CN104978580B - A kind of insulator recognition methods for unmanned plane inspection transmission line of electricity - Google Patents
A kind of insulator recognition methods for unmanned plane inspection transmission line of electricity Download PDFInfo
- Publication number
- CN104978580B CN104978580B CN201510330413.0A CN201510330413A CN104978580B CN 104978580 B CN104978580 B CN 104978580B CN 201510330413 A CN201510330413 A CN 201510330413A CN 104978580 B CN104978580 B CN 104978580B
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- munderover
- msup
- msubsup
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 58
- 239000012212 insulator Substances 0.000 title claims abstract description 57
- 230000005540 biological transmission Effects 0.000 title claims abstract description 39
- 230000005611 electricity Effects 0.000 title claims abstract description 23
- 238000007689 inspection Methods 0.000 title claims abstract description 20
- 238000012549 training Methods 0.000 claims abstract description 38
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 33
- 238000001514 detection method Methods 0.000 claims abstract description 14
- 238000012545 processing Methods 0.000 claims abstract description 13
- 238000009413 insulation Methods 0.000 claims abstract description 9
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 238000013135 deep learning Methods 0.000 claims abstract description 5
- 238000012856 packing Methods 0.000 claims abstract description 5
- 230000001629 suppression Effects 0.000 claims abstract description 4
- 230000008569 process Effects 0.000 claims description 20
- 230000006870 function Effects 0.000 claims description 14
- 230000004913 activation Effects 0.000 claims description 6
- 238000011478 gradient descent method Methods 0.000 claims description 5
- 230000007423 decrease Effects 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 230000007935 neutral effect Effects 0.000 claims description 3
- 238000003475 lamination Methods 0.000 claims description 2
- 238000012360 testing method Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 description 6
- 239000004020 conductor Substances 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000003213 activating effect Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/19—Recognition using electronic means
- G06V30/192—Recognition using electronic means using simultaneous comparisons or correlations of the image signals with a plurality of references
- G06V30/194—References adjustable by an adaptive method, e.g. learning
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses one kind to be efficiently used for unmanned plane inspection electric transmission line isolator recognition methods, including image variants：Extraction for trained subgraph and carries out preliminary treatment from electric transmission line isolator image, forms training dataset；Packing processing, the corresponding label of addition image are carried out to the subgraph for being used for training extracted；Data are trained using convolutional neural networks (CNN) algorithm in deep learning, obtain the detection model for insulator；Detect insulator target area：Transmission line of electricity image is detected, obtains the candidate frame of insulation sub-goal；Non-maxima suppression is carried out to candidate frame, obtains final insulator candidate frame；Fitting a straight line operation is carried out to obtained final insulator candidate frame, obtains central point, the angle and size information of candidate frame, are finally labeled on electric transmission line isolator image.The image that the application obtains inspection screens, and mitigates the burden of artificial examination, has broad application prospects.
Description
Technical field
The present invention relates to Digital Image Processing and mode identification technology in transmission line equipment detection technique field, especially relates to
A kind of and efficient recognition methods for being used for unmanned plane inspection electric transmission line isolator.
Background technology
Insulator is part important in overhead transmission line, for support and fixed busbar with energized conductor, simultaneously
Make have enough distances and insulation between energized conductor or between conductor and the earth.Since overhead transmission line is chronically exposed to nature
In environment, influenced be subject to nature or human factor, there are aging circuit and destroy the problems such as, if not to these problems into
Row, which is inspected periodically and overhauled, may cause serious accident.
Artificial line walking detection efficiency is low, and dangerous high.With the development of unmanned air vehicle technique, pass through unmanned plane skill
Art gathers high-tension line image and these image informations is handled, it is possible to reduce personnel cost and the peace for ensureing construction personnel
Entirely, while work efficiency can be improved.
Since electric power line pole tower position geographical environment is complicated, cause the image background of acquisition also relative complex, this
Identification to succeeding target causes difficulty with positioning, and the transmission line status detection technique based on Aerial Images is still in
Starting stage, refers to document and achievement in research is less.
In the prior art, a kind of method is to utilize colouring information, using max-thresholds method, maximum variance between clusters to colour
Image is split.There are the shortcomings that：This kind of method is influenced than more serious by illumination, and natural environment residing for transmission line of electricity
Complexity, there is the complex background such as trees, river, road so that this kind of method Detection accuracy is not high.
Another kind of method is using the oval information of the laminated structure of insulator, and ellipse is detected with Hough transform.In the presence of
The shortcomings that：The problem of due to shooting angle, there is a situation where to block between piece and piece, cause detection error.
The content of the invention
To solve the shortcomings of the prior art, the invention discloses one kind to be efficiently used for unmanned plane inspection transmission line of electricity
Insulator recognition methods, the technology using deep learning method by learning to the transmission line of electricity image document taken photo by plane,
Then the angle information of insulator is calculated using line fitting method, can be completed to insulator in unmanned plane line walking image
Identification location technology requirement, improve the accuracy and robustness to isolator detecting.
To achieve the above object, concrete scheme of the invention is as follows：
A kind of insulator recognition methods for unmanned plane inspection transmission line of electricity, comprises the following steps：
Step 1：Image variants：Extraction is gone forward side by side for the subgraph of training from electric transmission line isolator image
Row preliminary treatment, forms training dataset；
Step 2：Packing processing, the corresponding label of addition image are carried out to the subgraph for being used for training extracted；
Step 3：Using convolutional neural networks (CNN) algorithm in deep learning to the data in step 1 data set into
Row training, obtains the detection model for insulator；
Step 4：Detect insulator target area：Transmission line of electricity image is detected, obtains the candidate of insulation sub-goal
Frame；
Step 5：Non-maxima suppression is carried out to candidate frame, obtains final insulator candidate frame；
Step 6：Fitting a straight line operation is carried out to obtained final insulator candidate frame, obtains central point, candidate frame
Angle and size information, are finally labeled on electric transmission line isolator image.
In the step 1, image variants process is：Insulation sub portion is extracted in electric transmission line isolator image
The area image of part, shaft tower and background, processing is zoomed in and out by the image extracted, and angle is carried out to the insulator extracted
Rotation process.
In the step 3, when carrying out model training with convolutional neural networks algorithm：
The template parameter that uses of initial training is set first, including the number of plies of convolutional neural networks, convolution kernel it is big
It is small, the initial weight of each node, if handled using down-sampling, per the number of layer data input and output, activation primitive, every layer
The learning efficiency that the gradient of convolutional neural networks declines.
In the step 3, specifically the isolator detecting model training process based on convolutional neural networks includes following step
Suddenly：
3-1) forward conduction (Feedforward Pass)：Divide the coloured image of input to RGB triple channels extraction pixel letter
Breath, the input information as convolutional neural networks；
3-2) conduction (BackPropagation Pass) backward：The class label that optimization forward conduction obtains and sample are real
Loss function between the label of border.
The step 3-1) in, the structure of convolutional neural networks, uses six layers of convolutional neural networks structured training template；
First layer is convolutional layer, and third layer is convolutional layer, and layer 5 is convolutional layer, and convolutional layer use sets the volume of pixel size
Product core carries out convolution operation with input information and obtains feature vector；
The second layer is down-sampled layer, and the 4th layer is down-sampled layer, and definition of the down-sampled layer in parameterized template, is dropped
Sampling operation, is maximized as output using in the image block of the pixel size of setting, has by down-sampled process
Effect reduces data processing amount on the basis of useful information is preserved；
Layer 6 is full articulamentum：The feature vector that layer 5 is obtained is integrated, and is formed a long vector, is passed it to
Activation primitive obtains the judgement to input sample classification, selects label of the output valve of maximum as input sample.
The step 3-2) in, it is assumed that for training set form it is { (x(1),y(1)),(x(2),y(2)),...,(x(n),y(n))},x(i)Represent i-th of training data, y(i)Represent data x(i)Corresponding data label, training dataset include n sample.
For single sample (x, y), the result learnt is hw,b(x), its loss function is：
So whole loss function J (W, b) is：
In formula (2), sl, sl+1Represent the number of l layers and l+1 layers neuron node, nlRepresent the number of plies of neutral net,Represent the weight coefficient of connection i-th of node of l layers and l+1 j-th of node of layer in l group weight parameters,For l layers
The biasing of i-th of node, Section 1 J (W, b；x(i),y(i)) it is a mean square deviation item, Section 2 is regularization term, for reducing
The amplitude of weight, prevents over-fitting, γ coefficients in order to control.
Parameter W and b are updated with gradient descent method：
Wherein, the α in formula (3) is learning rate, for the speed for controlling gradient to decline.
In the step 4, the detection of insulator is carried out using convolutional neural networks, method detailed is as follows：
4-1) extract trained convolutional neural networks template, including the network structure that weight, biasing and training use；
Then according to these parameter initialization test programs；
Image 4-2) is loaded into, since the image that unmanned plane collects is larger, image is zoomed in and out to accelerate follow-up fortune
Calculate, in order to accurately position the position of insulator, add multi-scale method, slider bar operation is carried out on multiple scales, obtain
Specific target image block；
4-3) using target image block as input, forward conduction operation is carried out, obtains the generic of object block；
4-4) preserve the other target block message of insulator class, start position and length and width information including object block.
In the step 6, insulator is marked using line fitting method, detailed process is as follows：
6-1) the m of the t classification obtained in obtaining step fivei, i={ 1 ..., t } a candidate frame information, calculates each wait
The center position of frame is selected, and is preserved；
6-2) center position is (P, Q), the center of each classification can be intended with the mode of linear fit
Close, accurately position the position of each insulator, can solve the problems, such as this with unitary linear fit here：
Y=kX+b'(4)
Least square fitting method 6-3) is used, calculates a straight line that can most reflect X and Y relations.
The step 6-3) in, defining loss function is：
Wherein, (pj,qj) be belong to i classes j-th of rectangle frame central point.
By seeking local derviation to b' and k, to obtain the optimal solution of b' and k：
The last solution for obtaining k and b' is
Wherein, (pj,qj) be belong to i classes j-th of rectangle frame central point, miRepresent the number of the rectangle frame of the i-th class.
Beneficial effects of the present invention：
The application uses convolutional layer, down-sampled layer and full linking layer：, will after three-layer coil product and two layers of down-sampled operation
All characteristic blocks carry out full attended operation, have thus obtained describing the final feature of image block.
Effective identification technology of electric transmission line isolator, which can be realized effectively, determines the insulation sub-goal in inspection image
Position, basis is provided for subsequent defective diagnosis.Meanwhile the image that this technology can also obtain inspection screens, mitigate artificial
The burden of examination, has broad application prospects.
Brief description of the drawings
Fig. 1 CNN structure charts；
Fig. 2 convolution and down-sampled process；
Fig. 3 unmanned plane transmission line of electricity images；
Transmission line of electricity image after Fig. 4 candidate frames mark；
The final isolator detecting result images of Fig. 5；
Fig. 6 trains flow chart；
Fig. 7 target detections and positioning flow figure.
Embodiment：
The present invention is described in detail below in conjunction with the accompanying drawings：
The training flow chart of experimentation is as shown in fig. 6, extraction data set, data set packing, start training process, instructing
Be trained hierarchical structure and initiation parameter before practicing, during training, train forward, reverse feedback, judge maximum iteration/
Required precision, if it is, output template, otherwise, continues to train.
For target identification with positioning flow figure as shown in fig. 7, input picture, initializes detection model, sliding window extracts subgraph,
Judge whether it is insulator, if it is not, sliding window extraction subgraph process is returned to, if so, storage candidate frame, is fitted candidate
Frame, in artwork mark position, output.
One kind is efficiently used for unmanned plane inspection electric transmission line isolator recognition methods, and specific steps include：
1) training set：Subgraph of the extraction for training from transmission line of electricity image, forms training dataset.
2) packing processing is carried out to the insulation subgraph extracted, adds the label of correspondence image；
3) utilize convolutional neural networks (CNN) algorithm in deep learning to be trained data, obtain being directed to insulator
Detection model；
4) insulator target area is detected：Transmission line of electricity image is detected, obtains the candidate frame of insulation sub-goal；
5) non-maxima suppression is carried out to candidate frame, obtains final insulator candidate frame.
6) fitting a straight line operation is carried out to obtained candidate frame, obtains central point, the angle and size information of frame, finally exist
It is labeled in artwork.
The image acquisition process of the step 1) is：The administrative division map of insulator part, shaft tower and background is extracted in artwork
Picture, in order to meet trained requirement, zooms in and out processing, size zooms to 64*64 pixels by the image extracted.In reality
In the situation of border, due to shooting angle reason obtains shaft tower image may be there are inclined situation, the insulation for this to extracting
Son carries out the rotation process of low-angle, to increase the diversity of data set, improves the robustness of training pattern.
In the step 3) step, the specific method that model training is carried out with CNN algorithms is as follows：
(1) template parameter for setting initial training to use is first had to, including the number of plies of convolutional neural networks, convolution
The size of core, the initial weight of each node, if handled using down-sampling, per the number of layer data input and output, activate letter
Number etc., also has the setting of the learning efficiency of the gradient decline of corresponding templates.
(2) network structure, here we use six layers of convolutional neural networks structured training template.CNN in form such as Fig. 1
Shown in structure chart.
Specifically the isolator detecting model training process based on CNN mainly has following two step：
A. forward conduction：RGB triple channels are divided to extract Pixel Information, the input as convolutional network the coloured image of input
Information, first layer are convolutional layer (Convolution), and convolution operation is carried out with the convolution kernel and input information of 5*5 pixel sizes.
It can strengthen original signal by convolution operation, and reduce the influence of noise.The second layer is down-sampled layer, according to parameterized template
In definition, carry out it is down-sampled operation (SubSampling/Pooling), taken most using in the image block of 2*2 here
Big value is as output.Data processing amount effectively can be reduced on the basis of useful information is preserved by down-sampled process；The
Three-layer coil lamination, the 4th layer of down-sampled layer, layer 5 convolutional layer carry out similar operation；Layer 6 is full articulamentum：By the 5th
The feature vector that layer obtains is integrated, and forms a long vector, is passed it to activation primitive and is obtained and input sample classification is sentenced
It is disconnected.As shown in Figure 2.
Convolutional layer, down-sampled layer and full linking layer are further described below below：
Input color image, carries out convolution algorithm, then plus one partially with a trainable wave filter fx and image
Bx is held, this has just extracted a characteristics of image, obtains the Cx of convolutional layer；In order to reduce data volume, pixel in four neighborhood parts
Change is maximized as output, then by the nonlinear transformation of the complete paired data of ReLu activation primitives, nonlinear transformation can be with
Reduce the linear relationship between feature and feature, the expressive faculty of Enhanced feature.Moreover, the processing of down-sampled layer in this way
Afterwards, the data volume obtained is only a quarter of convolutional layer.
After three-layer coil product and two layers of down-sampled operation, all characteristic blocks are subjected to full attended operation, thus
Arrive and the final feature of image block has been described.For the classification of prognostic chart picture, by obtained Feature Descriptor into line activating letter
Number calculation process, selects label of the output valve of maximum as image block.
B. to conduction after：The class label obtained by Feedforward Pass is with sample physical tags come counting loss
Function.Assuming that it is { (x that we, which are used for training set form,(1),y(1)),(x(2),y(2)),...,(x(n),y(n)), training dataset bag
Containing n sample, for single sample (x, y), the result learnt is hw,b(x), its loss function is：
So whole loss function is：
Formula (2) slRepresent the number of neuron in l layers,Represent to connect l layers in l group weight parameters i-th
The weight coefficient of node and l+1 j-th of node of layer.Section 1 J (W, b；x(i),y(i)) it is a mean square deviation item, Section 2 is
Regularization term, for reducing the amplitude of weight, prevents over-fitting, γ coefficients in order to control.Updated with each iteration of gradient descent method
Parameter W and b are updated：
α in formula (3) is learning rate, for the speed for controlling gradient to decline.
In following pseudocode, Δ W(l)It is one and matrix W(l)The identical matrix of dimension, Δ b(l)It is one and b(l)
The identical vector of dimension.In the following, provide an iteration process realized in gradient descent method：
1. couple all layers of l, make Δ W(l):=0, Δ b(l):=0；
2. couple i=1 to n：
A. calculated using back-propagation algorithm
B. calculate
C. calculate
3. update weight parameter：
In this way, what can be repeated reaches the value for reducing loss function J (W, b) using gradient descent method to iterate to calculate,
And then solve whole neutral net.
In the step 4), the detection of insulator is carried out using CNN, method detailed is as follows：
(1) trained CNN templates, including the network structure that weight, biasing and training use are extracted；Then according to this
A little parameter initialization test program frames；
(2) be loaded into image (as shown in Figure 3), the image collected due to unmanned plane is larger (5184*3456), to image into
Row is scaled to accelerate follow-up computing.In order to accurately position the position of insulator, multi-scale method is added, on multiple scales
Slider bar operation is carried out, obtains specific target image block.
(3) using target image block as input, CNN Feedforward Pass operations is carried out, obtain the affiliated of object block
Classification.(4) the other target block message of insulator class, start position and length and width information including object block are preserved, Fig. 4 is warp
Cross the image for marking insulator；
In the step 6), insulator is marked using approximating method, detailed process is as follows：
(1) m for the t classification that obtaining step (5) obtainsi, i={ 1 ..., t } a candidate frame information, calculates each frame
Central point position
Put, and preserve.
(2) center position is (P, Q), the center of each classification can be intended with the mode of linear fit
Close, accurately position the position of each insulator.Here this can be solved the problems, such as with unitary linear fit：
Y=kX+b'(4)
(3) least square fitting method is used, calculates a straight line that can most reflect X and Y relations：
Defining loss function is：
By seeking local derviation to b' and k, to obtain the optimal solution of b' and k：
The last solution for obtaining k and b' is
The angle information of insulator is thus determined, marks the specific position of insulator in artwork according to angle information
Put.If Fig. 5 is that final insulator marks image.
Although above-mentioned be described the embodiment of the present invention with reference to attached drawing, model not is protected to the present invention
The limitation enclosed, those skilled in the art should understand that, on the basis of technical scheme, those skilled in the art are not
Need to make the creative labor the various modifications that can be made or deformation still within protection scope of the present invention.
Claims (10)
1. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity, it is characterized in that, comprise the following steps：
Step 1：Image variants：Extraction for trained subgraph and carries out just from electric transmission line isolator image
Step processing, forms training dataset；
Step 2：Packing processing is carried out to the subgraph for being used for training extracted, adds the label of correspondence image；
Step 3：The data in step 1 data set are trained using the convolutional neural networks algorithm in deep learning, are obtained
To the detection model for insulator；
Step 4：Detect insulator target area：Transmission line of electricity image is detected, obtains the candidate frame of insulation sub-goal；
Step 5：Non-maxima suppression is carried out to candidate frame, obtains final insulator candidate frame；
Step 6：Fitting a straight line operation is carried out to obtained final insulator candidate frame, obtains central point, the angle of candidate frame
And size information, finally it is labeled on electric transmission line isolator image.
2. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity as claimed in claim 1, it is characterized in that, institute
State in step 1, image variants process is：Insulator part, shaft tower and the back of the body are extracted in electric transmission line isolator image
The area image of scape, processing is zoomed in and out by the image extracted, and the rotation process of angle is carried out to the insulator extracted.
3. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity as claimed in claim 1, it is characterized in that, institute
State in step 3, when carrying out model training with convolutional neural networks algorithm：
The template parameter for setting initial training to use first, including the number of plies of convolutional neural networks, the size of convolution kernel,
The initial weight of each node, if handled using down-sampling, per the number of layer data input and output, activation primitive, every layer of volume
The learning efficiency that the gradient of product neutral net declines.
4. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity as described in claim 1 or 3, its feature
It is that in the step 3, specifically the isolator detecting model training process based on convolutional neural networks comprises the following steps：
3-1) forward conduction：RGB triple channels are divided to extract Pixel Information the coloured image of input, as the defeated of convolutional neural networks
Enter information；
3-2) backward conduction：The class label obtained by forward conduction is with sample physical tags come counting loss function.
5. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity as claimed in claim 4, it is characterized in that, institute
State step 3-1) in, the structure of convolutional neural networks, using six layers of convolutional neural networks structured training template, first layer is volume
Lamination, third layer are convolutional layers, and layer 5 is convolutional layer, and convolutional layer is carried out with the convolution kernel of setting pixel size with input information
Convolution operation obtains feature vector；
The second layer is down-sampled layer, and the 4th layer is down-sampled layer, and definition of the down-sampled layer in parameterized template, carries out down-sampled
Operation, is maximized as output using in the image block of the pixel size of setting, effective by down-sampled process
Data processing amount is reduced on the basis of useful information is preserved；
Layer 6 is full articulamentum：The feature vector that layer 5 is obtained is integrated, and is formed a long vector, is passed it to activation
Function obtains the judgement to input sample classification, selects label of the output valve of maximum as image block.
6. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity as claimed in claim 4, it is characterized in that, institute
State step 3-2) in, it is assumed that for training set form it is { (x(1),y(1)),(x(2),y(2)),...,(x(n),y(n))},x(i)Represent
I-th of training data, y(i)Represent data x(i)Corresponding data label, training dataset include n sample；
For single sample (x, y), the result learnt is hw,b(x), its loss function is：
<mrow>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>W</mi>
<mo>,</mo>
<mi>b</mi>
<mo>;</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>h</mi>
<mrow>
<mi>w</mi>
<mo>,</mo>
<mi>b</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mi>y</mi>
<mo>|</mo>
<msup>
<mo>|</mo>
<mn>2</mn>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
So whole loss function is：
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>W</mi>
<mo>,</mo>
<mi>b</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mo>&lsqb;</mo>
<mfrac>
<mn>1</mn>
<mi>n</mi>
</mfrac>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>W</mi>
<mo>,</mo>
<mi>b</mi>
<mo>;</mo>
<msup>
<mi>x</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>,</mo>
<msup>
<mi>y</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
<mo>+</mo>
<mfrac>
<mi>&gamma;</mi>
<mn>2</mn>
</mfrac>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>l</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<msub>
<mi>n</mi>
<mi>l</mi>
</msub>
<mo>-</mo>
<mn>1</mn>
</mrow>
</munderover>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>s</mi>
<mi>l</mi>
</msub>
</munderover>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>s</mi>
<mrow>
<mi>l</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>W</mi>
<mrow>
<mi>j</mi>
<mi>i</mi>
</mrow>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>)</mo>
</mrow>
</msubsup>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>=</mo>
<mo>&lsqb;</mo>
<mfrac>
<mn>1</mn>
<mi>n</mi>
</mfrac>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<mrow>
<mo>(</mo>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>h</mi>
<mrow>
<mi>w</mi>
<mo>,</mo>
<mi>b</mi>
</mrow>
</msub>
<mo>(</mo>
<msup>
<mi>x</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>)</mo>
<mo>-</mo>
<msup>
<mi>y</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</msup>
<mo>|</mo>
<msup>
<mo>|</mo>
<mn>2</mn>
</msup>
<mo>)</mo>
</mrow>
<mo>&rsqb;</mo>
<mo>+</mo>
<mo>+</mo>
<mfrac>
<mi>&gamma;</mi>
<mn>2</mn>
</mfrac>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>l</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<msub>
<mi>n</mi>
<mi>l</mi>
</msub>
<mo>-</mo>
<mn>1</mn>
</mrow>
</munderover>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>s</mi>
<mi>l</mi>
</msub>
</munderover>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>s</mi>
<mrow>
<mi>l</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msub>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>W</mi>
<mrow>
<mi>j</mi>
<mi>i</mi>
</mrow>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>)</mo>
</mrow>
</msubsup>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Formula (2) slRepresent the number of neuron in l layers,Represent in l group weight parameters connect i-th of node of l layers with
The weight coefficient of l+1 j-th of node of layer,For the biasing of l i-th of node of layer, Section 1 J (W, b；x(i),y(i)) it is one
A mean square deviation item, Section 2 are regularization terms, for reducing the amplitude of weight, prevent over-fitting, γ coefficients in order to control.
7. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity as claimed in claim 6, it is characterized in that, use
The each iteration renewal of gradient descent method is updated parameter W and b：
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>W</mi>
<mrow>
<mi>j</mi>
<mi>i</mi>
</mrow>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>)</mo>
</mrow>
</msubsup>
<mo>=</mo>
<msubsup>
<mi>W</mi>
<mrow>
<mi>j</mi>
<mi>i</mi>
</mrow>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>)</mo>
</mrow>
</msubsup>
<mo>-</mo>
<mi>&alpha;</mi>
<mfrac>
<mo>&part;</mo>
<mrow>
<mo>&part;</mo>
<msubsup>
<mi>W</mi>
<mrow>
<mi>j</mi>
<mi>i</mi>
</mrow>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>)</mo>
</mrow>
</msubsup>
</mrow>
</mfrac>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>W</mi>
<mo>,</mo>
<mi>b</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>b</mi>
<mi>i</mi>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>)</mo>
</mrow>
</msubsup>
<mo>=</mo>
<msubsup>
<mi>b</mi>
<mi>i</mi>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>)</mo>
</mrow>
</msubsup>
<mo>-</mo>
<mi>&alpha;</mi>
<mfrac>
<mo>&part;</mo>
<mrow>
<mo>&part;</mo>
<msubsup>
<mi>b</mi>
<mi>i</mi>
<mrow>
<mo>(</mo>
<mi>l</mi>
<mo>)</mo>
</mrow>
</msubsup>
</mrow>
</mfrac>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>W</mi>
<mo>,</mo>
<mi>b</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>3</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, the α in formula (3) is learning rate, for the speed for controlling gradient to decline.
8. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity as claimed in claim 1, it is characterized in that, institute
State in step 4, the detection of insulator is carried out using convolutional neural networks, method detailed is as follows：
4-1) extract trained convolutional neural networks template, including the network structure that weight, biasing and training use；Then
According to these parameter initialization test programs；
Image 4-2) is loaded into, since the graphical rule that unmanned plane collects is larger, image is zoomed in and out to accelerate follow-up fortune
Calculate, in order to accurately position the position of insulator, add multi-scale method, slider bar operation is carried out on multiple scales, obtain
Specific target image block；
4-3) using target image block as input, forward conduction operation is carried out, obtains the generic of object block；
4-4) preserve the other target block message of insulator class, start position and length and width information including object block.
9. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity as claimed in claim 1, it is characterized in that, institute
State in step 6, insulator is marked using line fitting method, detailed process is as follows：
6-1) the m of the t classification obtained in obtaining step fivei, i={ 1 ..., t } a candidate frame information, calculates each candidate frame
Center position, and preserve, miRepresent the number of the rectangle frame of the i-th class；
6-2) center position is (P, Q), the center of each classification is fitted with the mode of linear fit, accurately
The position of each insulator is positioned, this is solved the problems, such as with unitary linear fit：
Y=kX+b'(4)
Least square fitting method 6-3) is used, calculates a straight line that can most reflect X and Y relations.
10. a kind of insulator recognition methods for unmanned plane inspection transmission line of electricity as claimed in claim 9, it is characterized in that,
The step 6-3) in, defining loss function is：
<mrow>
<mi>L</mi>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>q</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<mo>(</mo>
<mrow>
<msup>
<mi>b</mi>
<mo>&prime;</mo>
</msup>
<mo>+</mo>
<msub>
<mi>kp</mi>
<mi>j</mi>
</msub>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>5</mn>
<mo>)</mo>
</mrow>
</mrow>
By seeking local derviation to b' and k, to obtain the optimal solution of b' and k：
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<mfrac>
<mrow>
<mo>&part;</mo>
<mi>L</mi>
</mrow>
<mrow>
<mo>&part;</mo>
<msup>
<mi>b</mi>
<mo>&prime;</mo>
</msup>
</mrow>
</mfrac>
<mo>=</mo>
<mn>2</mn>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<mrow>
<mo>(</mo>
<msub>
<mi>q</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<mo>(</mo>
<mrow>
<msup>
<mi>b</mi>
<mo>&prime;</mo>
</msup>
<mo>+</mo>
<msub>
<mi>kp</mi>
<mi>j</mi>
</msub>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mfrac>
<mrow>
<mo>&part;</mo>
<mi>L</mi>
</mrow>
<mrow>
<mo>&part;</mo>
<mi>k</mi>
</mrow>
</mfrac>
<mo>=</mo>
<mo>-</mo>
<mn>2</mn>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<mrow>
<mo>(</mo>
<msub>
<mi>q</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<mo>(</mo>
<mrow>
<msup>
<mi>b</mi>
<mo>&prime;</mo>
</msup>
<mo>+</mo>
<msub>
<mi>kp</mi>
<mi>j</mi>
</msub>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
<msub>
<mi>p</mi>
<mi>j</mi>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>5</mn>
<mo>)</mo>
</mrow>
</mrow>
The last solution for obtaining k and b' is
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
<mo>*</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msub>
<mi>p</mi>
<mi>j</mi>
</msub>
<msub>
<mi>q</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msub>
<mi>p</mi>
<mi>j</mi>
</msub>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msub>
<mi>q</mi>
<mi>j</mi>
</msub>
</mrow>
<mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
<mo>*</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msubsup>
<mi>p</mi>
<mi>j</mi>
<mn>2</mn>
</msubsup>
<mo>-</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msubsup>
<mi>q</mi>
<mi>j</mi>
<mn>2</mn>
</msubsup>
</mrow>
</mfrac>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msup>
<mi>b</mi>
<mo>&prime;</mo>
</msup>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
<mo>*</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msubsup>
<mi>p</mi>
<mi>j</mi>
<mn>2</mn>
</msubsup>
<msub>
<mi>q</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msub>
<mi>p</mi>
<mi>j</mi>
</msub>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msub>
<mi>q</mi>
<mi>j</mi>
</msub>
</mrow>
<mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
<mo>*</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msubsup>
<mi>p</mi>
<mi>j</mi>
<mn>2</mn>
</msubsup>
<mo>-</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</munderover>
<msubsup>
<mi>q</mi>
<mi>j</mi>
<mn>2</mn>
</msubsup>
</mrow>
</mfrac>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>7</mn>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Wherein, (pj,qj) be belong to i classes j-th of rectangle frame central point, miRepresent the number of the rectangle frame of the i-th class.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510330413.0A CN104978580B (en) | 2015-06-15 | 2015-06-15 | A kind of insulator recognition methods for unmanned plane inspection transmission line of electricity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510330413.0A CN104978580B (en) | 2015-06-15 | 2015-06-15 | A kind of insulator recognition methods for unmanned plane inspection transmission line of electricity |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104978580A CN104978580A (en) | 2015-10-14 |
| CN104978580B true CN104978580B (en) | 2018-05-04 |
Family
ID=54275065
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510330413.0A Active CN104978580B (en) | 2015-06-15 | 2015-06-15 | A kind of insulator recognition methods for unmanned plane inspection transmission line of electricity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104978580B (en) |
Families Citing this family (51)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105551036B (en) * | 2015-12-10 | 2019-10-08 | 中国科学院深圳先进技术研究院 | A kind of training method and device of deep learning network |
| CN105528595A (en) * | 2016-02-01 | 2016-04-27 | 成都通甲优博科技有限责任公司 | Method for identifying and positioning power transmission line insulators in unmanned aerial vehicle aerial images |
| CN105844627B (en) * | 2016-03-21 | 2019-02-05 | 华中科技大学 | A kind of sea-surface target image background suppressing method based on convolutional neural networks |
| CN106228158A (en) * | 2016-07-25 | 2016-12-14 | 北京小米移动软件有限公司 | The method and apparatus of picture detection |
| CN106503742B (en) * | 2016-11-01 | 2019-04-26 | 广东电网有限责任公司电力科学研究院 | A kind of visible images insulator recognition methods |
| CN108154072A (en) * | 2016-12-02 | 2018-06-12 | 天津工业大学 | Insulator breakdown of taking photo by plane based on depth convolutional neural networks detects automatically |
| CN106682592B (en) * | 2016-12-08 | 2023-10-27 | 北京泛化智能科技有限公司 | Image automatic identification system and method based on neural network method |
| CN106774389A (en) * | 2016-12-09 | 2017-05-31 | 武汉科技大学 | A kind of four rotor wing unmanned aerial vehicles electricity tower method for inspecting based on motor learning |
| CN106595551B (en) * | 2016-12-13 | 2019-01-04 | 武汉大学 | Ice covering thickness detection method in powerline ice-covering image based on deep learning |
| CN106895824A (en) * | 2017-01-24 | 2017-06-27 | 国网四川省电力公司检修公司 | Unmanned plane localization method based on computer vision |
| CN106934346B (en) * | 2017-01-24 | 2019-03-15 | 北京大学 | A kind of method of target detection performance optimization |
| CN106980365A (en) * | 2017-02-21 | 2017-07-25 | 华南理工大学 | The first visual angle dynamic gesture identification method based on depth convolutional neural networks framework |
| CN106934418B (en) * | 2017-03-09 | 2021-05-14 | 国家电网公司 | Insulator infrared diagnosis method based on convolution recursive network |
| CN106874890A (en) * | 2017-03-16 | 2017-06-20 | 天津大学 | A kind of method of insulator missing in identification transmission line of electricity based on Aerial Images |
| CN107145846B (en) * | 2017-04-26 | 2018-10-19 | 贵州电网有限责任公司输电运行检修分公司 | A kind of insulator recognition methods based on deep learning |
| CN107133943B (en) * | 2017-04-26 | 2018-07-06 | 贵州电网有限责任公司输电运行检修分公司 | A kind of visible detection method of stockbridge damper defects detection |
| CN107230205A (en) * | 2017-05-27 | 2017-10-03 | 国网上海市电力公司 | A kind of transmission line of electricity bolt detection method based on convolutional neural networks |
| CN108496188A (en) * | 2017-05-31 | 2018-09-04 | 深圳市大疆创新科技有限公司 | Method, apparatus, computer system and the movable equipment of neural metwork training |
| CN107316004A (en) * | 2017-06-06 | 2017-11-03 | 西北工业大学 | Space Target Recognition based on deep learning |
| CN107479368B (en) | 2017-06-30 | 2021-09-21 | 北京百度网讯科技有限公司 | Method and system for training unmanned aerial vehicle control model based on artificial intelligence |
| CN107369291A (en) * | 2017-07-13 | 2017-11-21 | 南京理工大学 | The anti-external force damage alarm system and method for high-tension line based on deep learning |
| CN107292885A (en) * | 2017-08-08 | 2017-10-24 | 广东工业大学 | A kind of product defects classifying identification method and device based on autocoder |
| CN107729808B (en) * | 2017-09-08 | 2020-05-01 | 国网山东省电力公司电力科学研究院 | Intelligent image acquisition system and method for unmanned aerial vehicle inspection of power transmission line |
| CN108037133B (en) * | 2017-12-27 | 2021-01-26 | 武汉市智勤创亿信息技术股份有限公司 | Intelligent electric power equipment defect identification method and system based on unmanned aerial vehicle inspection image |
| CN108229587B (en) * | 2018-02-06 | 2021-04-06 | 贵州电网有限责任公司 | Autonomous transmission tower scanning method based on hovering state of aircraft |
| CN108648169B (en) * | 2018-03-20 | 2020-11-10 | 中国科学院自动化研究所 | Method and device for automatically identifying defects of high-voltage power transmission tower insulator |
| CN108881841A (en) * | 2018-07-03 | 2018-11-23 | 国网通用航空有限公司 | Power transmission line intelligent identifying system and method based on helicopter platform |
| CN109118479B (en) * | 2018-07-26 | 2022-07-19 | 中睿能源（北京）有限公司 | Capsule network-based insulator defect identification and positioning device and method |
| CN109118456B (en) * | 2018-09-26 | 2021-07-23 | 北京字节跳动网络技术有限公司 | Image processing method and device |
| CN109242801B (en) * | 2018-09-26 | 2021-07-02 | 北京字节跳动网络技术有限公司 | Image processing method and device |
| CN109389160A (en) * | 2018-09-27 | 2019-02-26 | 南京理工大学 | Electric insulation terminal defect inspection method based on deep learning |
| CN109685762A (en) * | 2018-11-09 | 2019-04-26 | 五邑大学 | A kind of Downtilt measurement method based on multiple dimensioned deep semantic segmentation network |
| CN109813276B (en) * | 2018-12-19 | 2021-01-26 | 五邑大学 | Base station antenna downward inclination angle measuring method and system |
| CN110133443B (en) * | 2019-05-31 | 2020-06-16 | 中国科学院自动化研究所 | Power transmission line component detection method, system and device based on parallel vision |
| CN110309865A (en) * | 2019-06-19 | 2019-10-08 | 上海交通大学 | A kind of unmanned plane patrolling power transmission lines pin defect system image-recognizing method |
| CN110503029A (en) * | 2019-08-21 | 2019-11-26 | 云南电网有限责任公司电力科学研究院 | A kind of detection method of transmission line of electricity glass insulator state |
| CN110367239A (en) * | 2019-08-26 | 2019-10-25 | 华北电力大学（保定） | A kind of bird trouble on transmission line control method based on Internet of Things |
| CN110688925B (en) * | 2019-09-19 | 2022-11-15 | 国网智能科技股份有限公司 | Cascade target identification method and system based on deep learning |
| CN110866548A (en) * | 2019-10-31 | 2020-03-06 | 国网江苏省电力有限公司电力科学研究院 | Infrared intelligent matching identification and distance measurement positioning method and system for insulator of power transmission line |
| CN110956176B (en) * | 2019-11-14 | 2023-06-16 | 国网山西省电力公司大同供电公司 | Method and detection device for fitting transmission line between tower bases |
| CN110794861A (en) * | 2019-11-14 | 2020-02-14 | 国网山东省电力公司电力科学研究院 | Autonomous string falling method and system for flying on-line and off-line insulator string detection robot |
| CN111507958B (en) * | 2020-04-15 | 2023-05-26 | 全球能源互联网研究院有限公司 | Target detection method, training method of detection model and electronic equipment |
| CN111598872A (en) * | 2020-05-15 | 2020-08-28 | 山东万腾智能科技有限公司 | Insulator crack detection method and system |
| WO2021253247A1 (en) * | 2020-06-16 | 2021-12-23 | 深圳市大疆创新科技有限公司 | Inspection method and apparatus for movable platform, and movable platform and storage medium |
| CN112200178B (en) * | 2020-09-01 | 2022-10-11 | 广西大学 | Transformer substation insulator infrared image detection method based on artificial intelligence |
| CN113487541B (en) * | 2021-06-15 | 2022-05-03 | 三峡大学 | Insulator detection method and device |
| CN114359541B (en) * | 2021-11-24 | 2024-09-27 | 上海电力大学 | Fault insulator detection method based on candidate frame generation and attention coordination mechanism |
| CN114359156A (en) * | 2021-12-07 | 2022-04-15 | 国网湖北省电力有限公司宜昌供电公司 | Insulator string-dropping fault detection method based on image recognition |
| CN115061490B (en) * | 2022-05-30 | 2024-04-05 | 广州中科云图智能科技有限公司 | Unmanned aerial vehicle-based reservoir inspection method, unmanned aerial vehicle-based reservoir inspection device, unmanned aerial vehicle-based reservoir inspection equipment and storage medium |
| CN115588139B (en) * | 2022-11-22 | 2023-02-28 | 东北电力大学 | Power grid safety intelligent cruise detection method |
| CN116843909B (en) * | 2023-05-12 | 2024-03-08 | 国家电网有限公司华东分部 | Power line extraction method and device, storage medium and computer equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7149353B2 (en) * | 2003-09-23 | 2006-12-12 | Amazon.Com, Inc. | Method and system for suppression of features in digital images of content |
| CN103440495A (en) * | 2013-07-31 | 2013-12-11 | 华北电力大学（保定） | Method for automatically identifying hydrophobic grades of composite insulators |
| CN103488801A (en) * | 2013-10-17 | 2014-01-01 | 中国电子科技集团公司第五十四研究所 | Geographical information space database-based airport target detection method |
| WO2014060001A1 (en) * | 2012-09-13 | 2014-04-24 | FRENKEL, Christina | Multitransmitter model of the neural network with an internal feedback |
| CN104112263A (en) * | 2014-06-28 | 2014-10-22 | 南京理工大学 | Method for fusing full-color image and multispectral image based on deep neural network |
-
2015
- 2015-06-15 CN CN201510330413.0A patent/CN104978580B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7149353B2 (en) * | 2003-09-23 | 2006-12-12 | Amazon.Com, Inc. | Method and system for suppression of features in digital images of content |
| WO2014060001A1 (en) * | 2012-09-13 | 2014-04-24 | FRENKEL, Christina | Multitransmitter model of the neural network with an internal feedback |
| CN103440495A (en) * | 2013-07-31 | 2013-12-11 | 华北电力大学（保定） | Method for automatically identifying hydrophobic grades of composite insulators |
| CN103488801A (en) * | 2013-10-17 | 2014-01-01 | 中国电子科技集团公司第五十四研究所 | Geographical information space database-based airport target detection method |
| CN104112263A (en) * | 2014-06-28 | 2014-10-22 | 南京理工大学 | Method for fusing full-color image and multispectral image based on deep neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104978580A (en) | 2015-10-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104978580B (en) | A kind of insulator recognition methods for unmanned plane inspection transmission line of electricity | |
| Zhao et al. | Transfer learning with fully pretrained deep convolution networks for land-use classification | |
| CN108549893A (en) | A kind of end-to-end recognition methods of the scene text of arbitrary shape | |
| CN106504233A (en) | Image electric power widget recognition methodss and system are patrolled and examined based on the unmanned plane of Faster R CNN | |
| CN109886225A (en) | A kind of image gesture motion on-line checking and recognition methods based on deep learning | |
| CN108564097A (en) | A kind of multiscale target detection method based on depth convolutional neural networks | |
| CN110348376A (en) | A kind of pedestrian's real-time detection method neural network based | |
| CN107341452A (en) | Human bodys' response method based on quaternary number space-time convolutional neural networks | |
| CN109241913A (en) | In conjunction with the ship detection method and system of conspicuousness detection and deep learning | |
| CN107016405A (en) | A kind of insect image classification method based on classification prediction convolutional neural networks | |
| CN106682569A (en) | Fast traffic signboard recognition method based on convolution neural network | |
| CN108090403A (en) | Face dynamic identification method and system based on 3D convolutional neural network | |
| CN108280397A (en) | Human body image hair detection method based on depth convolutional neural networks | |
| CN110097029B (en) | Identity authentication method based on high way network multi-view gait recognition | |
| CN106991147A (en) | A kind of Plant identification and recognition methods | |
| CN108021947A (en) | A kind of layering extreme learning machine target identification method of view-based access control model | |
| CN108288269A (en) | Bridge pad disease automatic identifying method based on unmanned plane and convolutional neural networks | |
| CN110222760A (en) | A kind of fast image processing method based on winograd algorithm | |
| CN109002752A (en) | A kind of complicated common scene rapid pedestrian detection method based on deep learning | |
| CN104298974A (en) | Human body behavior recognition method based on depth video sequence | |
| CN110163069A (en) | Method for detecting lane lines for assisting driving | |
| CN109325915A (en) | A kind of super resolution ratio reconstruction method for low resolution monitor video | |
| CN108122003A (en) | A kind of Weak target recognition methods based on deep neural network | |
| CN113780132A (en) | Lane line detection method based on convolutional neural network | |
| CN108664994A (en) | A kind of remote sensing image processing model construction system and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: Wang Yue Central Road Ji'nan City, Shandong province 250002 City No. 2000 Co-patentee after: National Network Intelligent Technology Co., Ltd. Patentee after: Electric Power Research Institute of State Grid Shandong Electric Power Company Co-patentee after: State Grid Corporation of China Address before: Wang Yue Central Road Ji'nan City, Shandong province 250002 City No. 2000 Co-patentee before: Shandong Luneng Intelligent Technology Co., Ltd. Patentee before: Electric Power Research Institute of State Grid Shandong Electric Power Company Co-patentee before: State Grid Corporation of China |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20201030 Address after: 250101 Electric Power Intelligent Robot Production Project 101 in Jinan City, Shandong Province, South of Feiyue Avenue and East of No. 26 Road (ICT Industrial Park) Patentee after: National Network Intelligent Technology Co.,Ltd. Address before: Wang Yue Central Road Ji'nan City, Shandong province 250002 City No. 2000 Patentee before: ELECTRIC POWER RESEARCH INSTITUTE OF STATE GRID SHANDONG ELECTRIC POWER Co. Patentee before: National Network Intelligent Technology Co.,Ltd. Patentee before: STATE GRID CORPORATION OF CHINA |