CN112446870A - Pipeline damage detection method, device, equipment and storage medium - Google Patents
Pipeline damage detection method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN112446870A CN112446870A CN202011385962.5A CN202011385962A CN112446870A CN 112446870 A CN112446870 A CN 112446870A CN 202011385962 A CN202011385962 A CN 202011385962A CN 112446870 A CN112446870 A CN 112446870A
- Authority
- CN
- China
- Prior art keywords
- pipeline
- detection
- damage
- frame
- preset
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 346
- 238000007689 inspection Methods 0.000 claims abstract description 78
- 238000000034 method Methods 0.000 claims abstract description 29
- 238000000605 extraction Methods 0.000 claims description 44
- 230000000875 corresponding effect Effects 0.000 claims description 38
- 238000011156 evaluation Methods 0.000 claims description 34
- 238000010586 diagram Methods 0.000 claims description 26
- 230000010354 integration Effects 0.000 claims description 25
- 230000004927 fusion Effects 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 8
- 238000012423 maintenance Methods 0.000 claims description 7
- 238000012935 Averaging Methods 0.000 claims description 5
- 230000002596 correlated effect Effects 0.000 claims description 5
- 230000007935 neutral effect Effects 0.000 claims description 5
- 238000012800 visualization Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 abstract description 8
- 238000013473 artificial intelligence Methods 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 4
- 238000002790 cross-validation Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 208000025174 PANDAS Diseases 0.000 description 1
- 208000021155 Paediatric autoimmune neuropsychiatric disorders associated with streptococcal infection Diseases 0.000 description 1
- 240000004718 Panda Species 0.000 description 1
- 235000016496 Panda oleosa Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000001052 transient 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20132—Image cropping
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention relates to the field of artificial intelligence and discloses a pipeline damage detection method, a pipeline damage detection device, pipeline damage detection equipment and a storage medium. The method comprises the following steps: acquiring a pipeline inspection video to be detected; inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection, and outputting a detection result; if the detection result is that pipeline damage exists in the current video frame, calling a preset OpenCV interface, and visualizing the five-dimensional vector in the detection result into a detection frame; and combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type. The method performs targeted optimization on the specific task of pipeline damage detection, has better applicability to the task of pipeline damage detection, and greatly improves the efficiency of pipeline damage detection.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a pipeline damage detection method, a pipeline damage detection device, pipeline damage detection equipment and a storage medium.
Background
With the rapid development of computer technology, computer vision has become an important field of artificial intelligence, and plays an increasingly important role in the aspects of people's life. The computer vision technology is widely applied, and the target detection method in the computer vision technology is adopted to identify the target object, so that the key targets in the picture or the video can be effectively extracted, and the identification effect is achieved.
Traditional pipeline inspection mainly relies on manual experience, and the efficiency of utilizing manual experience to carry out damage appraisal very easy mistake and inspection is very low, can't satisfy effective, timely maintenance requirement of a large amount of sewer pipes in the city.
Disclosure of Invention
The invention mainly aims to solve the technical problem of low pipeline damage detection efficiency at present.
The invention provides a pipeline damage detection method in a first aspect, which comprises the following steps:
acquiring a pipeline inspection video to be detected;
inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection, and outputting a detection result;
if the detection result is that pipeline damage exists in the current video frame, calling a preset OpenCV interface, and visualizing the five-dimensional vector in the detection result into a detection frame;
and combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type.
Optionally, in a first implementation manner of the first aspect of the present invention, before the acquiring the video of the sewer line to be detected, the method further includes:
acquiring a plurality of pipeline inspection video samples, and marking the pipeline inspection video samples with damage information frame by frame to obtain damaged positive sample images and non-damaged negative sample images;
inputting the positive sample image and the negative sample image into a preset target detection network for feature extraction to obtain a sample feature map;
according to the sample feature map, calling a preset Autofusion algorithm to search evaluation indexes of a feature extraction layer connecting part of the target detection network, and taking the target detection network corresponding to the combination with the highest evaluation index as an optimal target detection network;
and calling a preset Stacking integration algorithm to integrate the optimal target detection network to obtain a pipeline damage detection model.
Optionally, in a second implementation manner of the first aspect of the present invention, the inputting the positive sample image and the negative sample image into a preset target detection network for feature extraction, and obtaining a sample feature map includes:
inputting the positive sample image and the negative sample image into a preset input layer for data enhancement to obtain an enhanced sample picture;
scaling and cutting the size of the enhanced sample picture to obtain a standard sample picture;
inputting the standard sample picture into a preset CSP network for feature extraction to obtain sample feature information;
and inputting the sample characteristic information into a preset neutral network for characteristic fusion to obtain a sample characteristic diagram.
Optionally, in a third implementation manner of the first aspect of the present invention, the invoking a preset AutoFusion algorithm to search for an evaluation index of a feature extraction layer connection part of the target detection network according to the sample feature map, and taking the target detection network corresponding to a combination with a highest evaluation index as an optimal target detection network includes:
calling a preset Autofusion algorithm, and carrying out unary operation and maintenance operation on the feature extraction layer connecting part of the target detection network to obtain an unary operation value;
inputting the unary operation value into a preset operation layer to perform amplitude function operation to obtain an operand;
combining the unary operation value and the operand to obtain a combination of evaluation indexes;
and taking the target detection network corresponding to the combination with the highest evaluation index as the optimal target detection network.
Optionally, in a fourth implementation manner of the first aspect of the present invention, the invoking a preset Stacking integration algorithm to integrate the optimal target detection network to obtain the pipeline damage detection model includes:
calling a preset Stacking integration algorithm, inputting the sample feature graph into the optimal target detection network for integration operation, and obtaining a first-layer meta-feature;
averaging the first layer meta-feature and inputting the average into the optimal target detection network for integrated operation to obtain a second layer meta-feature;
and according to the second layer meta-feature, performing parameter adjustment on the optimal target detection network until the optimal target detection network converges to obtain a pipeline damage detection model.
Optionally, in a fifth implementation manner of the first aspect of the present invention, the inputting the pipeline inspection video into a preset pipeline damage detection model for performing frame-by-frame detection, and outputting a detection result includes:
inputting the pipeline inspection video into a CSP network in a preset pipeline damage detection model to perform feature extraction frame by frame to obtain feature information;
inputting the characteristic information into a hack network in a preset pipeline damage detection model for characteristic fusion to obtain a characteristic diagram;
and analyzing the category information and the position information of the characteristic diagram, and outputting a detection result.
Optionally, in a sixth implementation manner of the first aspect of the present invention, after the combining the detection frame with a corresponding video frame in the pipeline inspection video to obtain a pipeline inspection marking video marked with a pipeline damage position and a damage type, the method further includes:
playing the pipeline inspection marking video, and judging whether pipeline damage exists in a current video frame;
if so, screenshot is carried out on the current video frame to obtain a pipeline damage picture, and pipeline damage information in the pipeline damage picture is extracted;
and storing the pipeline damage picture, the pipeline damage information and the current video playing time point in a correlated manner, and outputting a CSV format file containing the pipeline damage information.
A second aspect of the present invention provides a pipe damage detection apparatus, including:
the acquisition module is used for acquiring a pipeline inspection video to be detected;
the detection module is used for inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection and outputting a detection result;
the visualization module is used for calling a preset OpenCV interface if the detection result indicates that pipeline damage exists in the current video frame, and visualizing the five-dimensional vector in the detection result into a detection frame;
and the output module is used for combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type.
Optionally, in a first implementation manner of the second aspect of the present invention, the pipe damage detection apparatus further includes:
the system comprises a marking unit, a storage unit and a processing unit, wherein the marking unit is used for acquiring a plurality of pipeline inspection video samples and marking the pipeline inspection video samples with damage information frame by frame to obtain damaged positive sample images and damaged negative sample images;
the characteristic extraction unit is used for inputting the positive sample image and the negative sample image into a preset target detection network for characteristic extraction to obtain a sample characteristic diagram;
the network optimization unit is used for calling a preset Autofusion algorithm to search evaluation indexes of a feature extraction layer connecting part of the target detection network according to the sample feature map, and taking the target detection network corresponding to the combination with the highest evaluation index as an optimal target detection network;
and the model integration unit is used for calling a preset Stacking integration algorithm to integrate the optimal target detection network to obtain a pipeline damage detection model.
Optionally, in a second implementation manner of the second aspect of the present invention, the feature extraction unit is specifically configured to:
inputting the positive sample image and the negative sample image into a preset input layer for data enhancement to obtain an enhanced sample picture;
scaling and cutting the size of the enhanced sample picture to obtain a standard sample picture;
inputting the standard sample picture into a preset CSP network for feature extraction to obtain sample feature information;
and inputting the sample characteristic information into a preset neutral network for characteristic fusion to obtain a sample characteristic diagram.
Optionally, in a third implementation manner of the second aspect of the present invention, the network optimization unit is specifically configured to:
calling a preset Autofusion algorithm, and carrying out unary operation and maintenance operation on the feature extraction layer connecting part of the target detection network to obtain an unary operation value;
inputting the unary operation value into a preset operation layer to perform amplitude function operation to obtain an operand;
combining the unary operation value and the operand to obtain a combination of evaluation indexes;
and taking the target detection network corresponding to the combination with the highest evaluation index as the optimal target detection network.
Optionally, in a fourth implementation manner of the second aspect of the present invention, the model integration unit is specifically configured to:
calling a preset Stacking integration algorithm, inputting the sample feature graph into the optimal target detection network for integration operation, and obtaining a first-layer meta-feature;
averaging the first layer meta-feature and inputting the average into the optimal target detection network for integrated operation to obtain a second layer meta-feature;
and according to the second layer meta-feature, performing parameter adjustment on the optimal target detection network until the optimal target detection network converges to obtain a pipeline damage detection model.
Optionally, in a fifth implementation manner of the second aspect of the present invention, the detection module is specifically configured to:
inputting the pipeline inspection video into a CSP network in a preset pipeline damage detection model to perform feature extraction frame by frame to obtain feature information;
inputting the characteristic information into a hack network in a preset pipeline damage detection model for characteristic fusion to obtain a characteristic diagram;
and analyzing the category information and the position information of the characteristic diagram, and outputting a detection result.
Optionally, in a sixth implementation manner of the second aspect of the present invention, the pipe damage detection apparatus further includes:
the storage module is used for playing the pipeline inspection marking video and judging whether pipeline damage exists in the current video frame; if so, screenshot is carried out on the current video frame to obtain a pipeline damage picture, and pipeline damage information in the pipeline damage picture is extracted; and storing the pipeline damage picture, the pipeline damage information and the current video playing time point in a correlated manner, and outputting a CSV format file containing the pipeline damage information.
A third aspect of the present invention provides a pipe damage detection apparatus, including: a memory and at least one processor, the memory having instructions stored therein; the at least one processor invokes the instructions in the memory to cause the pipe damage detection apparatus to perform the pipe damage detection method described above.
A fourth aspect of the present invention provides a computer-readable storage medium having stored therein instructions, which, when run on a computer, cause the computer to execute the above-mentioned pipe damage detection method.
In the technical scheme provided by the invention, in view of the fact that the existing method for detecting the pipeline by means of artificial naked eyes has large workload and is easy to misjudge or miss detection, a machine learning mode is introduced to generate a model for automatically detecting the pipeline image, the pipeline video to be detected is input into the model for frame-by-frame detection, the model can realize rapid detection of the damage information on the image, the damage position and the damage type are directly calibrated, then the detection result is visualized through an OpenCV interface, the detected video is stored, and a user can rapidly know whether the damage, the damage type and the specific damage position only by watching the calibrated video. The pipeline damage detection model is constructed aiming at pipeline damage detection, has better applicability to pipeline damage detection tasks, and can greatly improve the efficiency of pipeline damage detection.
Drawings
FIG. 1 is a schematic diagram of a first embodiment of a pipeline damage detection method according to an embodiment of the invention;
FIG. 2 is a diagram of a second embodiment of a pipeline damage detection method according to an embodiment of the invention;
FIG. 3 is a schematic diagram of a third embodiment of a pipeline damage detection method according to an embodiment of the invention;
FIG. 4 is a schematic diagram of a fourth embodiment of a pipeline damage detection method according to the embodiment of the invention;
FIG. 5 is a schematic diagram of an embodiment of a pipeline damage detection device in an embodiment of the invention;
fig. 6 is a schematic diagram of an embodiment of a pipeline damage detection device in the embodiment of the invention.
Detailed Description
The embodiment of the invention provides a pipeline damage detection method, a pipeline damage detection device, pipeline damage detection equipment and a storage medium. The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that the embodiments described herein may be practiced otherwise than as specifically illustrated or described herein. Furthermore, the terms "comprises," "comprising," or "having," and any variations thereof, are intended to cover non-exclusive inclusions, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
For convenience of understanding, a specific flow of an embodiment of the present invention is described below, and referring to fig. 1, a first embodiment of a pipeline damage detection method in an embodiment of the present invention includes:
101. acquiring a pipeline inspection video to be detected;
it is to be understood that the execution subject of the present invention may be a pipeline damage detection apparatus, and may also be a terminal or a server, which is not limited herein. The embodiment of the present invention is described by taking a server as an execution subject.
In this embodiment, the pipeline through camera or other equipment shooting patrols and examines the video, and the video that will shoot obtains patrols and examines the video as the pipeline that detects.
102. Inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection, and outputting a detection result;
in this embodiment, the pipeline damage detection model detects a network structure from N (N >1) targets, for example: the model is constructed by a mainstream deep learning framework Pythrch, and is characterized in that the YooloV 5 network structure, the YooloV 5 are composed of a CSP network, a hack network and a damage information analysis layer. And the detection result comprises that the current frame has no damage, and when the current frame has damage, the position information and the damage type information corresponding to the damage point are used for generating a five-dimensional vector and outputting the five-dimensional vector as the detection result.
In this embodiment, the pipeline inspection video is input into a preset pipeline damage detection model to be detected frame by frame, so as to obtain a detection result, where the detection result includes that there is no damage in the current frame, and when there is damage, position information and damage type information corresponding to the damage point are generated into a five-dimensional vector and output as a detection result.
Optionally, in an embodiment, the inputting the pipeline inspection video into a preset pipeline damage detection model for performing frame-by-frame detection, and outputting a detection result includes:
inputting the pipeline inspection video into a CSP network in a preset pipeline damage detection model to perform feature extraction frame by frame to obtain feature information;
in this embodiment, the CSP network divides the original input into two branches, performs convolution operation to reduce the number of channels by half, then performs a Bottlenneck x N operation on the first branch, and then splices the first branch and the second branch in a tensor manner, so that the input and the output of the CSP network are the same size, and the CSP network can make the model extract more features.
Inputting the characteristic information into a hack network in a preset pipeline damage detection model for characteristic fusion to obtain a characteristic diagram;
in this embodiment, the tack network mainly functions to perform feature fusion on feature information extracted from the CSP network, and performs transfer fusion on feature information of a high layer by an upsampling method by using a common convolution operation to obtain a feature map for prediction, thereby enhancing the capability of network feature fusion.
And analyzing the category information and the position information of the characteristic diagram, and outputting a detection result.
In the embodiment, the pipeline inspection video is input into a CSP network in a preset pipeline damage detection model to perform feature extraction frame by frame to obtain feature information, the feature information is input into a hack network in the preset pipeline damage detection model to perform feature fusion to obtain a feature map, category information and position information of the feature map are analyzed, and a detection result is output. In this embodiment, if there is damage information in the detected video frame, the detection result is that there is damage, and the coordinates of the damage point and the damage type information are generated into a five-dimensional vector and output.
103. If the detection result is that pipeline damage exists in the current video frame, calling a preset OpenCV interface, and visualizing the five-dimensional vector in the detection result into a detection frame;
in this embodiment, the five-dimensional vector in the detection result is (c, x, y, w, h), where c is the type of the detection frame, x is the abscissa, y is the ordinate, w is the width, and h is the height, and the position of the damage in the picture and the type of the damage are marked according to the five-dimensional vector (c, x, y, w, h).
In this embodiment, if the detection result indicates that pipeline damage exists in the current video frame, a preset OpenCV interface is called, and a five-dimensional vector in the detection result is visualized as a detection frame. OpenCV is a BSD license (open source) based distributed cross-platform computer vision and machine learning software library that can run on Linux, Windows, Android, and Mac OS operating systems. OpenCV is light and efficient, provides interfaces of languages such as Python, Ruby, MATLAB and the like, and realizes a plurality of general algorithms in the aspects of image processing and computer vision.
104. And combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type.
In this embodiment, will detect the frame with the pipeline is patrolled and examined corresponding video frame in the video and is combined together, obtains to mark the pipeline that has pipeline damage position and damage kind and patrols and examines the mark video. And combining the detection frame with the corresponding video frame in the original video to generate a new video through an OpenCV interface to obtain the marked video containing the marking information. The detection result and the original video are combined, so that a user can effectively classify and store a large amount of video data, damaged pipeline information is filed, and the detection result and the original video are convenient to search and compare.
Optionally, in an embodiment, after the detecting frame is combined with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type, the method further includes:
playing the pipeline inspection marking video, and judging whether pipeline damage exists in a current video frame;
if so, screenshot is carried out on the current video frame to obtain a pipeline damage picture, and pipeline damage information in the pipeline damage picture is extracted;
and storing the pipeline damage picture, the pipeline damage information and the current video playing time point in a correlated manner, and outputting a CSV format file containing the pipeline damage information.
In this embodiment, a video frame with damage information in a marked video is subjected to screenshot, the damage information in the screenshot is extracted, the screenshot is stored, and the damage information corresponding to the screenshot is saved by calling a Panda tool to obtain a CSV file corresponding to the damage information.
In view of the fact that the pipeline is detected by means of the existing artificial naked eyes, the method is large in workload and easy to misjudge or miss detection, a machine learning mode is introduced to generate a model which can be used for automatically detecting pipeline images, pipeline videos to be detected are input into the model to be detected frame by frame to be detected, the model can achieve rapid detection of damage information on the images, damage positions and damage types can be directly calibrated, then detection results are visualized through an OpenCV interface, the detected videos are stored, and a user can rapidly know whether damage, damage types and specific damage positions exist or not only by watching the calibrated videos. The pipeline damage detection method and the pipeline damage detection system construct the pipeline damage detection model aiming at the specific task of pipeline damage detection, the model has better applicability to the pipeline damage detection task, and the pipeline damage detection efficiency can be greatly improved.
Referring to fig. 2, a second embodiment of the pipeline damage detection method according to the embodiment of the present invention includes:
201. acquiring a plurality of pipeline inspection video samples, and marking the pipeline inspection video samples with damage information frame by frame to obtain damaged positive sample images and non-damaged negative sample images;
in the embodiment, the preset labelme is called to inspect the pipeline inspection video frame by frame, when the video frame has damage, the coordinates of the damage points in the image are firstly extracted, the original image is converted into a binary image by the extraction of the damage points in the image, then the coordinates of the connected domain of the damage part are found, and the coordinates corresponding to the connected domain of the damage part in the image are stored as a mat file. And secondly, calling a preset img2json. And fusing the mat file and the json file by adopting a preset imate _ json.
202. Inputting the positive sample image and the negative sample image into a preset target detection network for feature extraction to obtain a sample feature map;
optionally, in an embodiment, the inputting the positive sample image and the negative sample image into a preset target detection network for feature extraction to obtain a sample feature map includes:
inputting the positive sample image and the negative sample image into a preset input layer for data enhancement to obtain an enhanced sample picture;
scaling and cutting the size of the enhanced sample picture to obtain a standard sample picture;
inputting the standard sample picture into a preset CSP network for feature extraction to obtain sample feature information;
and inputting the sample characteristic information into a preset neutral network for characteristic fusion to obtain a sample characteristic diagram.
In this embodiment, for the positive sample image and the negative sample image, besides the classical geometric distortion and the illumination distortion, the CutMix and Mosaic technology is used to perform data enhancement, so as to obtain an enhanced sample image. The target detection network needs to adjust the size of the original image for feature recognition, and the image in the model is scaled to 512 x 512. The CSP network solves the problem of repeated gradient information of network optimization in other large convolutional neural network frameworks (backbones), integrates the change of the gradient into the feature map from beginning to end, separates the feature map of the basic layer, effectively relieves the problem of gradient disappearance, supports feature propagation, encourages the network to reuse the features, and reduces the number of network parameters. The hack network is used to generate a feature pyramid. The feature pyramid can enhance the detection of the model on objects with different scaling scales, so that the same object with different sizes and scales can be identified, the features extracted by the CSP network are fused, and a feature picture is obtained. The detection speed and the detection precision of the target detection network are perfectly coordinated, and the obtained sample characteristic diagram has higher accuracy.
203. According to the sample feature map, calling a preset Autofusion algorithm to search evaluation indexes of a feature extraction layer connecting part of the target detection network, and taking the target detection network corresponding to the combination with the highest evaluation index as an optimal target detection network;
in this embodiment, the AutoFusion algorithm performs spatial search on the feature extraction layer connection part of the target detection network structureThree steps are total, firstly Unary ops operation is carried out to obtain op1、op2，op1、op2Is a unitary operation value, and then the unitary operation value is subjected to amplitude function operation to obtain mu1、μ2，μ1、μ2As an operand, and finally as a combined operation of these two steps Δ w ═ λ × (μ ═ b)1(op1),u2(op2) And the corresponding target detection network is the optimal target detection network when the value of the delta w is the highest.
204. A preset Stacking integration algorithm is called to integrate the optimal target detection network to obtain a pipeline damage detection model;
in the embodiment, the positive sample image and the negative sample image are input into a preset target detection network for feature extraction to obtain a sample feature map; according to the sample feature map, calling a preset Autofusion algorithm to search evaluation indexes of a feature extraction layer connecting part of the target detection network, and taking the target detection network corresponding to the combination with the highest evaluation index as an optimal target detection network; and calling a preset Stacking integration algorithm to integrate the optimal target detection network to obtain a pipeline damage detection model.
In this embodiment, the conventional calculation learning method includes three major parts, i.e., feature extraction, model design, and parameter tuning, and the automatic machine learning AutoFusion algorithm automatically completes the whole machine learning process, and only data needs to be input to obtain output. In this embodiment, the neural network structure search technology is to perform spatial search on the connection point of the feature extraction layers of the target detection network through an AutoFusion algorithm, search for an optimal evaluation index combination, and use the target detection network with the highest evaluation index as the optimal target detection network.
In this embodiment, the AutoFusion algorithm searches for a local maximum value, suppresses non-maximum value elements, finds a bounding box with a higher confidence coefficient according to coordinate information of the score matrix and the region, first performs descending sorting on the confidence coefficients of all detection boxes, selects a detection box with the highest confidence coefficient, determines whether the detection box with the highest confidence coefficient is correct, calculates an IOU value between the detection box with the highest confidence coefficient and other detection boxes if the detection box with the highest confidence coefficient is determined to be correct, removes the detection box with the highest confidence coefficient and other detection boxes according to the IOU value, removes the detection box with the highest overlap degree when the IOU value is greater than threshold, removes the detection box with the high overlap degree, continues to perform the sorting on the confidence coefficients until redundant detection boxes are removed, and finds the best position for detecting the damage.
In this embodiment, the Stacking integration algorithm trains a multi-layer learner structure, the first layer obtains a prediction result of the first layer by using N YoloV5 models, the prediction result of the first layer is merged into a YoloV5 model after a new feature input image is input into the learner, and a final prediction result of a pipeline damage model is obtained through output of a second prediction process.
205. Acquiring a pipeline inspection video to be detected;
206. inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection, and outputting a detection result;
207. if the detection result is that pipeline damage exists in the current video frame, calling a preset OpenCV interface, and visualizing the five-dimensional vector in the detection result into a detection frame;
208. and combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type.
In the embodiment of the invention, an Autofusion algorithm is adopted to optimize a target network structure to obtain an optimal target network structure, and a Stacking integration algorithm is adopted to integrate the optimal target network structure to obtain a final pipeline damage detection model. And optimizing the network structure by adopting an Autofusion optimization algorithm, so that the obtained pipeline damage detection model is more suitable for the specific task of pipeline damage detection.
Referring to fig. 3, a third embodiment of the pipeline damage detection method according to the embodiment of the present invention includes:
301. acquiring a plurality of pipeline inspection video samples, and marking the pipeline inspection video samples with damage information frame by frame to obtain damaged positive sample images and non-damaged negative sample images;
302. inputting the positive sample image and the negative sample image into a preset target detection network for feature extraction to obtain a sample feature map;
303. calling a preset Autofusion algorithm, and carrying out unary operation and maintenance operation on the feature extraction layer connecting part of the target detection network to obtain an unary operation value;
304. inputting the unary operation value into a preset operation layer to perform amplitude function operation to obtain an operand;
305. combining the unary operation value and the operand to obtain a combination of evaluation indexes;
306. taking the target detection network corresponding to the combination with the highest evaluation index as the optimal target detection network;
in this embodiment, the AutoFusion algorithm is adopted to optimize the target detection network structure, and three steps are total, and firstly, a unitary operation and maintenance operation is performed to obtain the op1、op2，op1、op2Is a unitary operation value, and then the unitary operation value is subjected to amplitude function operation to obtain mu1、μ2，μ1、μ2As an operand, the combination of evaluation indicators obtained by integrating the two steps and the integrated operation Δ w ═ λ × (μ) of the two steps are finally performed1(op1),u2(op2) And the corresponding target detection network is the optimal target detection network when the value of the delta w is the highest. The neural framework with good performance is generated by selecting the target detection network with the highest evaluation index from the search space, and the target detection network corresponding to the highest evaluation index in the evaluation index combination is used as the optimal target detection network, so that the detection speed of the optimal target detection network is improved, and the damage detection is more accurate.
307. Acquiring a pipeline inspection video to be detected;
308. inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection, and outputting a detection result;
309. if the detection result is that pipeline damage exists in the current video frame, calling a preset OpenCV interface, and visualizing the five-dimensional vector in the detection result into a detection frame;
310. and combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type.
In the embodiment of the invention, the Autofusion algorithm is adopted to optimize the target detection network structure, so that the optimization can be carried out automatically without the assistance of the outside, and a near-optimal network architecture and model for pipeline damage detection can still be obtained.
Referring to fig. 4, a fourth embodiment of the pipeline damage detection method according to the embodiment of the present invention includes:
401. acquiring a plurality of pipeline inspection video samples, and marking the pipeline inspection video samples with damage information frame by frame to obtain damaged positive sample images and non-damaged negative sample images;
402. inputting the positive sample image and the negative sample image into a preset target detection network for feature extraction to obtain a sample feature map;
403. according to the sample feature map, calling a preset Autofusion algorithm to search evaluation indexes of a feature extraction layer connecting part of the target detection network, and taking the target detection network corresponding to the combination with the highest evaluation index as an optimal target detection network;
404. calling a preset Stacking integration algorithm, inputting the sample feature graph into the optimal target detection network for integration operation, and obtaining a first-layer meta-feature;
405. averaging the first layer meta-feature and inputting the average into the optimal target detection network for integrated operation to obtain a second layer meta-feature;
406. according to the second layer meta-feature, parameter adjustment is carried out on the optimal target detection network until the optimal target detection network is converged to obtain a pipeline damage detection model;
in this embodiment, a Stacking method is used to train a meta-model, and the meta-model generates the final output according to the output result returned by the weak learner at the lower layer. In the Stacking method, N YoloV5 models are used in a first layer to obtain a prediction result of the first layer, the prediction result of the first layer is merged into a new YoloV5 model after feature input image input learning, the output of a second prediction process is used as a final detection result of a system, and the N YoloV5 models are integrated into a pipeline detection model, so that the advantages of a plurality of network structures can be integrated, and the detection speed of the integrated pipeline damage model is higher and the accuracy is higher. Performing cross validation by using YoloV5 as a basic model, wherein the cross validation comprises two processes, namely training the model based on a characteristic diagram; and secondly, predicting the feature map based on a model generated by feature map training. And obtaining a predicted value of the current feature map after the cross validation is completed, and performing the two steps twice to finally generate a detection result. And adjusting parameters of the optimal target detection network by adopting a binary cross entropy until the optimal target detection network is converged to obtain a pipeline damage detection model.
407. Acquiring a pipeline inspection video to be detected;
408. inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection, and outputting a detection result;
409. if the detection result is that pipeline damage exists in the current video frame, calling a preset OpenCV interface, and visualizing the five-dimensional vector in the detection result into a detection frame;
410. and combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type.
In the embodiment of the invention, the Stacking integration method combines the meta-models by training the meta-models and outputs a final prediction result according to the prediction results of different weak models, so that the frames can integrate the advantages of various frames, the method has better applicability to pipeline damage detection tasks, and the pipeline damage detection models integrated by the Stacking method can identify the position information and the category information of damage more accurately.
In the above description of the method for detecting pipeline damage in the embodiment of the present invention, referring to fig. 5, a pipeline damage detection apparatus in the embodiment of the present invention is described below, where an embodiment of the pipeline damage detection apparatus in the embodiment of the present invention includes:
the acquiring module 501 is used for acquiring a pipeline inspection video to be detected;
the detection module 502 is used for inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection and outputting a detection result;
a visualization module 503, configured to invoke a preset OpenCV interface if the detection result indicates that a pipeline damage exists in the current video frame, and visualize the five-dimensional vector in the detection result as a detection frame;
and the output module 504 is used for combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type.
Optionally, in an embodiment, the pipe damage detecting device further includes:
the system comprises a marking unit, a storage unit and a processing unit, wherein the marking unit is used for acquiring a plurality of pipeline inspection video samples and marking the pipeline inspection video samples with damage information frame by frame to obtain damaged positive sample images and damaged negative sample images;
the characteristic extraction unit is used for inputting the positive sample image and the negative sample image into a preset target detection network for characteristic extraction to obtain a sample characteristic diagram;
the network optimization unit is used for calling a preset Autofusion algorithm to search evaluation indexes of a feature extraction layer connecting part of the target detection network according to the sample feature map, and taking the target detection network corresponding to the combination with the highest evaluation index as an optimal target detection network;
and the model integration unit is used for calling a preset Stacking integration algorithm to integrate the optimal target detection network to obtain a pipeline damage detection model.
Optionally, in an embodiment, the feature extraction unit is specifically configured to:
inputting the positive sample image and the negative sample image into a preset input layer for data enhancement to obtain an enhanced sample picture;
scaling and cutting the size of the enhanced sample picture to obtain a standard sample picture;
inputting the standard sample picture into a preset CSP network for feature extraction to obtain sample feature information;
and inputting the sample characteristic information into a preset neutral network for characteristic fusion to obtain a sample characteristic diagram.
Optionally, in an embodiment, the network optimization unit is specifically configured to:
calling a preset Autofusion algorithm, and carrying out unary operation and maintenance operation on the feature extraction layer connecting part of the target detection network to obtain an unary operation value;
inputting the unary operation value into a preset operation layer to perform amplitude function operation to obtain an operand;
combining the unary operation value and the operand to obtain a combination of evaluation indexes;
and taking the target detection network corresponding to the combination with the highest evaluation index as the optimal target detection network.
Optionally, in an embodiment, the model integration unit is specifically configured to:
calling a preset Stacking integration algorithm, inputting the sample feature graph into the optimal target detection network for integration operation, and obtaining a first-layer meta-feature;
averaging the first layer meta-feature and inputting the average into the optimal target detection network for integrated operation to obtain a second layer meta-feature;
and according to the second layer meta-feature, performing parameter adjustment on the optimal target detection network until the optimal target detection network converges to obtain a pipeline damage detection model.
Optionally, in an embodiment, the detecting module 502 is specifically configured to:
inputting the pipeline inspection video into a CSP network in a preset pipeline damage detection model to perform feature extraction frame by frame to obtain feature information;
inputting the characteristic information into a hack network in a preset pipeline damage detection model for characteristic fusion to obtain a characteristic diagram;
and analyzing the category information and the position information of the characteristic diagram, and outputting a detection result.
Optionally, in an embodiment, the pipe damage detecting device further includes:
the storage module is used for playing the pipeline inspection marking video and judging whether pipeline damage exists in the current video frame; if so, screenshot is carried out on the current video frame to obtain a pipeline damage picture, and pipeline damage information in the pipeline damage picture is extracted; and storing the pipeline damage picture, the pipeline damage information and the current video playing time point in a correlated manner, and outputting a CSV format file containing the pipeline damage information.
In view of the fact that the pipeline is detected by means of the existing artificial naked eyes, the method is large in workload and easy to misjudge or miss detection, a machine learning mode is introduced to generate a model which can be used for automatically detecting pipeline images, pipeline videos to be detected are input into the model to be detected frame by frame to be detected, the model can achieve rapid detection of damage information on the images, damage positions and damage types can be directly calibrated, then detection results are visualized through an OpenCV interface, the detected videos are stored, and a user can rapidly know whether damage, damage types and specific damage positions exist or not only by watching the calibrated videos. The pipeline damage detection method and the pipeline damage detection system construct the pipeline damage detection model aiming at the specific task of pipeline damage detection, the model has better applicability to the pipeline damage detection task, and the pipeline damage detection efficiency can be greatly improved.
Fig. 5 describes the pipe damage detection apparatus in the embodiment of the present invention in detail from the perspective of the modular functional entity, and describes the pipe damage detection apparatus in the embodiment of the present invention in detail from the perspective of hardware processing.
Fig. 6 is a schematic structural diagram of a pipeline damage detection apparatus 600 according to an embodiment of the present invention, where the pipeline damage detection apparatus 600 may generate relatively large differences due to different configurations or performances, and may include one or more processors (CPUs) 610 (e.g., one or more processors) and a memory 620, and one or more storage media 630 (e.g., one or more mass storage devices) storing applications 633 or data 632. Memory 620 and storage medium 630 may be, among other things, transient or persistent storage. The program stored in the storage medium 630 may include one or more modules (not shown), each of which may include a series of instruction operations for the pipe damage detection apparatus 600. Still further, the processor 610 may be configured to communicate with the storage medium 630 to execute a series of instruction operations in the storage medium 630 on the pipeline damage detection apparatus 600.
The pipeline damage detection apparatus 600 may also include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input-output interfaces 660, and/or one or more operating systems 631, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, and so forth. Those skilled in the art will appreciate that the configuration of the pipeline damage detection apparatus shown in FIG. 6 does not constitute a limitation of the pipeline damage detection apparatus, and may include more or fewer components than shown, or some components in combination, or a different arrangement of components. The invention further provides a pipeline damage detection device, which includes a memory and a processor, where the memory stores computer readable instructions, and the computer readable instructions, when executed by the processor, cause the processor to execute the steps of the pipeline damage detection method in the above embodiments.
The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium, and which may also be a volatile computer-readable storage medium, having stored therein instructions, which, when run on a computer, cause the computer to perform the steps of the pipe damage detection method.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A pipeline damage detection method is characterized by comprising the following steps:
acquiring a pipeline inspection video to be detected;
inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection, and outputting a detection result;
if the detection result is that pipeline damage exists in the current video frame, calling a preset OpenCV interface, and visualizing the five-dimensional vector in the detection result into a detection frame;
and combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type.
2. The pipeline damage detection method according to claim 1, further comprising, before the acquiring the video of the sewer pipeline to be detected:
acquiring a plurality of pipeline inspection video samples, and marking the pipeline inspection video samples with damage information frame by frame to obtain damaged positive sample images and non-damaged negative sample images;
inputting the positive sample image and the negative sample image into a preset target detection network for feature extraction to obtain a sample feature map;
according to the sample feature map, calling a preset Autofusion algorithm to search evaluation indexes of a feature extraction layer connecting part of the target detection network, and taking the target detection network corresponding to the combination with the highest evaluation index as an optimal target detection network;
and calling a preset Stacking integration algorithm to integrate the optimal target detection network to obtain a pipeline damage detection model.
3. The pipeline damage detection method according to claim 2, wherein the inputting the positive sample image and the negative sample image into a preset target detection network for feature extraction to obtain a sample feature map comprises:
inputting the positive sample image and the negative sample image into a preset input layer for data enhancement to obtain an enhanced sample picture;
scaling and cutting the size of the enhanced sample picture to obtain a standard sample picture;
inputting the standard sample picture into a preset CSP network for feature extraction to obtain sample feature information;
and inputting the sample characteristic information into a preset neutral network for characteristic fusion to obtain a sample characteristic diagram.
4. The pipeline damage detection method according to claim 2, wherein the step of calling a preset Autofusion algorithm to search for an evaluation index of a feature extraction layer connection part of the target detection network according to the sample feature map, and using the target detection network corresponding to the combination with the highest evaluation index as the optimal target detection network comprises the steps of:
calling a preset Autofusion algorithm, and carrying out unary operation and maintenance operation on the feature extraction layer connecting part of the target detection network to obtain an unary operation value;
inputting the unary operation value into a preset operation layer to perform amplitude function operation to obtain an operand;
combining the unary operation value and the operand to obtain a combination of evaluation indexes;
and taking the target detection network corresponding to the combination with the highest evaluation index as the optimal target detection network.
5. The pipeline damage detection method according to claim 2, wherein the step of calling a preset Stacking integration algorithm to integrate the optimal target detection network to obtain a pipeline damage detection model comprises the steps of:
calling a preset Stacking integration algorithm, inputting the sample feature graph into the optimal target detection network for integration operation, and obtaining a first-layer meta-feature;
averaging the first layer meta-feature and inputting the average into the optimal target detection network for integrated operation to obtain a second layer meta-feature;
and according to the second layer meta-feature, performing parameter adjustment on the optimal target detection network until the optimal target detection network converges to obtain a pipeline damage detection model.
6. The pipeline damage detection method according to claim 3, wherein the pipeline inspection video is input into a preset pipeline damage detection model for frame-by-frame detection, and outputting a detection result comprises:
inputting the pipeline inspection video into a CSP network in a preset pipeline damage detection model to perform feature extraction frame by frame to obtain feature information;
inputting the characteristic information into a hack network in a preset pipeline damage detection model for characteristic fusion to obtain a characteristic diagram;
and analyzing the category information and the position information of the characteristic diagram, and outputting a detection result.
7. The pipeline damage detection method according to any one of claims 1 to 6, wherein after the detection frame is combined with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type, the method further comprises:
playing the pipeline inspection marking video, and judging whether pipeline damage exists in a current video frame;
if so, screenshot is carried out on the current video frame to obtain a pipeline damage picture, and pipeline damage information in the pipeline damage picture is extracted;
and storing the pipeline damage picture, the pipeline damage information and the current video playing time point in a correlated manner, and outputting a CSV format file containing the pipeline damage information.
8. A pipeline damage detection device, characterized in that pipeline damage detection device includes:
the acquisition module is used for acquiring a pipeline inspection video to be detected;
the detection module is used for inputting the pipeline inspection video into a preset pipeline damage detection model for frame-by-frame detection and outputting a detection result;
the visualization module is used for calling a preset OpenCV interface if the detection result indicates that pipeline damage exists in the current video frame, and visualizing the five-dimensional vector in the detection result into a detection frame;
and the output module is used for combining the detection frame with the corresponding video frame in the pipeline inspection video to obtain the pipeline inspection marking video marked with the pipeline damage position and the damage type.
9. A pipeline damage detection device, comprising: a memory and at least one processor, the memory having instructions stored therein;
the at least one processor invokes the instructions in the memory to cause the pipe damage detection apparatus to perform the pipe damage detection method of any of claims 1-7.
10. A computer readable storage medium having instructions stored thereon, wherein the instructions, when executed by a processor, implement the pipe damage detection method of any one of claims 1-7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011385962.5A CN112446870B (en) | 2020-12-02 | 2020-12-02 | Pipeline damage detection method, device, equipment and storage medium |
| PCT/CN2021/083573 WO2022116433A1 (en) | 2020-12-02 | 2021-03-29 | Pipeline damage detection method and apparatus, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011385962.5A CN112446870B (en) | 2020-12-02 | 2020-12-02 | Pipeline damage detection method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112446870A true CN112446870A (en) | 2021-03-05 |
| CN112446870B CN112446870B (en) | 2024-07-09 |
Family
ID=74740456
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011385962.5A Active CN112446870B (en) | 2020-12-02 | 2020-12-02 | Pipeline damage detection method, device, equipment and storage medium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112446870B (en) |
| WO (1) | WO2022116433A1 (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112966678A (en) * | 2021-03-11 | 2021-06-15 | 南昌航空大学 | Text detection method and system |
| CN113362285A (en) * | 2021-05-21 | 2021-09-07 | 同济大学 | Steel rail surface damage fine-grained image classification and detection method |
| CN114429639A (en) * | 2022-01-27 | 2022-05-03 | 广联达科技股份有限公司 | Pipeline marking identification method and device |
| WO2022116433A1 (en) * | 2020-12-02 | 2022-06-09 | 平安科技（深圳）有限公司 | Pipeline damage detection method and apparatus, device and storage medium |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116123988B (en) * | 2023-04-14 | 2023-07-11 | 国家石油天然气管网集团有限公司 | Pipeline geometric deformation detection method and system based on data processing |
| CN116382224B (en) * | 2023-06-05 | 2023-08-04 | 云印技术(深圳)有限公司 | Packaging equipment monitoring method and system based on data analysis |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110287798A (en) * | 2019-05-27 | 2019-09-27 | 魏运 | Vector network pedestrian detection method based on characteristic module and context fusion |
| WO2020140371A1 (en) * | 2019-01-04 | 2020-07-09 | 平安科技（深圳）有限公司 | Deep learning-based vehicle damage identification method and related device |
| CN111667011A (en) * | 2020-06-08 | 2020-09-15 | 平安科技（深圳）有限公司 | Damage detection model training method, damage detection model training device, damage detection method, damage detection device, damage detection equipment and damage detection medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110264444B (en) * | 2019-05-27 | 2020-07-17 | 阿里巴巴集团控股有限公司 | Damage detection method and device based on weak segmentation |
| CN110390666B (en) * | 2019-06-14 | 2023-06-27 | 平安科技（深圳）有限公司 | Road damage detection method, device, computer equipment and storage medium |
| CN110490079B (en) * | 2019-07-19 | 2022-06-03 | 万翼科技有限公司 | Patrol data processing method and device, computer equipment and storage medium |
| CN112446870B (en) * | 2020-12-02 | 2024-07-09 | 平安科技（深圳）有限公司 | Pipeline damage detection method, device, equipment and storage medium |
-
2020
- 2020-12-02 CN CN202011385962.5A patent/CN112446870B/en active Active
-
2021
- 2021-03-29 WO PCT/CN2021/083573 patent/WO2022116433A1/en active Application Filing
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020140371A1 (en) * | 2019-01-04 | 2020-07-09 | 平安科技（深圳）有限公司 | Deep learning-based vehicle damage identification method and related device |
| CN110287798A (en) * | 2019-05-27 | 2019-09-27 | 魏运 | Vector network pedestrian detection method based on characteristic module and context fusion |
| CN111667011A (en) * | 2020-06-08 | 2020-09-15 | 平安科技（深圳）有限公司 | Damage detection model training method, damage detection model training device, damage detection method, damage detection device, damage detection equipment and damage detection medium |
Non-Patent Citations (1)
| Title |
|---|
| 张宇维: "城市排水管内窥图像分类与病害智能检测研究", 中国优秀硕士学位论文全文数据库工程科技Ⅱ辑, no. 2, 15 February 2020 (2020-02-15), pages 038 - 1379 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022116433A1 (en) * | 2020-12-02 | 2022-06-09 | 平安科技（深圳）有限公司 | Pipeline damage detection method and apparatus, device and storage medium |
| CN112966678A (en) * | 2021-03-11 | 2021-06-15 | 南昌航空大学 | Text detection method and system |
| CN113362285A (en) * | 2021-05-21 | 2021-09-07 | 同济大学 | Steel rail surface damage fine-grained image classification and detection method |
| CN113362285B (en) * | 2021-05-21 | 2023-02-07 | 同济大学 | Steel rail surface damage fine-grained image classification and detection method |
| CN114429639A (en) * | 2022-01-27 | 2022-05-03 | 广联达科技股份有限公司 | Pipeline marking identification method and device |
| CN114429639B (en) * | 2022-01-27 | 2024-05-03 | 广联达科技股份有限公司 | Pipeline annotation recognition method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2022116433A1 (en) | 2022-06-09 |
| CN112446870B (en) | 2024-07-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112446870A (en) | Pipeline damage detection method, device, equipment and storage medium | |
| CN111325713B (en) | Neural network-based wood defect detection method, system and storage medium | |
| Hoang et al. | Metaheuristic optimized edge detection for recognition of concrete wall cracks: a comparative study on the performances of roberts, prewitt, canny, and sobel algorithms | |
| CN107230203B (en) | Casting defect identification method based on human eye visual attention mechanism | |
| CN111626176B (en) | Remote sensing target rapid detection method and system based on dynamic attention mechanism | |
| Vishwakarma et al. | Cnn model & tuning for global road damage detection | |
| KR20230124713A (en) | Fault detection methods, devices and systems | |
| CN110992349A (en) | Underground pipeline abnormity automatic positioning and identification method based on deep learning | |
| CN110264444B (en) | Damage detection method and device based on weak segmentation | |
| CN111738045B (en) | Image detection method and device, electronic equipment and storage medium | |
| US20220178841A1 (en) | Apparatus for optimizing inspection of exterior of target object and method thereof | |
| CN112215217B (en) | Digital image recognition method and device for simulating doctor to read film | |
| CN112949453B (en) | Training method of smoke and fire detection model, smoke and fire detection method and equipment | |
| Akhyar et al. | High efficient single-stage steel surface defect detection | |
| CN111815573B (en) | Coupling outer wall detection method and system based on deep learning | |
| CN113469971A (en) | Image matching method, image matching detection device and storage medium | |
| CN115578326A (en) | Road disease identification method, system, equipment and storage medium | |
| JP2011044016A (en) | Three dimensional cell image analyzer | |
| Nassar et al. | Assessing data quality of annotations with Krippendorff alpha for applications in computer vision | |
| JPWO2019215746A5 (en) | ||
| CN110472638A (en) | A kind of object detection method, device and equipment, storage medium | |
| CN114549548A (en) | Glass image segmentation method based on polarization clue | |
| De Roovere et al. | CenDerNet: Center and Curvature Representations for Render-and-Compare 6D Pose Estimation | |
| CN114782877A (en) | Valve fault detection method, device, equipment and storage medium | |
| JP2007139421A (en) | Morphological classification device and method, and variation region extraction device and 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 | ||
| GR01 | Patent grant |