CN114418963B - Battery plate defect detection method based on machine vision - Google Patents
Battery plate defect detection method based on machine vision Download PDFInfo
- Publication number
- CN114418963B CN114418963B CN202111632245.2A CN202111632245A CN114418963B CN 114418963 B CN114418963 B CN 114418963B CN 202111632245 A CN202111632245 A CN 202111632245A CN 114418963 B CN114418963 B CN 114418963B
- Authority
- CN
- China
- Prior art keywords
- pictures
- layer
- machine
- extreme learning
- groups
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 14
- 230000007547 defect Effects 0.000 title claims abstract description 11
- 230000006870 function Effects 0.000 claims abstract description 32
- 230000004913 activation Effects 0.000 claims abstract description 15
- 238000012706 support-vector machine Methods 0.000 claims abstract description 15
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 14
- 238000000605 extraction Methods 0.000 claims abstract description 6
- 238000000034 method Methods 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 12
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 6
- 238000004519 manufacturing process Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 5
- 230000002950 deficient Effects 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 238000013135 deep learning Methods 0.000 abstract description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 230000009286 beneficial 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
- 238000005070 sampling Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2431—Multiple classes
-
- 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
- G06N3/048—Activation functions
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/048—Fuzzy inferencing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/002—Image coding using neural networks
Abstract
The invention relates to the field of vision detection and discloses a battery polar plate defect detection method based on machine vision, which comprises the steps of firstly carrying out first derivative sharpening treatment on polar plate images to obtain feature extraction pictures; dividing the feature extraction pictures into a plurality of groups, inputting the groups into an extreme learning machine-automatic encoder for encoding, and outputting the encoded pictures; inputting the pictures after multi-channel coding into a full-connection layer, and summarizing the pictures into an atlas after the full-connection layer is activated by using an activation function and inputting the atlas into a fuzzy support vector machine; classifying by using a fuzzy support vector machine, determining classification conditions by a plurality of membership degrees, and dividing the pictures into a plurality of different categories to obtain a plurality of groups of pictures; and respectively uploading a plurality of groups of pictures to a total picture library to form an extreme learning machine-self-coding convolutional neural network, and applying the extreme learning machine-self-coding convolutional neural network to check the training correctness. The invention uses the extreme learning machine-automatic coding convolutional neural network to carry out deep learning, can greatly improve the recognition accuracy and can greatly reduce the polar plate defect problem.
Description
Technical Field
The invention belongs to the field of visual detection, and particularly relates to a polar plate defect detection method based on machine vision.
Background
The battery has wide application, the quality of the battery can greatly influence the normal use of various devices, and the quality of an electrode plate in the battery also seriously influences the service life of the battery and the stability of the battery. Therefore, the battery plate needs to be inspected in the production process, the traditional inspection of the battery plate is often manual sampling inspection, and the battery plate cannot be guaranteed to be qualified. According to the invention, the polar plates on the production line are inspected one by utilizing the high-definition camera, so that the quality of the polar plates in the production process of the battery polar plates is ensured.
Disclosure of Invention
The invention aims to: aiming at the problems in the prior art, the invention provides a safe and efficient polar plate detection method, which effectively solves the problems of low efficiency of the current manual spot check and instability of the manual spot check, extracts an image of a photographed polar plate by carrying a high-definition camera, and can effectively solve the problems of polar plate detection by deep learning of an extreme learning machine-self-coding neural network algorithm.
The technical content is as follows: the invention provides a battery plate defect detection method based on machine vision, which comprises the following steps:
step 1: shooting a polar plate in a production line, performing image processing on the shot picture, extracting the characteristics of the picture through first derivative sharpening processing to obtain a characteristic extraction picture, and numbering the polar plate;
step 2: dividing the characteristic extraction pictures into a plurality of groups, inputting the groups into an extreme learning machine-automatic encoder for encoding, and outputting the encoded pictures;
step 3: inputting the pictures after multi-channel coding into a full-connection layer, and summarizing the pictures into an atlas after the full-connection layer is activated by using an activation function and inputting the atlas into a fuzzy support vector machine;
step 4: classifying by using a fuzzy support vector machine, determining classification conditions by a plurality of membership degrees, and dividing the pictures into a plurality of different categories to obtain a plurality of groups of pictures;
step 5: and respectively uploading a plurality of groups of pictures to a total picture library, reserving the pictures to form an extreme learning machine-self-coding convolutional neural network, and applying the extreme learning machine-self-coding convolutional neural network to check the training correctness.
Further, the specific operation of performing the first derivative sharpening processing on the multiple pictures in the step 1 is as follows:
step 1.1: the gradient of the image f at coordinates (x, y) is defined as a two-dimensional vector:
the vector represents the maximum rate of change direction of the pixel point pointing to f as an important geometrical property at the position (x, y);
step 1.2: a gradient pattern is built up, expressed as M (x, y) with the magnitude of v f being the vector:
wherein f is a norm, indicating that the rate of change in the gradient vector direction is in (x, y), the size of M (x, y) is the same as the original image size;
step 1.3: instead of the sum of squares operation, absolute value operations are used:
M(x,y)≈|g x |+|g y |。
further, the structure of the extreme learning machine-automatic encoder in the step 2 is as follows:
the extreme learning machine-automatic encoder consists of a convolution layer and a pooling layer, wherein the convolution layer and the pooling layer consist of an input layer, an output layer and a hidden layer, the number of neurons of the input layer and the output layer is the same, the network layer is of a forward propagation loop-free structure, the input layer to the hidden layer are the encoding process of data, the hidden layer to the output layer are the decoding process of data, and the operation process of the extreme learning machine-automatic encoder is as follows:
wherein x is input, sigma a Sigma is an activation function, W a 、W s Weight, b a 、b s To bias R v 、R u Respectively an input set and an output set, which are convolution operations;
combining an extreme learning machine into a self-encoder, wherein the extreme learning machine has the same structure as the self-encoder, and the hidden layer input weight and bias need to be orthogonalized, namely:
W T W＝I,b T b＝I
extreme learning machine-self encoder combined hidden layer output H and reconstructed samplesThe relation of (2) is:
the solution method of the output weight beta of the extreme learning machine-self encoder is related to the number of network nodes, and when the number N of the input nodes is different from the number L of hidden layer nodes, the beta calculation formula is as follows:
when the node numbers are the same, the beta calculation formula is: beta=h -1 X。
Further, the full-connection layer in the step 3 includes a weight vector and an activation function, where the weight vector is a proportion of the multipath feature weighted residuals, and the activation function is f (x) =max (0:x).
Further, the specific operation of the fuzzy support vector machine in the step 4 is as follows:
step 4.1: introducing characteristics of samples, universal identifiers and membership degree of each sample, and setting training sets of each sample as follows: { (x) 1 ,y 1 ,μ 1 (x 1 )),(x 2 ,y 2 ,μ 2 (x 2 )),...,(x n ,y n ,μ n (x n ) -x) each training feature represented as x i ∈R n The home class identifier is y i ∈(-1,1)，ξ i To support the classification error term of the vector machine objective function, μ (x i )ξ i Is a weighted error term;
step 4.2: the objective function of the fuzzy support vector machine is determined as follows:
s.t.y i (ωx i +b)-1+ξ i ≥0,
ξ i ≥0,i＝1,2,...,n
wherein ω is a linear piecewise function, C is a penalty factor, and the corresponding discriminant function is:
step 4.3: calculating a membership function according to the distance from the sample to the class center, wherein the membership function is as follows:
wherein,is of class center, x i For the sample, r is the class radius and ε is a small positive number preset.
Further, the fuzzy vector machine selects three membership degrees to determine classification conditions, the pictures are divided into 3 groups of different categories, a plurality of groups of pictures are obtained, and the 3 groups of categories are respectively: defect free images, uncertain images, defective images.
The beneficial effects are that:
1. the invention can effectively reduce the complexity of the convolutional neural network on the picture data processing by utilizing the first derivative sharpening process, the first derivative sharpening process extracts the picture characteristics, reduces the interference of non-characteristic parts, effectively improves the recognition accuracy and reduces the training time.
2. Compared with the traditional self-encoder, the extreme learning machine-self-encoder does not need to adopt a gradient descent method for iterative fine adjustment, greatly shortens the training time, solves the problem of local optimization, ensures that network parameters do not need iterative fine adjustment, takes a least square solution as a global optimal solution, and ensures the generalization capability of an algorithm.
3. The fuzzy support vector machine is used for replacing the traditional Softmax classifier, so that the problems of overfitting, noise interference and the like of the traditional classifier are solved, and the recognition accuracy of the classifier is improved.
4. In the convolutional neural network, the improved activation function f (x) =max (0:x) is introduced into the full-connection layer, so that the pooled data can be effectively activated, the data is binarized, the operation amount can be greatly reduced, and compared with the activation function in the conventional convolutional algorithm, the method is more effective, the learning of the convolutional neural network is also faster, the learning progress can be greatly accelerated, and the programming quantity is reduced.
Drawings
FIG. 1 is a flow chart of a machine vision based polar plate detection method;
FIG. 2 is a flow chart of an extreme learning machine-self-encoding convolutional neural network;
FIG. 3 is a block diagram of an automatic encoder;
FIG. 4 is a block diagram of a multi-channel extreme learning machine self-encoding convolutional neural network.
Detailed Description
The present invention will be further described with reference to the accompanying drawings, and the following examples are only for more clearly illustrating the technical aspects of the present invention, and are not to be construed as limiting the scope of the present invention.
Referring to fig. 1, the invention discloses a method for detecting battery polar plates based on machine vision, which is based on machine vision, image processing and an extreme learning machine self-coding convolutional neural network.
According to the detection method, a high-definition camera is used for shooting an image of a polar plate at a fixed position on a production line, the shot image is numbered, and first derivative sharpening is carried out.
After the power tower and polar plate pictures are obtained, the image is subjected to first derivative sharpening, and the gradient of the image f at coordinates (x, y) is defined as a two-dimensional vector:
the vector represents that the important geometrical property at position (x, y) is that the pixel point points in the direction of maximum rate of change of f.
To facilitate pixel transformation, a gradient pattern is typically created, i.e. expressed as M (x, y) with the magnitude of vector f, where:
wherein f is a norm, and represents a value of a change rate in the gradient vector direction in (x, y), where M (x, y) is the same as the original image in size, so as to ensure that the image is not distorted.
For simplicity of program operand, and fast processing, absolute value operations are often used instead of sum of squares operations:
M(x,y)≈|g x |+|g y |
referring to fig. 2-4, the extreme learning machine-self-coding convolutional neural network is a main part of the present invention, and mainly comprises an extreme learning machine-self encoder, a full connection layer and a fuzzy support vector machine:
the extreme learning machine-automatic encoder is composed of a convolution layer and a pooling layer, the quantity of neurons of an input layer and an output layer is the same, the network layer is of a forward propagation loop-free structure, the input layer to the hidden layer are the coding process of data, the hidden layer to the output layer are the decoding process of data, and the operation process of the extreme learning machine-automatic encoder is as follows:
wherein x is input, sigma a Sigma is an activation function, W a 、W s Weight, b a 、b s To bias R v 、R u Respectively an input set and an output set, which are convolution operations.
Combining an extreme learning machine into a self-encoder, wherein the extreme learning machine has the same structure as the self-encoder, and the hidden layer input weight and bias are required to be orthogonalized
W T W＝I,b T b＝I
Extreme learning machine-self encoder combined hidden layer output H and reconstructed samplesThe relation of (2) is that
The solution method of the extreme learning machine-self encoder output weight beta is related to the number of network nodes. When the number N of the input nodes is different from the number L of the hidden layer nodes, the beta calculation formula is as follows
When the node numbers are the same, the beta calculation formula is as follows
β＝H -1 X
Compared with the traditional self-encoder, the extreme learning machine-self-encoder does not need to adopt a gradient descent method for iterative fine adjustment, and the training time is greatly shortened.
And activates the data at the input layer to the hidden layer and at the hidden layer to the output layer using an activation function σ=max {0, x }.
The full connection layer comprises a weight vector and an activation function, wherein the weight vector determines the proportion of the multipath characteristic weighted residual error, and in order to ensure the training speed of the deep neural network, the traditional activation function f (x) =tanh (x) andinstead f (x) =max (0:x).
Using a fuzzy support vector machine to replace a classification layer, introducing characteristics of samples, a universal identifier and membership degree of each sample when the fuzzy support vector machine classifies the samples, and setting a training set of each sample as
{(x 1 ,y 1 ,μ 1 (x 1 )),(x 2 ,y 2 ,μ 2 (x 2 )),...,(x n ,y n ,μ n (x n ))}
Each training feature is denoted as x i ∈R n The home class identifier is y i ∈(-1,1)，ξ i To support the classification error term of the vector machine objective function, μ (x i )ξ i For weighted error term, the objective function of the fuzzy support vector machine is
s.t.y i (ωx i +b)-1+ξ i ≥0,
ξ i ≥0,i＝1,2,...,n
Wherein ω is a linear piecewise function, C is a penalty factor, and the corresponding discriminant function is:
and calculating a membership function according to the distance from the sample to the class center as follows:
wherein,is of class center, x i For the sample, r is the class radius and ε is a small positive number preset.
In this embodiment, the fuzzy vector machine selects three membership degrees to determine classification conditions, and classifies the pictures into 3 groups of different categories to obtain multiple groups of pictures, wherein the 3 groups of categories are respectively: defect free images, uncertain images, defective images.
The foregoing embodiments are merely illustrative of the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the present invention and to implement the same, not to limit the scope of the present invention. All equivalent changes or modifications made according to the spirit of the present invention should be included in the scope of the present invention.
Claims (3)
1. The battery plate defect detection method based on machine vision is characterized by comprising the following steps of:
step 1: shooting a polar plate in a production line, performing image processing on the shot picture, extracting the characteristics of the picture through first derivative sharpening processing to obtain a characteristic extraction picture, and numbering the polar plate;
step 2: dividing the characteristic extraction pictures into a plurality of groups, inputting the multiple groups into an extreme learning machine-automatic encoder for encoding, and inputting the encoded pictures;
the extreme learning machine-automatic encoder consists of a convolution layer and a pooling layer, wherein the convolution layer and the pooling layer consist of an input layer, an output layer and a hidden layer, the number of neurons of the input layer and the output layer is the same, the network layer is of a forward propagation loop-free structure, the input layer to the hidden layer are the encoding process of data, the hidden layer to the output layer are the decoding process of data, and the operation process of the extreme learning machine-automatic encoder is as follows:
wherein x is input, sigma a Sigma is an activation function, W a 、W s Weight, b a 、b s To bias R v 、R u Respectively an input set and an output set, which are convolution operations;
combining an extreme learning machine into a self-encoder, wherein the extreme learning machine has the same structure as the self-encoder, and the hidden layer input weight and bias need to be orthogonalized, namely:
W T W＝I,b T b＝I
extreme learning machine-self encoder combined hidden layer output H and reconstructed samplesThe relation of (2) is:
the solution method of the output weight beta of the extreme learning machine-self encoder is related to the number of network nodes, and when the number N of the input nodes is different from the number L of hidden layer nodes, the beta calculation formula is as follows:
when the node numbers are the same, the beta calculation formula is: beta=h -1 X；
Step 3: inputting the pictures after multi-channel coding into a full-connection layer, and summarizing the pictures into an atlas after the full-connection layer is activated by using an activation function and inputting the atlas into a fuzzy support vector machine;
step 4: classifying by using a fuzzy support vector machine, determining classification conditions by a plurality of membership degrees, and dividing the pictures into a plurality of different categories to obtain a plurality of groups of pictures;
step 4.1: introducing characteristics of a sample, a generic identifier, and eachMembership of each sample, training set of each sample is set as follows: { (x) 1 ,y 1 ,μ 1 (x 1 )),(x 2 ,y 2 ,μ 2 (x 2 )),…,(x n ,y n ,μ n (x n ) -x) each training feature represented as x i ∈R n The home class identifier is y i ∈(-1,1)，ξ i To support the classification error term of the vector machine objective function, μ (x i )ξ i Is a weighted error term;
step 4.2: the objective function of the fuzzy support vector machine is determined as follows:
s.t.y i (ωx i +b)-1+ξ i ≥0,
ξ i ≥0,i＝1,2,...,n
wherein ω is a linear piecewise function, C is a penalty factor, and the corresponding discriminant function is:
step 4.3: calculating a membership function according to the distance from the sample to the class center, wherein the membership function is as follows:
wherein,is of class center, x i The method is characterized in that the method is used for a sample, r is a class radius, and epsilon is a very small positive number preset;
the fuzzy support vector machine selects three membership degrees to determine classification conditions, the pictures are divided into 3 groups of different categories, a plurality of groups of pictures are obtained, and the 3 groups of categories are respectively: a defect-free image, an uncertain image, a defective image;
step 5: and respectively uploading a plurality of groups of pictures to a total picture library, reserving the pictures to form an extreme learning machine-self-coding convolutional neural network, and applying the extreme learning machine-self-coding convolutional neural network to check the training correctness.
2. The machine vision-based battery plate defect detection method according to claim 1, wherein the specific operation of performing the first derivative sharpening process on the multiple pictures in step 1 is as follows:
step 1.1: the gradient of the image f at coordinates (x, y) is defined as a two-dimensional vector:
the vector represents the maximum rate of change direction of the pixel point pointing to f as an important geometrical property at the position (x, y);
step 1.2: a gradient pattern is built up, expressed as M (x, y) with the magnitude of v f being the vector:
wherein f is a norm, indicating that the rate of change in the gradient vector direction is in (x, y), the size of M (x, y) is the same as the original image size;
step 1.3: instead of the sum of squares operation, absolute value operations are used:
M(x,y)≈|g x |+|g y |。
3. the machine vision-based battery plate defect detection method according to claim 1, wherein the full connection layer in the step 3 includes a weight vector and an activation function, the weight vector is a proportion of multi-path feature weighted residuals, and the activation function is f (x) =max (0:x).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111632245.2A CN114418963B (en) | 2021-12-28 | 2021-12-28 | Battery plate defect detection method based on machine vision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111632245.2A CN114418963B (en) | 2021-12-28 | 2021-12-28 | Battery plate defect detection method based on machine vision |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114418963A CN114418963A (en) | 2022-04-29 |
| CN114418963B true CN114418963B (en) | 2023-12-05 |
Family
ID=81269202
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111632245.2A Active CN114418963B (en) | 2021-12-28 | 2021-12-28 | Battery plate defect detection method based on machine vision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114418963B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116125803A (en) * | 2022-12-28 | 2023-05-16 | 淮阴工学院 | Inverter backstepping fuzzy neural network control strategy based on extreme learning machine |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111598861A (en) * | 2020-05-13 | 2020-08-28 | 河北工业大学 | Improved Faster R-CNN model-based non-uniform texture small defect detection method |
| KR20200103150A (en) * | 2019-02-11 | 2020-09-02 | (주) 인텍플러스 | Visual inspection system |
| CN112270659A (en) * | 2020-08-31 | 2021-01-26 | 中国科学院合肥物质科学研究院 | Rapid detection method and system for surface defects of pole piece of power battery |
| CN113592845A (en) * | 2021-08-10 | 2021-11-02 | 深圳市华汉伟业科技有限公司 | Defect detection method and device for battery coating and storage medium |
| CN113658176A (en) * | 2021-09-07 | 2021-11-16 | 重庆科技学院 | Ceramic tile surface defect detection method based on interactive attention and convolutional neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109239075B (en) * | 2018-08-27 | 2021-11-30 | 北京百度网讯科技有限公司 | Battery detection method and device |
-
2021
- 2021-12-28 CN CN202111632245.2A patent/CN114418963B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200103150A (en) * | 2019-02-11 | 2020-09-02 | (주) 인텍플러스 | Visual inspection system |
| CN111598861A (en) * | 2020-05-13 | 2020-08-28 | 河北工业大学 | Improved Faster R-CNN model-based non-uniform texture small defect detection method |
| CN112270659A (en) * | 2020-08-31 | 2021-01-26 | 中国科学院合肥物质科学研究院 | Rapid detection method and system for surface defects of pole piece of power battery |
| CN113592845A (en) * | 2021-08-10 | 2021-11-02 | 深圳市华汉伟业科技有限公司 | Defect detection method and device for battery coating and storage medium |
| CN113658176A (en) * | 2021-09-07 | 2021-11-16 | 重庆科技学院 | Ceramic tile surface defect detection method based on interactive attention and convolutional neural network |
Non-Patent Citations (2)
| Title |
|---|
| Technical data-driven tool condition monitoring challenges for CNC milling: a review;Shi Yuen Wong et.al;《The International Journal of Advanced Manufacturing Technology volume》;第4837-4857页 * |
| 基于机器视觉的锂电池极片缺陷检测与分类系统;王露;《万方学位论文》;第1-64页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114418963A (en) | 2022-04-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110543878B (en) | Pointer instrument reading identification method based on neural network | |
| CN103927534B (en) | A kind of online visible detection method of coding character based on convolutional neural networks | |
| Rijal et al. | Ensemble of deep neural networks for estimating particulate matter from images | |
| CN109272500B (en) | Fabric classification method based on adaptive convolutional neural network | |
| CN111833352B (en) | Image segmentation method for improving U-net network based on octave convolution | |
| CN110827260B (en) | Cloth defect classification method based on LBP characteristics and convolutional neural network | |
| CN109829414B (en) | Pedestrian re-identification method based on label uncertainty and human body component model | |
| CN109344856B (en) | Offline signature identification method based on multilayer discriminant feature learning | |
| CN110543906B (en) | Automatic skin recognition method based on Mask R-CNN model | |
| CN114418963B (en) | Battery plate defect detection method based on machine vision | |
| CN111145145B (en) | Image surface defect detection method based on MobileNet | |
| CN112164033A (en) | Abnormal feature editing-based method for detecting surface defects of counternetwork texture | |
| CN114049305A (en) | Distribution line pin defect detection method based on improved ALI and fast-RCNN | |
| CN110555461A (en) | scene classification method and system based on multi-structure convolutional neural network feature fusion | |
| CN110348503A (en) | A kind of apple quality detection method based on convolutional neural networks | |
| CN111445484B (en) | Image-level labeling-based industrial image abnormal area pixel level segmentation method | |
| CN112132257A (en) | Neural network model training method based on pyramid pooling and long-term memory structure | |
| Zhou et al. | Defect detection method based on knowledge distillation | |
| CN107529647B (en) | Cloud picture cloud amount calculation method based on multilayer unsupervised sparse learning network | |
| CN110991247B (en) | Electronic component identification method based on deep learning and NCA fusion | |
| CN112634171A (en) | Image defogging method based on Bayes convolutional neural network and storage medium | |
| CN115761416A (en) | CS-YOLOv5s network-based insulator defect detection method | |
| CN114445726B (en) | Sample library establishing method and device based on deep learning | |
| CN114782322A (en) | YOLOv5 model arc additive manufacturing molten pool defect detection method | |
| CN116029957A (en) | Insulator image pollution identification method based on Markov chain Monte Carlo |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |