CN115239717A - Defect detection device in industrial detection - Google Patents
Defect detection device in industrial detection Download PDFInfo
- Publication number
- CN115239717A CN115239717A CN202211154478.0A CN202211154478A CN115239717A CN 115239717 A CN115239717 A CN 115239717A CN 202211154478 A CN202211154478 A CN 202211154478A CN 115239717 A CN115239717 A CN 115239717A
- Authority
- CN
- China
- Prior art keywords
- tensor
- encoder
- channel
- discriminator
- channel tensor
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 60
- 230000007547 defect Effects 0.000 title claims abstract description 45
- 238000012549 training Methods 0.000 claims abstract description 36
- 238000000605 extraction Methods 0.000 claims abstract description 18
- 238000005070 sampling Methods 0.000 claims abstract description 16
- 238000007689 inspection Methods 0.000 claims description 15
- 238000000034 method Methods 0.000 description 25
- 230000008569 process Effects 0.000 description 13
- 238000013527 convolutional neural network Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 6
- 230000002159 abnormal effect Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 230000014759 maintenance of location Effects 0.000 description 4
- 239000000463 material Substances 0.000 description 4
- 238000013459 approach Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000000717 retained effect Effects 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009977 dual effect Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000003042 antagnostic effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000037361 pathway Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4007—Scaling of whole images or parts thereof, e.g. expanding or contracting based on interpolation, e.g. bilinear interpolation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4046—Scaling of whole images or parts thereof, e.g. expanding or contracting using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of industrial detection, and provides a defect detection device in industrial detection, which comprises: an extractor for obtaining a multi-channel tensor; generating an confrontation model, wherein the confrontation model comprises an encoder, a generator and a discriminator, the encoder is used for generating a single-channel tensor, the generator is used for generating a multi-dimensional feature tensor according to the multi-channel tensor, the discriminator is used for discriminating the multi-channel tensor and the single-channel tensor to generate a first discrimination result, and the multi-dimensional feature tensor and the single-channel tensor are discriminated to generate a second discrimination result, so that the generated confrontation model is subjected to model training according to the first discrimination result and the second discrimination result; and the discriminator is also used for discriminating according to the multi-channel tensor and the multi-dimensional feature tensor after the model training is finished and generating a third discrimination result. The invention effectively sums the advantages of the extractor and the encoder, realizes double promotion of feature extraction and sampling rate, and further improves the accuracy of defect detection.
Description
Technical Field
The invention relates to the technical field of industrial detection, in particular to a defect detection device and a defect detection method in industrial detection.
Background
When the defects of industrial products are detected, the commonly used artificial neural network model can only effectively identify the types of the trained defects. In other words, if a defect does not exist in the training dataset or the probability of the defect is very low, the trained model cannot effectively identify the defect. In industrial production, new or rare defects do occur due to instability of the production line and equipment. As described above, since such defects cannot be effectively trained, the artificial neural network recognition effect is far lower than expected.
At present, the anomaly detection research aiming at industrial quality inspection is rare, and the anomaly detection research facing to an open data set is relatively common. The algorithms published at present basically follow a two-step framework of sampling-discrimination. In this framework, the related pathways can be divided into two categories: firstly, sampling an image by using a convolutional neural network, and then performing two-class discrimination on the acquired image characteristic data by using a machine learning algorithm to judge a normal or abnormal image; second, a GAN (generic adaptive Networks) model is used to provide a capability of generating a good-quality image, and the generated image and the input image are determined by a discriminator in the model to predict whether the input image is abnormal.
When experiments are carried out on public data sets, the two approaches show certain abnormality detection capability, but have many problems. In the first approach, although normal data is required for training the sampled model, the precision of performing the second classification on the sampled feature data is affected by the existing abnormal data amount, normal data amount and the proportion thereof, and the performance on novel abnormal defects is poor. The second approach usually requires a plurality of neural network models to be trained simultaneously due to the adopted GAN, and this training method requires a large amount of data to be supplied on one hand, and on the other hand, the conditions of 'pattern collapse' and the like which directly affect the training quality of the generator may occur during training.
Disclosure of Invention
In order to solve the above technical problems, a first objective of the present invention is to provide a defect detection apparatus in industrial detection, which effectively sums the advantages of an extractor and an encoder, realizes dual promotion of feature extraction and sampling rate, and further improves the accuracy of defect detection.
The second purpose of the invention is to provide a defect detection method in industrial detection.
The technical scheme adopted by the invention is as follows:
an embodiment of the first aspect of the present invention provides a defect detection apparatus in industrial inspection, including: the extractor is used for receiving an industrial detection image and down-sampling the industrial detection image to obtain a multi-channel tensor; generating an countermeasure model, wherein the generated countermeasure model comprises an encoder, a generator and a discriminator, an input end of the encoder is connected with an output end of the extractor and an input end of the discriminator, an output end of the encoder is respectively connected with an input end of the generator and an input end of the discriminator, the encoder is used for performing feature extraction on the multi-channel tensor to generate a single-channel tensor, the generator is used for generating a multi-dimensional feature tensor according to the multi-channel tensor, the multi-dimensional feature tensor is the same as the multi-channel tensor in size, the discriminator is used for discriminating the multi-channel tensor and the single-channel tensor to generate a first discrimination result, the first discrimination result is fed back to the encoder, the multi-dimensional feature tensor and the single-channel tensor are discriminated to generate a second discrimination result, and the second discrimination result is fed back to the generator, so that the generated countermeasure model is trained according to the first discrimination result and the second discrimination result; and the discriminator is also used for discriminating according to the multi-channel tensor and the multi-dimensional feature tensor after model training is finished and generating a third discrimination result so as to detect defects according to the third discrimination result.
The defect detection device in industrial detection provided by the invention can also have the following additional technical characteristics:
according to an embodiment of the invention, the extractor comprises a CNN (Convolutional Neural Network) implementation.
According to an embodiment of the present invention, the encoder is a transform (a method for implementing data feature extraction without using a convolutional neural network) encoder.
According to one embodiment of the invention, when the encoder, the generator and the discriminator are subjected to model training, the industrial detection image is a good-quality image.
An embodiment of the second aspect of the present invention provides a defect detection method in industrial inspection, including the following steps: receiving an industrial detection image and performing down-sampling on the industrial detection image to obtain a multi-channel tensor; performing feature extraction on the multi-channel tensor to obtain a single-channel tensor; generating a multi-dimensional feature tensor according to the multi-channel tensor, wherein the multi-dimensional feature tensor is the same as the multi-channel tensor in size; judging the multi-channel tensor and the single-channel tensor to generate a first judgment result; judging the multi-dimensional feature tensor and the single-channel tensor to generate a second judgment result; performing model training according to the first judgment result and the second judgment result; and after model training is finished, judging according to the multi-channel tensor and the multi-dimensional feature tensor to generate a third judgment result, and detecting defects according to the third judgment result.
The defect detection method in industrial detection provided by the invention can also have the following additional technical characteristics:
according to one embodiment of the invention, the industrial detection image is a good-quality image during model training.
The invention has the beneficial effects that:
the invention firstly utilizes the extractor to carry out incomplete feature extraction, the extracted result is trained by a generating network consisting of the encoder, the generator and the discriminator, the encoder and the extractor are separated, the extractor does not participate in the training process of generating the network and only serves as an image scaling tool, the retention rate of the features in the process of reducing the size of the graph is effectively improved, the output multi-channel tensor is used as input data of the encoder, the encoder participates in the training process and continuously optimizes the super parameters of the encoder, the features of the original image can be maximally retained in sampling, the features entering the discriminator and the generator can better reflect the original image and the problem of mode collapse of the generator can be solved, therefore, the advantages of the extractor and the encoder are effectively integrated, the dual improvement of feature extraction and sampling rate is realized, and the accuracy of defect detection is further improved.
The CNN is adopted to replace the linear scaling of the graph, so that the retention rate of the features in the process of reducing the size of the graph can be effectively improved.
By using the Transformer as an encoder, the efficiency and the precision of a feature extraction stage can be obviously improved, and the efficiency and the precision optimization of the integral model reasoning can be positively influenced.
Drawings
FIG. 1 is a schematic configuration diagram of a defect detecting apparatus in industrial inspection according to a first embodiment of the present invention;
FIG. 2 is a schematic structural diagram of a defect detecting apparatus in industrial inspection according to a second embodiment of the present invention;
FIG. 3 is a schematic diagram of a Transformer according to one embodiment of the invention;
FIG. 4 is a flow diagram of a method of defect detection in industrial inspection according to one embodiment of the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a schematic configuration diagram of a defect detection apparatus in industrial inspection according to an embodiment of the present invention, and fig. 2 is a schematic configuration diagram of a defect detection apparatus in industrial inspection according to a second embodiment of the present invention. As shown in fig. 1-2, the defect detecting apparatus includes: and the extractor T generates a confrontation model, and the generation of the confrontation model comprises an encoder E, a generator G and a discriminator D.
The extractor T is used for receiving the industrial detection image and down-sampling the industrial detection image to obtain a multi-channel tensor; the input of encoder E links to each other with the output of extractor T and arbiter D's input, the output of encoder E links to each other with generator G's input and arbiter D's input respectively, encoder E is used for carrying out the feature extraction to the multichannel tensor, in order to generate the single channel tensor, generator G is used for generating multidimensional feature tensor according to the multichannel tensor, and multidimensional feature tensor is the same with the size of multichannel tensor, as shown in FIG. 1, arbiter D is used for distinguishing multichannel tensor and single channel tensor in order to generate first discrimination
And the first discrimination result is used
Fed back to the encoder E to discriminate the multi-dimensional feature tensor and the single-channel tensor to generate a second discrimination result
And the second judgment result is compared with the first judgment result
Feeding back to the generator G to generate a confrontation model according to the first discrimination result
And a second discrimination result
Carrying out model training; as shown in fig. 2, the discriminator D is further configured to discriminate according to the multi-channel tensor and the multi-dimensional feature tensor and generate a third discrimination result after the model training is completed, so as to discriminate according to the third discrimination result
And carrying out defect detection.
Further, in an embodiment of the present invention, the extractor is implemented by CNN. The encoder is a transform encoder. And when the encoder, the generator and the discriminator are subjected to model training, the industrial detection image is a good-quality image.
Specifically, as shown in fig. 1, the original image (industrial inspection image) is represented as
Will be
By means of extractors
Down-sampling to obtain
Is shown as
. The extractor T may adopt an incomplete CNN classifier structure, which may implement incomplete extraction of image features, and the result obtained after extraction should be a small-sized multi-channel tensor, which can store more defect information when processing a defective industrial detection image, compared with linear scaling of an original image. The extractor T may be trained or trained separately in advance.
Scaled multi-channel tensor
Inputting the single channel tensor to a Transformer coder E to obtain a single channel tensor,is shown in FIG. 1 as
. In deep learning based computer vision, as shown in fig. 2, a standard Transformer is divided into three parts: graph partitioning flattening, a transform encoder, and a multi-layer perceptron, the encoder E in this application employs the transform encoder in fig. 2. Because the image after the zooming of the extractor T is already a multi-channel small-size tensor, which is similar to the output structure after the graph segmentation and flattening processing in FIG. 2, meanwhile, the invention also adopts a discriminator D to replace the function of the multi-layer perceptron in FIG. 2. Therefore, the invention adopts a Transformer encoder to realize more efficient feature extraction. The output of the Transformer encoder E is
，
Is a single-channel tensor that contains the feature data of the input image and may contain more precise defect information than the result after CNN sampling.
Output of the encoder E
Also as an input value to the generator G, expressed as
I.e. the expression of the single-channel tensor of the input generator G is
Generator G is based on
The expression of the generated multi-dimensional feature tensor is
，
Should be in accordance with the multi-channel tensor
The same is true. In the invention, the generator G does not generate a three-channel or single-channel image, but generates an AND
A corresponding multi-dimensional feature tensor.
The discriminator D carries out two times of classification discrimination in the training process, and the input of the encoder is discriminated for the first time
And output
Is shown as
Feeding back the judgment result to the encoder; input to the second decision generator G
And output
Is shown as
And the judged result is fed back to the generator G.
That is, the whole work flow of the defect detecting apparatus is divided into a training phase and an inference phase (detection phase), as shown in fig. 1, in the training phase, only the encoder E, the generator G and the discriminator D participate in the training loop, the extractor T does not participate in the training loop, and in the training process, the discriminator performs discrimination twice, that is, the first discrimination result is the above-mentioned first discrimination result
And a second discrimination result
And carrying out parameter optimization according to the judgment result to further complete model training, and in the training process, adopting a good product image for the industrial detection image so that the generator G can generate the good product image after inputting any tensor, and the discriminator D can better identify abnormity.
After the training is completed, as shown in fig. 2, in the inference stage, the industrial detection image of the industrial site is input into the extractor T, and the extractor T generates the corresponding multi-channel tensor
Inputting X into a discriminator D while performing multi-channel tensor
Inputting the single-channel tensor into the encoder E, generating a corresponding single-channel tensor, inputting the single-channel tensor into the generator G, and generating a corresponding multi-dimensional feature tensor by the generator G
Will be
Input to a discriminator D based on the multi-channel tensor
And multidimensional feature tensor
Generating a third discrimination result
Comparing the difference between the input numerical value and the generated numerical value, and judging that the input image is a good product image if the input numerical value and the generated numerical value are close to each other; if not, the image is judged to be abnormal. Thus, finish outAnd (4) detecting the sink.
In short, the training stage is based on the first discrimination result
And a second discrimination result
Training the generation antagonistic model according to a third discrimination result in the inference stage
And (5) detecting.
In summary, according to the defect detection apparatus in industrial detection in the embodiment of the present invention, the extractor is first used to perform incomplete feature extraction, the extracted result is trained by the generation network formed by the encoder, the generator and the discriminator, the encoder and the extractor are separated, the extractor does not participate in the training process of generating the network, and only serves as an image scaling tool, so that the retention rate of features in the process of reducing the graphic size is effectively improved, the output multi-channel tensor is used as the input data of the encoder, the encoder participates in the training process, and continuously optimizes the hyper-parameters of the encoder, so that the features of the original image can be retained to the maximum extent in sampling, the features entering the discriminator and the generator can better reflect the original image, and the problem of "mode collapse" of the generator can be solved, thereby effectively integrating the advantages of the extractor and the encoder, realizing double improvement of feature extraction and sampling rate, and further improving the accuracy of defect detection; the CNN is adopted to replace the linear scaling of the graph, so that the retention rate of the features in the process of reducing the size of the graph can be effectively improved; by using the Transformer as an encoder, the efficiency and the precision of a feature extraction stage can be obviously improved, and the efficiency and the precision optimization of the integral model reasoning can be positively influenced.
Based on the defect detection device in industrial detection, the invention also provides a defect detection method in industrial detection. Since the method embodiment of the present invention is based on the apparatus embodiment, details that are not disclosed in the method embodiment may refer to the method embodiment, and are not described again in the present invention.
Fig. 3 is a flowchart of a defect detection method in industrial inspection according to an embodiment of the present invention, as shown in fig. 3, the method includes the steps of:
s1, receiving an industrial detection image and down-sampling the industrial detection image to obtain a multi-channel tensor.
And S2, performing feature extraction on the multi-channel tensor to obtain a single-channel tensor.
And S3, generating a multi-dimensional feature tensor according to the multi-channel tensor, wherein the multi-dimensional feature tensor and the multi-channel tensor have the same size.
And S4, judging the multi-channel tensor and the single-channel tensor to generate a first judgment result.
And S5, judging the multi-dimensional feature tensor and the single-channel tensor to generate a second judgment result.
And S6, performing model training according to the first judgment result and the second judgment result.
And S7, after the model training is finished, judging according to the multi-channel tensor and the multi-dimensional feature tensor and generating a third judgment result so as to detect the defects according to the third judgment result.
According to one embodiment of the invention, the industrial detection image is a good-quality image when model training is carried out.
According to the defect detection method in industrial detection provided by the embodiment of the invention, training and judgment are carried out according to the multi-channel tensor, the single-channel tensor and the generated multi-dimensional feature tensor which correspond to the industrial detection image, so that the features of the original image can be retained to the maximum extent in sampling, the problem of mode collapse of a generator can be solved, double promotion of feature extraction and sampling rate can be realized, and the accuracy of defect detection is further improved.
In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implying any number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. The meaning of "plurality" is two or more unless specifically limited otherwise.
In the description of the specification, reference to the description of "one embodiment," "some embodiments," "an example," "a specific example," or "some examples" or the like means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Moreover, various embodiments or examples and features of various embodiments or examples described in this specification can be combined and combined by one skilled in the art without being mutually inconsistent. In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing steps of a custom logic function or process, and alternate implementations are included within the scope of the preferred embodiment of the present invention in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present invention.
The logic and/or steps represented in the flowcharts or otherwise described herein, e.g., an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
It should be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, which may be stored in a computer readable storage medium, and when the program is executed, the program includes one or a combination of the steps of the method embodiments.
In addition, functional units in the embodiments of the present invention may be integrated into one processing module, or each unit may exist alone physically, or two or more units are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode. The integrated module, if implemented in the form of a software functional module and sold or used as a stand-alone product, may also be stored in a computer readable storage medium.
The storage medium mentioned above may be a read-only memory, a magnetic or optical disk, etc. Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (4)
1. A defect detection apparatus in industrial inspection, comprising:
an extractor for receiving an industrial inspection image and down-sampling the industrial inspection image to obtain a multi-channel tensor;
generating an countermeasure model, wherein the generated countermeasure model comprises an encoder, a generator and a discriminator, the input end of the encoder is connected with the output end of the extractor and the input end of the discriminator, the output end of the encoder is respectively connected with the input end of the generator and the input end of the discriminator, the encoder is used for performing feature extraction on the multi-channel tensor to generate a single-channel tensor, the generator is used for generating a multi-dimensional feature tensor according to the multi-channel tensor, the multi-dimensional feature tensor is the same as the multi-channel tensor in size, and the discriminator is used for discriminating the multi-channel tensor and the single-channel tensor to generate a first discrimination result and feeding the first discrimination result back to the encoder;
judging the multi-dimensional feature tensor and the single-channel tensor to generate a second judgment result, and feeding the second judgment result back to the generator to enable the generated countermeasure model to carry out model training according to the first judgment result and the second judgment result;
and the discriminator is also used for discriminating according to the multi-channel tensor and the multi-dimensional feature tensor after model training is finished and generating a third discrimination result so as to detect defects according to the third discrimination result.
2. The apparatus for detecting defects in industrial inspections according to claim 1, wherein said extractor is implemented by CNN.
3. The apparatus of claim 1, wherein the encoder is a transform encoder.
4. The apparatus of claim 1, wherein the industrial inspection image is a good image when the encoder, the generator and the discriminator are model-trained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211154478.0A CN115239717B (en) | 2022-09-22 | 2022-09-22 | Defect detection device in industrial detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211154478.0A CN115239717B (en) | 2022-09-22 | 2022-09-22 | Defect detection device in industrial detection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115239717A true CN115239717A (en) | 2022-10-25 |
| CN115239717B CN115239717B (en) | 2023-01-03 |
Family
ID=83667420
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211154478.0A Active CN115239717B (en) | 2022-09-22 | 2022-09-22 | Defect detection device in industrial detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115239717B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112017182A (en) * | 2020-10-22 | 2020-12-01 | 北京中鼎高科自动化技术有限公司 | Industrial-grade intelligent surface defect detection method |
| CN113554630A (en) * | 2021-07-30 | 2021-10-26 | 西安电子科技大学 | Chip surface defect detection method, system, computer device and storage medium |
| CN114266308A (en) * | 2021-12-21 | 2022-04-01 | 浙江网商银行股份有限公司 | Detection model training method and device, and image detection method and device |
| CN114491083A (en) * | 2022-04-01 | 2022-05-13 | 江苏智云天工科技有限公司 | Knowledge graph library construction method and knowledge graph library construction device in industrial detection |
-
2022
- 2022-09-22 CN CN202211154478.0A patent/CN115239717B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112017182A (en) * | 2020-10-22 | 2020-12-01 | 北京中鼎高科自动化技术有限公司 | Industrial-grade intelligent surface defect detection method |
| CN113554630A (en) * | 2021-07-30 | 2021-10-26 | 西安电子科技大学 | Chip surface defect detection method, system, computer device and storage medium |
| CN114266308A (en) * | 2021-12-21 | 2022-04-01 | 浙江网商银行股份有限公司 | Detection model training method and device, and image detection method and device |
| CN114491083A (en) * | 2022-04-01 | 2022-05-13 | 江苏智云天工科技有限公司 | Knowledge graph library construction method and knowledge graph library construction device in industrial detection |
Non-Patent Citations (2)
| Title |
|---|
| GAOWEI ZHANG 等: "Semi-supervised learning with GAN for automatic defect detection from images", 《AUTOMATION IN CONSTRUCTION》 * |
| 视觉菜鸟LEONARDO: "（无监督）基于无监督学习的模板缺陷检测方法研究", 《HTTPS://BLOG.CSDN.NET/QQ_60609496/ARTICLE/DETAILS/126739264》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115239717B (en) | 2023-01-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111931868B (en) | Time series data abnormity detection method and device | |
| CN102473660B (en) | Automatic fault detection and classification in a plasma processing system and methods thereof | |
| US11449711B2 (en) | Machine learning-based defect detection of a specimen | |
| CN114600154B (en) | BBP-assisted defect detection procedure for SEM images | |
| CN110348435B (en) | Target detection method and system based on regional candidate network | |
| JP2022108260A (en) | Device, program, and method for anomaly detection and classification | |
| US20220012866A1 (en) | Cross layer common-unique analysis for nuisance filtering | |
| Zhou | Research on image-based automatic wafer surface defect detection algorithm | |
| Lutz et al. | Evaluation of deep learning for semantic image segmentation in tool condition monitoring | |
| CN113695713A (en) | Online monitoring method and device for welding quality of inner container of water heater | |
| CN117131110B (en) | Method and system for monitoring dielectric loss of capacitive equipment based on correlation analysis | |
| CN112597831A (en) | Signal abnormity detection method based on variational self-encoder and countermeasure network | |
| CN116824256A (en) | Visual inspection system and method for grinding machine | |
| Sharma et al. | A semi-supervised generalized vae framework for abnormality detection using one-class classification | |
| Basima et al. | Enhanced leucocyte classification for leukaemia detection | |
| CN116777917B (en) | Defect detection method and system for optical cable production | |
| CN114020811A (en) | Data anomaly detection method and device and electronic equipment | |
| US11727052B2 (en) | Inspection systems and methods including image retrieval module | |
| CN115239717B (en) | Defect detection device in industrial detection | |
| CN110770753B (en) | Device and method for real-time analysis of high-dimensional data | |
| CN116977708A (en) | Bearing intelligent diagnosis method and system based on self-adaptive aggregation visual view | |
| CN111695582A (en) | Flutter texture detection method and device | |
| US11507616B2 (en) | Inspection systems and methods including image retrieval module | |
| CN112419304B (en) | Multi-stage target detection method and device for one-dimensional data | |
| Wei et al. | Research on tool wear state identification method driven by multi-source information fusion and multi-dimension attention mechanism |
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 |