consuming and low in accuracy. The existing convolutional neural network identification technology based on the candidate region cannot meet the real-time processing requirement and is difficult to bear the mass data detection task of the geological radar.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides the railway roadbed disease radar map real-time detection method based on the convolutional neural network, which utilizes the depth residual error network to extract radar image characteristics, and adopts the multi-scale prediction network to replace a pooling layer and a full connection layer, thereby reducing the calculated amount, ensuring the detection speed while effectively improving the detection precision of the neural network, and ensuring the detection speed, wherein the detection speed of the method reaches 0.038 seconds per sheet (26 frames per second) under the NVIDIA RTX2080 hardware environment, so that the method can meet the real-time engineering requirements, and can be used in the scientific fields of railway roadbed disease identification and the like.
The technical scheme adopted by the invention is as follows:
marking the roadbed disease image, and generating an XML marking file corresponding to the included object;
step (2), dividing the disease image and the XML file obtained after marking in the step (1) into a training set and a testing set, wherein the training set accounts for 90%, the testing set accounts for 10%, and the training set and the testing set both contain the radar disease image and the disease marking information in the XML format;
step (3) building a convolutional neural network, and taking the radar map and the label file as the input of the network, the type of the disease, the position coordinate of the disease and the confidence coefficient of the disease as the output of the network;
step (4), expanding the disease image of the training set obtained in the step (2) and the XML file, sending the expanded disease image and the XML file into the convolutional neural network in the step (3) as input, and obtaining a railway roadbed disease detection model through iterative calculation of a gradient descent method;
step (5), the test set obtained in the step (2) is used for checking the model effect in the step (4), and the average detection precision and the number of frames per second are used as indexes for evaluating the quality of the model;
and (6) using the model obtained in the step (4) as a railway roadbed disease detection model to realize intelligent detection of railway roadbed diseases.
The method has the beneficial effects that compared with the existing method for detecting the railway roadbed disease radar map, the method for detecting the railway roadbed disease radar map based on the convolutional neural network has the advantages that the neural network can realize real-time detection while ensuring high-precision detection.
Drawings
The invention is further illustrated with reference to the following figures and examples.
FIG. 1 is a general flow diagram according to one embodiment of the invention;
FIG. 2 is a diagram of a convolutional neural network structure, according to one embodiment of the present invention;
fig. 3 is an exemplary diagram of a detection result of a sinking fault of a railway roadbed according to an embodiment of the invention;
fig. 4 is an exemplary diagram of a detection result of the railroad bed slurry pumping according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings. The embodiments described with reference to the drawings are illustrative and are intended to be illustrative of the invention and should not be construed as limiting the invention.
Before introducing the method for detecting the railway roadbed disease radar map in real time based on the convolutional neural network, the data selected in the embodiment is introduced. The railway roadbed defect map in the data set is a railway roadbed radar image obtained by a vehicle-mounted ground penetrating radar (RIS vehicle radar of Italy IDS company), and the data set is constructed by 403 pictures of the railway roadbed defect map.
FIG. 1 is a general flow diagram according to one embodiment of the invention;
as shown in fig. 1, the method for detecting the railway roadbed disease radar map based on the convolutional neural network in real time comprises the following steps:
s1010: marking the damaged area of the image by using a railway roadbed damaged radar image acquired from a vehicle-mounted ground penetrating radar, wherein the marking information comprises the upper left corner coordinate, the lower right corner coordinate, the size of the damaged image and the type of the damaged area, and packaging the marking information to generate XML files, and the file name of each XML file is the same as the file name of the damaged image.
S1020: and randomly dividing the disease graph and the XML file marked in the S1010 into a training set and a testing set, wherein the training set accounts for 90%, and the testing set accounts for 10%.
S1030: and randomly expanding the roadbed disease image in the S1020, wherein the expanding method comprises the following steps: horizontal mirroring, random cutting and contrast adjustment, and the neural network can have better robustness and improve the identification precision by expanding a data set.
S1040: the structure of the depth residual error network is shown in fig. 2, before the depth residual error network is sent to a neural network, a disease graph is taken as a target, the resolution of 416 × 416 is taken as network input, a radar image is divided into S × S grids, if a disease center to be detected falls in a certain grid, the grid is responsible for predicting the disease, then forward propagation is carried out in the convolutional neural network for feature extraction, a pooling layer and a full connection layer do not exist in the network, size transformation of the feature graph is achieved by changing the step size of a convolution kernel, the network takes the whole graph as the input of the network, then feature extraction is carried out through the whole convolutional neural network, and no step of generating a candidate area is carried out in the process.
S1110, S1210, S1310: as shown in fig. 2, as the network goes deep, three different scales of feature maps are generated, where the first scale is 52 × 52, the second scale is 26 × 26, and the third scale is 13 × 13, and predictions of different scales are performed on three scales of the neural network.
S1410: as shown in fig. 2, feature fusion is performed on three scales by convolution and upsampling, and detection is performed on the fusion feature maps of multiple scales independently, so that detection of disease targets of different sizes is realized.
S1420: the neural network outputs vectors with dimensions of S multiplied by (3 multiplied by 5+ T) on three scales of S1110, S1210 and S1310; wherein S represents the number of the cells in the corresponding horizontal row of each image, 3 represents the predicted number of the boundary frames of each cell, 5 represents the position and confidence of the predicted detection frame, and T represents the predicted disease category. In the process of minimizing the loss function by continuously updating the neural network parameters through the gradient descent method, the loss function is:
in the formula, Loss is a Loss function, S represents the number of horizontal cells corresponding to each graph, B represents the number of predictable bounding boxes of each cell, and lambdacoordAnd λnoobjFor the penalty term coefficient, set to λcoord5 and λnoobj＝0.5，
The true values of the center abscissa, ordinate, width and height of the position of the ith disease (x)i,yi,wi,hi) Is a predicted value of the center abscissa, ordinate, width and height of the position of the ith disease,
representing whether the ith prediction box of the ith lattice has a target of 1 or not and no 0,
the area intersection ratio of the bounding box in the label and the predicted bounding box, CiAs the product of the confidence and the intersection ratio of the areas of the bounding box and the predicted bounding box in the label, classes represents the target class,probability of belonging to a given class for a target, 1, not 0, pi(c) To predict that a class belongs to a given probability, the range is between 0 and 1. After the bounding box is obtained, the detection result is obtained by non-maximum suppression (eliminating the bounding box with the overlapping rate higher than 40%).
S1430: and (3) evaluating the model generated in the step (1420) by using the test set data divided in the step (1020), and using the average precision and the number of frames per second as indexes for evaluating the quality of the model, wherein the average precision is the average value of the areas under the PR curves (precision and recall curves) of all categories, and the number of frames per second is the number of pictures which can be predicted in each second.
S1440: and (4) the optimal model obtained in the S1430 is used as a railway roadbed disease detection evaluation model to realize intelligent detection of railway roadbed diseases.
In order to illustrate the superiority of the detection effect of the method, in this embodiment, the detection speed reaches 0.038 seconds per sheet (26 frames per second) under an ubuntu18.04 operating system and an NVIDIARTX2080 platform, and the average detection precision reaches 0.95, so that the precision is guaranteed and the real-time detection can be realized at the same time.
Fig. 3 and 4 show radar map detection results of a railway roadbed damage according to an embodiment of the invention, and as can be seen from fig. 3 and 4, the invention can effectively detect railway roadbed subsidence and slurry-dumping and mud-pumping damage.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Claims (3)
1. The method for detecting the railway roadbed disease radar map in real time based on the convolutional neural network is established on the theoretical basis of radar data on railway roadbed detection imaging, and is characterized by sequentially comprising the following steps in the process of identifying the railway roadbed radar image:
marking the roadbed disease image, and generating an XML marking file corresponding to the included object;
step (2), dividing the disease image and the XML file obtained after marking in the step (1) into a training set and a testing set, wherein the training set accounts for 90%, the testing set accounts for 10%, and the training set and the testing set both contain the radar disease image and the disease marking information in the XML format;
step (3) building a convolutional neural network, and taking the radar map and the label file as the input of the network, the type of the disease, the position coordinate of the disease and the confidence coefficient of the disease as the output of the network;
step (4), expanding the disease image of the training set obtained in the step (2) and the XML file, sending the expanded disease image and the XML file into the convolutional neural network in the step (3) as input, and obtaining a railway roadbed disease detection model through iterative calculation of a gradient descent method;
step (5), the test set obtained in the step (2) is used for checking the model effect in the step (4), and the average detection precision and the number of frames per second are used as indexes for evaluating the quality of the model;
and (6) using the model obtained in the step (4) as a railway roadbed disease detection model to realize intelligent detection of railway roadbed diseases.
2. The convolutional neural network-based radar signal railway roadbed disease detection method as claimed in claim 1, wherein in the step (3), the convolutional neural network has a specific structure that radar image features are extracted through a depth residual error network, then a plurality of convolutional layers are added behind the residual error network and divided into three branches to form a multi-scale prediction network, then feature maps extracted through the depth residual error network are up-sampled and feature fusion is performed on three scales, detection is independently performed on the fusion feature maps of the multiple scales, so that detection on targets with different sizes is realized, and finally a detection frame with a high coincidence rate is removed by using a non-maximum suppression method to obtain a final railway roadbed disease detection model.
3. The method for detecting the railway roadbed fault based on the radar signal of the convolutional neural network as claimed in claim 2, wherein the convolutional neural network is free of a pooling layer and a full connection layer, the size transformation of the feature map is realized by changing the step length of a convolutional kernel, the network takes the whole map as the input of the network, then the feature extraction is carried out through the whole convolutional neural network, the final model directly regresses a bounding box at an output layer under three different scales, the fault category and the fault confidence coefficient are obtained, the whole process is free of the step of generating a candidate area, and multi-scale feature fusion is adopted, so that the detection precision is greatly improved, the detection time is shortened, and the real-time detection is realized.
When the model is used for detection, firstly, a radar image is divided into S multiplied by S grids, if a center of a disease to be detected falls in a certain grid, the grid is responsible for predicting the disease, after the disease is transmitted to an output layer through a convolutional neural network in a forward direction, each grid can output 3 vectors, finally, the convolutional neural network outputs a vector of S multiplied by (3 multiplied by 5+ T) dimension on three scales respectively, wherein S represents the number of horizontal-row cells corresponding to each image, 3 represents the number of predicted boundary frames of each cell, 5 represents the position and confidence coefficient of a predicted detection frame, and T represents the type of the predicted disease.
In the above step, in the process of minimizing the loss function by continuously updating the parameters by the gradient descent method, the loss function is:
in the formula, Loss is a Loss function, S represents the number of horizontal cells corresponding to each graph, B represents the number of predictable bounding boxes of each cell, and lambdacoordAnd λnoobjFor the penalty term coefficient, set to λcoord5 and λnoobj＝0.5，
The true values of the center abscissa, ordinate, width and height of the position of the ith disease (x)i,yi,wi,hi) Is a predicted value of the center abscissa, ordinate, width and height of the position of the ith disease,
representing whether the ith prediction box of the ith lattice has a target of 1 or not and no 0,
the area intersection ratio of the bounding box in the label and the predicted bounding box, CiAs the product of the confidence and the intersection ratio of the areas of the bounding box and the predicted bounding box in the label, classes represents the target class,
probability of belonging to a given class for a target, 1, not 0, pi(c) To predict that a class belongs to a given probability, the range is between 0 and 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910883896.5A CN110706211A (en) | 2019-09-17 | 2019-09-17 | Convolutional neural network-based real-time detection method for railway roadbed disease radar map |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910883896.5A CN110706211A (en) | 2019-09-17 | 2019-09-17 | Convolutional neural network-based real-time detection method for railway roadbed disease radar map |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110706211A true CN110706211A (en) | 2020-01-17 |
Family
ID=69194908
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910883896.5A Pending CN110706211A (en) | 2019-09-17 | 2019-09-17 | Convolutional neural network-based real-time detection method for railway roadbed disease radar map |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110706211A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111340764A (en) * | 2020-02-20 | 2020-06-26 | 江苏东印智慧工程技术研究院有限公司 | Automatic inhaul cable apparent disease counting method based on deep sort |
| CN112380997A (en) * | 2020-11-16 | 2021-02-19 | 武汉巨合科技有限公司 | Model identification and undercarriage retraction and extension detection method based on deep learning |
| CN112396128A (en) * | 2020-12-08 | 2021-02-23 | 中国铁路设计集团有限公司 | Automatic labeling method for railway external environment risk source sample |
| CN112462346A (en) * | 2020-11-26 | 2021-03-09 | 西安交通大学 | Ground penetrating radar roadbed defect target detection method based on convolutional neural network |
| CN113516053A (en) * | 2021-05-28 | 2021-10-19 | 西安空间无线电技术研究所 | Ship target refined detection method with rotation invariance |
| CN114818998A (en) * | 2022-06-28 | 2022-07-29 | 浙江大学 | Method for judging mud pumping disease state of ballastless track foundation bed during slurry turning |
| CN116229396A (en) * | 2023-02-17 | 2023-06-06 | 广州丰石科技有限公司 | High-speed pavement disease identification and warning method |
| CN116363530A (en) * | 2023-03-14 | 2023-06-30 | 北京天鼎殊同科技有限公司 | Method and device for positioning expressway pavement diseases |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107621626A (en) * | 2017-10-09 | 2018-01-23 | 中国矿业大学(北京) | Radar signal Railway Roadbed detection method based on depth convolutional neural networks |
| CN109902677A (en) * | 2019-01-30 | 2019-06-18 | 深圳北斗通信科技有限公司 | A kind of vehicle checking method based on deep learning |
-
2019
- 2019-09-17 CN CN201910883896.5A patent/CN110706211A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107621626A (en) * | 2017-10-09 | 2018-01-23 | 中国矿业大学(北京) | Radar signal Railway Roadbed detection method based on depth convolutional neural networks |
| CN109902677A (en) * | 2019-01-30 | 2019-06-18 | 深圳北斗通信科技有限公司 | A kind of vehicle checking method based on deep learning |
Non-Patent Citations (3)
| Title |
|---|
| JOSEPH REDMON ET AL.: "You Only Look Once:Unified, Real-Time Object Detection", 《ARXIV:1506.02640V5[CS.CV]》 * |
| 肥羊汤: "YOLOV3", 《CSDN》 * |
| 薛金利等: "基于低空无人机影像和 YOLOv3 实现棉田杂草检测", 《石河子大学学报（自然科学版）》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111340764A (en) * | 2020-02-20 | 2020-06-26 | 江苏东印智慧工程技术研究院有限公司 | Automatic inhaul cable apparent disease counting method based on deep sort |
| CN112380997A (en) * | 2020-11-16 | 2021-02-19 | 武汉巨合科技有限公司 | Model identification and undercarriage retraction and extension detection method based on deep learning |
| CN112462346B (en) * | 2020-11-26 | 2023-04-28 | 西安交通大学 | Ground penetrating radar subgrade disease target detection method based on convolutional neural network |
| CN112462346A (en) * | 2020-11-26 | 2021-03-09 | 西安交通大学 | Ground penetrating radar roadbed defect target detection method based on convolutional neural network |
| CN112396128A (en) * | 2020-12-08 | 2021-02-23 | 中国铁路设计集团有限公司 | Automatic labeling method for railway external environment risk source sample |
| CN112396128B (en) * | 2020-12-08 | 2022-06-03 | 中国铁路设计集团有限公司 | Automatic labeling method for railway external environment risk source sample |
| CN113516053A (en) * | 2021-05-28 | 2021-10-19 | 西安空间无线电技术研究所 | Ship target refined detection method with rotation invariance |
| CN114818998A (en) * | 2022-06-28 | 2022-07-29 | 浙江大学 | Method for judging mud pumping disease state of ballastless track foundation bed during slurry turning |
| CN114818998B (en) * | 2022-06-28 | 2022-09-13 | 浙江大学 | Method for judging mud pumping disease state of ballastless track foundation bed during slurry turning |
| CN116229396A (en) * | 2023-02-17 | 2023-06-06 | 广州丰石科技有限公司 | High-speed pavement disease identification and warning method |
| CN116229396B (en) * | 2023-02-17 | 2023-11-03 | 广州丰石科技有限公司 | High-speed pavement disease identification and warning method |
| CN116363530A (en) * | 2023-03-14 | 2023-06-30 | 北京天鼎殊同科技有限公司 | Method and device for positioning expressway pavement diseases |
| CN116363530B (en) * | 2023-03-14 | 2023-11-03 | 北京天鼎殊同科技有限公司 | Method and device for positioning expressway pavement diseases |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110706211A (en) | Convolutional neural network-based real-time detection method for railway roadbed disease radar map | |
| CN110059554A (en) | A kind of multiple branch circuit object detection method based on traffic scene | |
| CN111178206B (en) | Building embedded part detection method and system based on improved YOLO | |
| CN108875595A (en) | A kind of Driving Scene object detection method merged based on deep learning and multilayer feature | |
| CN113256562B (en) | Road underground hidden danger detection method and system based on radar images and artificial intelligence | |
| CN103268358B (en) | Multi-source control point image database builds and update method | |
| CN107833213A (en) | A kind of Weakly supervised object detecting method based on pseudo- true value adaptive method | |
| CN106226779A (en) | An easily Prediction of Landslide based on remote sensing technology | |
| CN105205808A (en) | Multi-vision image dense coupling fusion method and system based on multiple characteristics and multiple constraints | |
| CN113283802A (en) | Landslide risk assessment method for complex and difficult mountain area | |
| Hacıefendioğlu et al. | Landslide detection using visualization techniques for deep convolutional neural network models | |
| Wu et al. | Enhanced precision in dam crack width measurement: Leveraging advanced lightweight network identification for pixel-level accuracy | |
| CN104166163A (en) | Method for automatically extracting fault curved surface based on three-dimensional large-data-volume seismic data cube | |
| CN104103076A (en) | Nuclear power plant planned restricted zone remote sensing inspecting method based on high-resolution remote sensing images | |
| CN113012063B (en) | Dynamic point cloud repairing method and device and computer equipment | |
| CN113610070A (en) | Landslide disaster identification method based on multi-source data fusion | |
| CN116168246A (en) | Method, device, equipment and medium for identifying waste slag field for railway engineering | |
| Han et al. | Effective complex airport object detection in remote sensing images based on improved end-to-end convolutional neural network | |
| Li et al. | 3D map system for tree monitoring in hong kong using google street view imagery and deep learning | |
| CN112036424B (en) | Submarine landslide risk analysis method based on unsupervised machine learning | |
| Zhu et al. | YOLO-Rip: A modified lightweight network for Rip currents detection | |
| CN117351321A (en) | Single-stage lightweight subway lining cavity recognition method and related equipment | |
| CN114519819B (en) | Remote sensing image target detection method based on global context awareness | |
| KR20160120955A (en) | Module for multi diensional vedio information visualization | |
| CN112907567B (en) | SAR image ordered artificial structure extraction method based on spatial reasoning method |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20200117 |