CN111986188A - Capsule robot drainage pipe network defect identification method based on Resnet and LSTM - Google Patents
Capsule robot drainage pipe network defect identification method based on Resnet and LSTM Download PDFInfo
- Publication number
- CN111986188A CN111986188A CN202010874511.1A CN202010874511A CN111986188A CN 111986188 A CN111986188 A CN 111986188A CN 202010874511 A CN202010874511 A CN 202010874511A CN 111986188 A CN111986188 A CN 111986188A
- Authority
- CN
- China
- Prior art keywords
- lstm
- network
- long
- inputting
- capsule robot
- 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
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
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Abstract
The invention discloses a method for identifying defects of a drainage pipe network of a capsule robot based on Resnet and LSTM, and relates to the technical field of drainage pipe detection; the method comprises the following steps: s1, inputting and matrix converting an image sequence, inputting the marked drainage pipeline defect image sequence collected by the capsule robot into a long-term convolution network LRCN, and performing matrix conversion on the images before inputting the images into the long-term convolution network LRCN; s2, extracting image features, namely, after inputting a drainage pipeline defect image sequence, dividing the drainage pipeline defect image sequence into a plurality of single-frame images, and then inputting the single-frame images into a residual error network Resnet50 to extract the image features; s3, processing the Bi-LSTM; s4, classifying prediction results, predicting the video category of each time step by the BI-directional long-short term memory network BI-LSTM, and integrating the prediction results into final classification; the invention has the beneficial effects that: the time of manual detection can be obviously reduced, and the working efficiency of the capsule robot is improved.
Description
Technical Field
The invention relates to the technical field of drainage pipeline detection, in particular to a Resnet and LSTM capsule robot drainage pipeline network defect identification method.
Background
The drainage system is an important channel for urban rain and sewage discharge and is also a life line for maintaining urban safe operation. However, in the drainage pipeline operation maintenance and management process, more and more drainage pipeline problems are continuously exposed due to the factors that the load flow rate is far beyond the design standard, the pipeline facilities are aged, the subway and other heavy and newly built underground projects are adopted, the underground detection means are insufficient and the like; therefore, the detection and evaluation of the drainage pipeline is important.
However, CCTV, a main technology for detecting drainage pipelines, has a series of disadvantages, such as high cost, low efficiency, and complex operation. For this reason, a capsule robot, a low-cost drainage pipeline inspection apparatus floating on the water surface, has been developed. The capsule robot has the advantages of lower cost, simpler operation, higher efficiency and wider measurement range, and meets the requirements of large-range and high-frequency defect detection of the urban drainage system.
However, the manual interpretation of a large number of capsule robot detection videos is time-consuming and labor-consuming, not only considerable professional knowledge is needed, but also the objectivity and accuracy of defect identification cannot be guaranteed. The traditional automatic drain pipeline defect identification method mostly adopts a traditional machine learning method based on a feature extractor, such as HOG, SIFT and the like, and the methods are usually designed only aiming at specific tasks, so that the generalization capability in practical application is poor. In addition, the image quality problem caused by detection modes such as fisheye distortion, shaking blur and the like is poor, and the feasibility of the capsule robot using the traditional method is far inferior to that of CCTV.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a method for identifying the defects of the drainage pipe network of the capsule robot based on Resnet and LSTM, which can obviously reduce the time, labor and material cost of manual detection and improve the working efficiency of the capsule robot.
The technical scheme adopted by the invention for solving the technical problems is as follows: in a method for identifying defects of a drainage pipe network of a capsule robot based on Resnet and LSTM, the improvement comprising the steps of:
s1, inputting and matrix converting an image sequence, inputting the marked drainage pipeline defect image sequence collected by the capsule robot into a long-term convolution network LRCN, performing matrix conversion on the images before inputting the images into the long-term convolution network LRCN, and converting the images into a three-dimensional matrix of 224 x 3 in a jpg format;
s2, extracting image features, namely, after inputting a drainage pipeline defect image sequence, dividing the drainage pipeline defect image sequence into a plurality of single-frame images, inputting the single-frame images into a residual error network Resnet50 to extract the image features, wherein the process of extracting the image features comprises the following steps:
s21, the input image is first passed through a convolution kernel of 7 × 7, the step size is 2, and the image output is 112 × 64;
s22, passing through a max pooling layer, kernel size 3 × 3, step size 2, image output 56 × 64;
s23, passing through 16 residual blocks, gradually reducing the size of the feature map, increasing the number of channels, and outputting an image of 7 × 2048;
s24, finally, converting the data into 2048-dimensional vectors through an average pooling layer;
s25, three full-connection layers comprising 512, 256 and 64 neurons are arranged behind the average pooling layer and are used for reducing the dimension of the image features into 64-dimensional vectors;
s3, processing the bidirectional long and short term memory network Bi-LSTM, inputting the image characteristics with time correlation into the subsequent bidirectional long and short term memory network Bi-LSTM for processing;
and S4, classifying the prediction results, predicting the video category of each time step by the BI-directional long-short term memory network BI-LSTM, and integrating the prediction results into the final classification.
Further, in step S2, the residual error network Resnet50 has 50 layers in total, and is composed of 49 convolutional layers and 1 fully-connected layer, where the fully-connected layer is used to convert the output of the convolutional layers into a multi-dimensional vector.
Further, the convolutional layer includes conv1, conv2_ x, conv3_ x, conv4_ x and conv5_ x, wherein:
the conv1 consists of a single convolution layer of 7 x 64 for down-sampling the input image size 224 x 224;
the conv2_ x, conv3_ x, conv4_ x and conv5_ x are respectively composed of 3, 4, 6 and 3 residual blocks, and each residual block is respectively composed of different numbers of 1 × 1, 3 × 3 and 1 × 1 convolution layers.
Further, in step S1, each drainage pipeline defect image sequence is composed of 24 consecutive frames, i.e. converted into 24 matrices of 224 × 3.
Further, in step S3, the Bi-directional long-short term memory network Bi-LSTM is composed of two layers of long-short term memory networks LSTM, which are respectively used for learning information of future direction and past direction, and more comprehensively capturing feature information of the sequence vector.
Further, the outputs of the two layers of long and short term memory networks LSTM are combined into serial connection, after the vectors of Bi-LSTM are connected in series through the bidirectional long and short term memory network, the output vectors complete the classification of the defect sequence through the full connection layer and SoftMax.
Further, in step S4, the final classification includes, but is not limited to, pipe breakage, pipe misconnection, corrosion fouling, and pipe deformation.
The invention has the beneficial effects that: the algorithm is applied to the capsule robot, and is optimized according to the characteristic that the capsule robot shoots images; secondly, the capsule robot image sequence is identified as a whole, so that the problem of poor image quality in the prior art is solved; the time, manpower and material resource cost of manual detection can be obviously reduced, and the working efficiency of the capsule robot is improved.
Drawings
FIG. 1 is a schematic flow chart of a method for identifying defects of a drainage pipe network of a capsule robot based on Resnet and LSTM.
FIG. 2 is a detailed schematic diagram of a method for identifying defects of a drainage pipe network of a capsule robot based on Resnet and LSTM.
Fig. 3 is a detailed structure diagram of the residual error network Resnet50 according to the present invention.
Fig. 4 is a flow chart of the structure of Resnet50 according to the present invention.
Fig. 5 is a basic structure diagram of the network elements of the long short term memory network LSTM of the present invention.
FIG. 6 is a schematic diagram of the Bi-LSTM structure of the bidirectional long-short term memory network of the present invention.
Detailed Description
The invention is further illustrated with reference to the following figures and examples.
The conception, the specific structure, and the technical effects produced by the present invention will be clearly and completely described below in conjunction with the embodiments and the accompanying drawings to fully understand the objects, the features, and the effects of the present invention. It is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments, and those skilled in the art can obtain other embodiments without inventive effort based on the embodiments of the present invention, and all embodiments are within the protection scope of the present invention. In addition, all the connection/connection relations referred to in the patent do not mean that the components are directly connected, but mean that a better connection structure can be formed by adding or reducing connection auxiliary components according to specific implementation conditions. All technical characteristics in the invention can be interactively combined on the premise of not conflicting with each other.
Referring to fig. 1, the invention discloses a method for identifying defects of a drainage pipe network of a capsule robot based on Resnet and LSTM, and the method is used for identifying a capsule robot image sequence as a whole. The method can obviously reduce the time, labor and material cost of manual detection and improve the working efficiency of the capsule robot.
Long-term Convolutional Networks (LRCNs) are deep Convolutional Networks that identify and classify sequences of images as a whole. The long-term convolutional network model was originally proposed by Jeff Donahue, and aims to solve the tasks of reference video identification, image description and retrieval, and the challenges of video interpretation. The model consists of a deep visual feature extractor (e.g., CNNs) and a neural network (e.g., RNNs) capable of processing time series data. The invention adopts a residual error network Resnet50 to extract visual characteristics of a drainage pipeline image sequence, adopts a bidirectional long-short term memory network Bi-LSTM to learn and identify time characteristics, and is used for a drainage pipeline defect classification task.
Referring to fig. 1 and 2, the method for identifying the defects of the drainage pipe network of the capsule robot based on the Resnet and the LSTM of the present invention specifically includes the following steps:
s1, inputting and matrix converting an image sequence, inputting the marked drainage pipeline defect image sequence collected by the capsule robot into a long-term convolution network LRCN, performing matrix conversion on the images before inputting the images into the long-term convolution network LRCN, and converting the images into a three-dimensional matrix of 224 x 3 in a jpg format; in this embodiment, each drainage pipeline defect image sequence is composed of 24 consecutive frames, i.e., converted into 24 matrices 224 × 3;
s2, extracting image features, namely, after inputting a drainage pipeline defect image sequence, dividing the drainage pipeline defect image sequence into a plurality of single-frame images, and then inputting the single-frame images into a residual error network Resnet50 to extract the image features; the process of extracting image features is as follows:
s21, the input image is first passed through a convolution kernel of 7 × 7, the step size is 2, and the image output is 112 × 64;
s22, passing through a max pooling layer, kernel size 3 × 3, step size 2, image output 56 × 64;
s23, passing through 16 residual blocks, gradually reducing the size of the feature map, increasing the number of channels, and outputting an image of 7 × 2048;
and S24, finally, converting the data into vectors of 2048 dimensions through an average pooling layer.
And S25, setting three full-connection layers comprising 512, 256 and 64 neurons behind the average pooling layer, and reducing the dimension of the image features into 64-dimensional vectors.
Since an image sequence contains 24 frames, the output of Resnet50 for each sequence takes the form of 24 64-dimensional vectors that are input into the subsequent decoder in chronological order. The flow chart of the structure is shown in figure 4.
S3, processing the bidirectional long and short term memory network Bi-LSTM, inputting the image characteristics with time correlation into the subsequent bidirectional long and short term memory network Bi-LSTM for processing;
s4, classifying prediction results, predicting the video category of each time step by the BI-directional long-short term memory network BI-LSTM, and integrating the prediction results into final classification; in this embodiment, the final classification results include, but are not limited to, pipe rupture, misconnection, corrosion fouling, and pipe deformation.
In the above step S2, the residual error network Resnet50 has 50 layers in total, and is composed of 49 convolutional layers and 1 fully-connected layer, where the fully-connected layer is used to convert the output of the convolutional layers into a multidimensional vector; the convolution layers comprise conv1, conv2_ x, conv3_ x, conv4_ x and conv5_ x, wherein the conv1 consists of 7 by 64 single convolution layers and is used for down-sampling an input image with the size of 224 by 224; the conv2_ x, conv3_ x, conv4_ x and conv5_ x are respectively composed of 3, 4, 6 and 3 residual blocks, and each residual block is respectively composed of different numbers of 1 × 1, 3 × 3 and 1 × 1 convolution layers. The last layer is a fully-connected layer for converting the output of the convolutional layer into a multi-dimensional vector. In the original Resnet, the vectors output by the fully-connected layer are passed to the softmax layer for classification. The specific structure of the residual error network Resnet50 is shown in FIG. 3.
As shown in fig. 5, which is a basic structure diagram of a network element of the long-short term memory network LSTM, the key of the long-short term memory network LSTM is the state of the element and the horizontal lines passing through the element. The data are sequentially input into the long-short term memory network LSTM in the form of vector sequence. The cells in the long short term memory network LSTM record these vectors and pass them to the next long short term memory network LSTM. Specifically, the cells in the current long-short term memory network LSTM record a state vector summarizing the past input data, update the current state according to the current input data, and then pass it to the next long-short term memory network LSTMThe network LSTM. The long and short term memory network LSTM updates the current state through a gating mechanism, so that the network selectively forgets the past state, selectively remembers the current input, and decides what to output as the current state. CtAs the state of the cell at time step t, its update is determined by the following parameters:
ft＝σ(Wf·[ht-1，xt]+bf)
it＝σ(Wi·[ht-1，xt]+bi)
ot＝σ(Wo·[ht-1，xt]+bo)
ht＝ot×tanh(Ct)
where σ is sigmoid function, Wf、Wi、Wo、WcWeight matrix being a linear transformation, bi、bf、bo、bcAs a bias vector, itFor input of gate vector, ftTo forget the gate vector, otTo output the gate vector, htTo output a hidden state vector.
However, the conventional long-short term memory network LSTM only receives and learns information in the future direction, which means that there is a certain loss of information in the past direction. As shown in FIG. 6, the Bi-directional long-short term memory network Bi-LSTM derived from the conventional model is composed of two layers of long-short term memory networks LSTM, which are respectively used for learning information of the future direction and the past direction, and more comprehensively capturing the feature information of the sequence vector. And the output of the two layers of long and short term memory networks LSTM is used for merging into a maximum value, a minimum value, an average value, multiplication and series connection, and after the vector connection of the Bi-directional long and short term memory network Bi-LSTM, the output vector finishes the classification of the defect sequence through the full connection layer and SoftMax.
Even if a deep learning method is used in the traditional drainage pipeline detection method, a single-frame picture is identified, and a video sequence is not regarded as a whole. In the invention, the algorithm is firstly applied to the capsule robot, and the algorithm is optimized according to the characteristic of the capsule robot that shoots the image; secondly, the capsule robot image sequence is identified as a whole, so that the problem of poor image quality in the prior art is solved; the time, manpower and material resource cost of manual detection can be obviously reduced, the working efficiency of the capsule robot is improved, and the requirement of large-range and high-frequency defect detection of the urban drainage system is met.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (7)
1. A method for identifying defects of a capsule robot drainage pipe network based on Resnet and LSTM is characterized by comprising the following steps:
s1, inputting and matrix converting an image sequence, inputting the marked drainage pipeline defect image sequence collected by the capsule robot into a long-term convolution network LRCN, performing matrix conversion on the images before inputting the images into the long-term convolution network LRCN, and converting the images into a three-dimensional matrix of 224 x 3 in a jpg format;
s2, extracting image features, namely, after inputting a drainage pipeline defect image sequence, dividing the drainage pipeline defect image sequence into a plurality of single-frame images, inputting the single-frame images into a residual error network Resnet50 to extract the image features, wherein the process of extracting the image features comprises the following steps:
s21, the input image is first passed through a convolution kernel of 7 × 7, the step size is 2, and the image output is 112 × 64;
s22, passing through a max pooling layer, kernel size 3 × 3, step size 2, image output 56 × 64;
s23, passing through 16 residual blocks, gradually reducing the size of the feature map, increasing the number of channels, and outputting an image of 7 × 2048;
s24, finally, converting the data into 2048-dimensional vectors through an average pooling layer;
s25, three full-connection layers comprising 512, 256 and 64 neurons are arranged behind the average pooling layer and are used for reducing the dimension of the image features into 64-dimensional vectors;
s3, processing the bidirectional long and short term memory network Bi-LSTM, inputting the image characteristics with time correlation into the subsequent bidirectional long and short term memory network Bi-LSTM for processing;
and S4, classifying the prediction results, predicting the video category of each time step by the BI-directional long-short term memory network BI-LSTM, and integrating the prediction results into the final classification.
2. The method for identifying defects in a drainpipe network of a capsule robot based on Resnet and LSTM as claimed in claim 1, wherein in step S2, the residual error network Resnet50 has 50 layers, which is composed of 49 convolutional layers and 1 fully-connected layer, and the fully-connected layer is used for converting the output of the convolutional layers into multi-dimensional vectors.
3. The Resnet and LSTM based capsule robot drainpipe network defect identification method of claim 2, wherein the convolutional layer comprises conv1, conv2_ x, conv3_ x, conv4_ x and conv5_ x, wherein:
the conv1 consists of a single convolution layer of 7 x 64 for down-sampling the input image size 224 x 224;
the conv2_ x, conv3_ x, conv4_ x and conv5_ x are respectively composed of 3, 4, 6 and 3 residual blocks, and each residual block is respectively composed of different numbers of 1 × 1, 3 × 3 and 1 × 1 convolution layers.
4. The method for identifying defects in a drainpipe network of a capsule robot based on Resnet and LSTM as claimed in claim 1, wherein in step S1, each drainpipe defect image sequence is composed of 24 consecutive frames, i.e. converted into 24 matrices of 224 × 3.
5. The method for identifying defects in a drainage pipe network of a capsule robot based on Resnet and LSTM according to claim 1, wherein in step S3, the Bi-directional long-short term memory network Bi-LSTM is composed of two layers of long-short term memory networks LSTM, and is used for learning information in future direction and past direction respectively, and capturing feature information of sequence vectors more comprehensively.
6. The Resnet and LSTM based capsule robot drainage pipe network defect identification method of claim 5, wherein the outputs of the two layers of long and short term memory networks LSTM are combined into a series, and after the vectors of Bi-LSTM are connected in series through the bidirectional long and short term memory network, the output vectors complete the classification of defect sequences through the full connection layer and SoftMax.
7. The method for identifying defects in a drainpipe network of a capsule robot based on Resnet and LSTM as claimed in claim 1, wherein in step S4, the final classification includes but is not limited to pipe rupture, misconnection of branch pipes, corrosion and fouling, and deformation of pipes.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010874511.1A CN111986188A (en) | 2020-08-27 | 2020-08-27 | Capsule robot drainage pipe network defect identification method based on Resnet and LSTM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010874511.1A CN111986188A (en) | 2020-08-27 | 2020-08-27 | Capsule robot drainage pipe network defect identification method based on Resnet and LSTM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111986188A true CN111986188A (en) | 2020-11-24 |
Family
ID=73440612
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010874511.1A Pending CN111986188A (en) | 2020-08-27 | 2020-08-27 | Capsule robot drainage pipe network defect identification method based on Resnet and LSTM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111986188A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112487989A (en) * | 2020-12-01 | 2021-03-12 | 重庆邮电大学 | Video expression recognition method based on capsule-long-and-short-term memory neural network |
| CN112784914A (en) * | 2021-01-29 | 2021-05-11 | 长兴云尚科技有限公司 | Pipe gallery video intelligent attribute detection method and system based on cloud processing |
| CN113326790A (en) * | 2021-06-10 | 2021-08-31 | 深圳市环水管网科技服务有限公司 | Capsule robot drain pipe disease detection method based on abnormal detection thinking |
| CN113763382A (en) * | 2021-11-09 | 2021-12-07 | 常州微亿智造科技有限公司 | Detection device in industrial quality inspection |
| CN113780305A (en) * | 2021-08-10 | 2021-12-10 | 西安电子科技大学 | Saliency target detection method based on interaction of two clues |
| CN114445712A (en) * | 2022-01-29 | 2022-05-06 | 东南大学 | Expressway pavement disease identification method based on improved YOLOv5 model |
| CN117115147A (en) * | 2023-10-19 | 2023-11-24 | 山东华盛创新纺织科技有限公司 | Textile detection method and system based on machine vision |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108932500A (en) * | 2018-07-09 | 2018-12-04 | 广州智能装备研究院有限公司 | A kind of dynamic gesture identification method and system based on deep neural network |
| CN109559302A (en) * | 2018-11-23 | 2019-04-02 | 北京市新技术应用研究所 | Pipe video defect inspection method based on convolutional neural networks |
| CN111046808A (en) * | 2019-12-13 | 2020-04-21 | 江苏大学 | Analysis method of drinking and playing waterer for raising pigs by adopting residual convolutional neural network and long-short term memory classification group |
| CN111091045A (en) * | 2019-10-25 | 2020-05-01 | 重庆邮电大学 | Sign language identification method based on space-time attention mechanism |
-
2020
- 2020-08-27 CN CN202010874511.1A patent/CN111986188A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108932500A (en) * | 2018-07-09 | 2018-12-04 | 广州智能装备研究院有限公司 | A kind of dynamic gesture identification method and system based on deep neural network |
| CN109559302A (en) * | 2018-11-23 | 2019-04-02 | 北京市新技术应用研究所 | Pipe video defect inspection method based on convolutional neural networks |
| CN111091045A (en) * | 2019-10-25 | 2020-05-01 | 重庆邮电大学 | Sign language identification method based on space-time attention mechanism |
| CN111046808A (en) * | 2019-12-13 | 2020-04-21 | 江苏大学 | Analysis method of drinking and playing waterer for raising pigs by adopting residual convolutional neural network and long-short term memory classification group |
Non-Patent Citations (2)
| Title |
|---|
| 傅真珍: ""基于深度学习的多方位角SAR目标识别研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 谭越等: ""基于 ResNet 和 BiLSTM 的僵尸网络检测方法"", 《通信技术》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112487989A (en) * | 2020-12-01 | 2021-03-12 | 重庆邮电大学 | Video expression recognition method based on capsule-long-and-short-term memory neural network |
| CN112487989B (en) * | 2020-12-01 | 2022-07-15 | 重庆邮电大学 | Video expression recognition method based on capsule-long-and-short-term memory neural network |
| CN112784914A (en) * | 2021-01-29 | 2021-05-11 | 长兴云尚科技有限公司 | Pipe gallery video intelligent attribute detection method and system based on cloud processing |
| CN112784914B (en) * | 2021-01-29 | 2023-04-07 | 长兴云尚科技有限公司 | Pipe gallery video intelligent attribute detection method and system based on cloud processing |
| CN113326790A (en) * | 2021-06-10 | 2021-08-31 | 深圳市环水管网科技服务有限公司 | Capsule robot drain pipe disease detection method based on abnormal detection thinking |
| CN113780305A (en) * | 2021-08-10 | 2021-12-10 | 西安电子科技大学 | Saliency target detection method based on interaction of two clues |
| CN113780305B (en) * | 2021-08-10 | 2024-03-12 | 西安电子科技大学 | Significance target detection method based on interaction of two clues |
| CN113763382A (en) * | 2021-11-09 | 2021-12-07 | 常州微亿智造科技有限公司 | Detection device in industrial quality inspection |
| CN113763382B (en) * | 2021-11-09 | 2022-02-11 | 常州微亿智造科技有限公司 | Detection device in industrial quality inspection |
| CN114445712A (en) * | 2022-01-29 | 2022-05-06 | 东南大学 | Expressway pavement disease identification method based on improved YOLOv5 model |
| CN117115147A (en) * | 2023-10-19 | 2023-11-24 | 山东华盛创新纺织科技有限公司 | Textile detection method and system based on machine vision |
| CN117115147B (en) * | 2023-10-19 | 2024-01-26 | 山东华盛创新纺织科技有限公司 | Textile detection method and system based on machine vision |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111986188A (en) | Capsule robot drainage pipe network defect identification method based on Resnet and LSTM | |
| CN108961235B (en) | Defective insulator identification method based on YOLOv3 network and particle filter algorithm | |
| CN112381788B (en) | Part surface defect increment detection method based on double-branch matching network | |
| CN110826588A (en) | Drainage pipeline defect detection method based on attention mechanism | |
| CN110648310A (en) | Weak supervision casting defect identification method based on attention mechanism | |
| CN109635763B (en) | Crowd density estimation method | |
| CN111368690A (en) | Deep learning-based video image ship detection method and system under influence of sea waves | |
| CN116485717B (en) | Concrete dam surface crack detection method based on pixel-level deep learning | |
| CN110599459A (en) | Underground pipe network risk assessment cloud system based on deep learning | |
| CN114155474A (en) | Damage identification technology based on video semantic segmentation algorithm | |
| CN115272777B (en) | Semi-supervised image analysis method for power transmission scene | |
| Li et al. | Image quality assessment using deep convolutional networks | |
| CN114943689A (en) | Method for detecting components of steel cold-rolling annealing furnace based on semi-supervised learning | |
| CN110599458A (en) | Underground pipe network detection and evaluation cloud system based on convolutional neural network | |
| CN115147418B (en) | Compression training method and device for defect detection model | |
| CN114283137A (en) | Photovoltaic module hot spot defect detection method based on multi-scale characteristic diagram inference network | |
| CN111160389A (en) | Lithology identification method based on fusion of VGG | |
| CN111652240A (en) | Image local feature detection and description method based on CNN | |
| CN113313031A (en) | Deep learning-based lane line detection and vehicle transverse positioning method | |
| CN115131747A (en) | Knowledge distillation-based power transmission channel engineering vehicle target detection method and system | |
| Jiang et al. | Attention M-net for automatic pixel-level micro-crack detection of photovoltaic module cells in electroluminescence images | |
| CN110942026B (en) | Deep learning-based capsule robot drain pipe disease detection method and system | |
| CN116824352A (en) | Water surface floater identification method based on semantic segmentation and image anomaly detection | |
| CN113034445B (en) | Multi-scale connection image defogging algorithm based on UNet3+ | |
| CN115082650A (en) | Implementation method of automatic pipeline defect labeling tool based on convolutional neural network |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201124 |
|
| RJ01 | Rejection of invention patent application after publication |