CN113792506A - MOCVD (Metal organic chemical vapor deposition) intracavity state identification method based on image processing and machine learning - Google Patents
MOCVD (Metal organic chemical vapor deposition) intracavity state identification method based on image processing and machine learning Download PDFInfo
- Publication number
- CN113792506A CN113792506A CN202111133298.XA CN202111133298A CN113792506A CN 113792506 A CN113792506 A CN 113792506A CN 202111133298 A CN202111133298 A CN 202111133298A CN 113792506 A CN113792506 A CN 113792506A
- Authority
- CN
- China
- Prior art keywords
- flow
- training
- mocvd
- image
- image processing
- 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
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000010801 machine learning Methods 0.000 title claims abstract description 18
- 238000005229 chemical vapour deposition Methods 0.000 title claims abstract description 8
- 239000002184 metal Substances 0.000 title claims abstract description 8
- 238000013528 artificial neural network Methods 0.000 claims abstract description 29
- 239000012530 fluid Substances 0.000 claims abstract description 10
- 238000010586 diagram Methods 0.000 claims abstract description 6
- 239000000758 substrate Substances 0.000 claims description 31
- 230000007704 transition Effects 0.000 claims description 9
- 230000009466 transformation Effects 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 6
- 238000003709 image segmentation Methods 0.000 claims description 4
- 230000017105 transposition Effects 0.000 claims description 4
- 239000000463 material Substances 0.000 abstract description 2
- 239000004065 semiconductor Substances 0.000 abstract description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/28—Design optimisation, verification or simulation using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/08—Fluids
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
Abstract
The invention discloses an MOCVD (metal organic chemical vapor deposition) intracavity state identification method based on image processing and machine learning, which comprises the following steps of: constructing an MOCVD reaction cavity model based on computational fluid mechanics software and generating a training streamline diagram; carrying out image processing and classification on the training streamline graph to obtain a flow state classification result; training the BP neural network based on the classification result of the flowing state to obtain the trained BP neural network; and calculating the critical stable pressure based on the trained BP neural network and drawing a stability map. By using the method, engineering personnel can conveniently and quickly search the gas flow state corresponding to the process parameters, thereby realizing the epitaxial growth of the semiconductor material with higher quality. The MOCVD intracavity state identification method based on image processing and machine learning can be widely applied to the MOCVD state identification field.
Description
Technical Field
The invention belongs to the field of MOCVD state identification, and particularly relates to an MOCVD (metal organic chemical vapor deposition) intracavity state identification method based on image processing and machine learning.
Background
The gas flow in the MOCVD reaction cavity can be divided into a plurality of states, for example, unstable flow can be divided into buoyancy flow and rotating flow, stable flow can be divided into piston flow, piston-rotating transition flow and the like, the change of the streamline shape of each flow state is continuous without obvious boundary lines, the flow state judgment according to the MOCVD reaction cavity streamline diagram at present mainly depends on the experience of researchers, because no uniform judgment standard exists, the judgment of different persons or the same person at different time can be different, and a great deal of energy can be consumed when more samples are judged, so that the efficiency and the accuracy are reduced.
Disclosure of Invention
In order to solve the technical problems, the invention aims to provide an MOCVD (metal organic chemical vapor deposition) intracavity state identification method based on image processing and machine learning, which has more uniform classification standards and higher identification precision and efficiency.
The technical scheme adopted by the invention is as follows: an MOCVD (metal organic chemical vapor deposition) intracavity state identification method based on image processing and machine learning comprises the following steps:
s1, constructing an MOCVD reaction cavity model based on computational fluid dynamics software and generating a training streamline diagram;
s2, carrying out image processing and classification on the training streamline graph to obtain a flow state classification result;
s3, training the BP neural network based on the flow state classification result to obtain the trained BP neural network;
and S4, calculating the critical stable pressure based on the trained BP neural network and drawing a stability map.
Further, the step of constructing an MOCVD reaction cavity model based on computational fluid dynamics software and generating a training flow chart specifically comprises:
s11, constructing an MOCVD reaction chamber model based on computational fluid dynamics software;
s12, setting a plurality of groups of process parameters, obtaining a corresponding flow state based on the MOCVD reaction chamber model, and generating a training flow chart;
the process parameters include pressure, temperature, rotational speed, and flow rate.
Further, the step of performing image processing and classification on the training flow chart to obtain a flow state classification result specifically includes:
s21, carrying out image preprocessing for eliminating irrelevant information on the training streamline graph to obtain a preprocessed image;
s22, carrying out image segmentation on the preprocessed image based on the gray threshold value to obtain a segmented image;
s23, performing feature extraction on the segmented image to obtain image features;
and S24, classifying the corresponding process parameters in the training flow chart according to the image characteristics to obtain a flow state classification result.
Further, the step of performing image preprocessing for eliminating irrelevant information on the training flow chart to obtain a preprocessed image specifically includes:
carrying out graying processing on the training streamline graph;
carrying out geometric transformation processing on the training streamline graph, wherein the geometric transformation comprises translation, transposition, mirroring, rotation and scaling;
carrying out image enhancement processing on the training streamline graph;
and obtaining a preprocessed image.
Further, the step of performing feature extraction on the segmented image to obtain image features specifically includes:
s231, extracting features of the segmented image, wherein the features comprise vortexes between the edge of the reaction cavity and the substrate, vortexes above the substrate and the scale of the vortexes;
and S232, expressing the features into a numerical value and vector form to obtain the image features.
Further, the flow state includes a piston-rotation transition flow, a rotation flow, a buoyancy flow and a piston flow, and the step of classifying corresponding process parameters in the training flow line graph according to the image features to obtain a flow state classification result specifically includes:
judging that a vortex exists between the edge of the reaction cavity and the substrate, the scale of the vortex exceeds the edge of the substrate, and the flowing state is classified as a rotational flow;
judging that a vortex exists between the edge of the reaction cavity and the substrate, the scale of the vortex does not exceed the edge of the substrate, and the flowing state is classified into a rotating transition flow;
judging that no vortex exists between the edge of the reaction cavity and the substrate and a vortex exists above the substrate, and classifying the flow state into buoyancy flow;
when no vortex was found between the edge of the reaction chamber and the substrate and no vortex was found above the substrate, the flow state was classified as plug flow.
Further, the step of training the BP neural network based on the flow state classification result to obtain a trained BP neural network specifically includes:
and training the BP neural network by taking four process parameters of pressure, temperature, rotating speed and flow as input and taking a corresponding flow state classification result as output, thereby obtaining an output result of any point on a continuous variable space and the trained BP neural network.
Further, the step of calculating a critical stable pressure based on the trained BP neural network and drawing a stability map specifically includes:
s41, fixing the temperature and the flow, taking a point in a value interval of the rotating speed and calculating the critical stable pressure;
s42, recording the critical stable pressure of each point in the rotating speed interval and then tracing a point connecting line to obtain a stability curve;
and S43, adjusting the temperature or the flow rate, and returning to the step S41 until the preset cycle number is reached, so as to obtain a stability map under the condition of multiple process parameters.
The method and the system have the beneficial effects that: the method firstly carries out analog simulation in a large-range process parameter to obtain a streamline graph, then identifies and classifies the gas streamline graphs under different process conditions by means of an image identification technology, is more uniform in standard compared with manual classification, and higher in identification precision and efficiency, and finally learns the classification result through a neural network to obtain a stability curve under multiple process parameters (pressure, temperature, rotating speed and flow), so that engineering personnel can conveniently and quickly search the gas flow state of the corresponding process parameter, and further the epitaxial growth of the semiconductor material with higher quality is realized.
Drawings
FIG. 1 is a flow chart of steps of an embodiment of the present invention;
FIG. 2 is a conventional flow chart of the MOCVD intracavity state recognition method based on image processing and machine learning according to the present invention;
FIG. 3 is a schematic illustration of a rotating flow according to an embodiment of the present invention;
FIG. 4 is a schematic illustration of a piston-spin transition flow in accordance with an embodiment of the present invention;
FIG. 5 is a schematic illustration of buoyancy flow according to an embodiment of the present invention;
FIG. 6 is a schematic illustration of plug flow in accordance with an embodiment of the present invention;
FIG. 7 is a graphical illustration of the P- ω stability curve at a fixed flow rate and different temperatures for a particular embodiment of the present invention;
FIG. 8 is a graph illustrating P- ω stability curves at different flow rates at a fixed temperature for an embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the figures and the specific embodiments. The step numbers in the following embodiments are provided only for convenience of illustration, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.
Referring to fig. 1 and 2, the invention provides an MOCVD (metal organic chemical vapor deposition) intracavity state identification method based on image processing and machine learning, which comprises the following steps:
s1, constructing an MOCVD reaction cavity model based on computational fluid dynamics software and generating a training streamline diagram;
s2, carrying out image processing and classification on the training streamline graph to obtain a flow state classification result;
s3, training the BP neural network based on the flow state classification result to obtain the trained BP neural network;
and S4, calculating the critical stable pressure based on the trained BP neural network and drawing a stability map.
Specifically, in the P- ω stability map, a stable flow state (plug flow) region is below the curve, an unstable flow state region is above the curve, wherein a buoyancy flow state region is above the curve, and a rotational flow state region is above the curve. The curve represents the critical steady state region.
Further, as a preferred embodiment of the method, the step of constructing the MOCVD reaction chamber model based on the computational fluid dynamics software and generating the training flow chart specifically includes:
s11, constructing an MOCVD reaction chamber model based on computational fluid dynamics software;
s12, setting a plurality of groups of process parameters, obtaining a corresponding flow state based on the MOCVD reaction chamber model, and generating a training flow chart;
the process parameters include pressure, temperature, rotational speed, and flow rate.
Specifically, simulation is performed based on CFD software, and a plurality of process parameters are set within a certain range: the pressure was set to 2Torr, 4Torr, 6Torr, 8Torr and 10Torr, the temperature was set to 500K, 600K, 700K, 800K, 900K and 1000K, the rotation speed was set to 500rpm, 600rpm, 700rpm, 800rpm, 900rpm and 1000rpm, the flow rate was set to 1500sccm, 1750sccm, 2000sccm, 2250sccm, 2500sccm, 2750sccm and 3000sccm, respectively, and a flow chart was generated in accordance with the same image parameters (resolution, pattern angle, line width, etc.).
Further, as a preferred embodiment of the method, the step of performing image processing and classification on the training flow chart to obtain a flow state classification result specifically includes:
s21, carrying out image preprocessing for eliminating irrelevant information on the training streamline graph to obtain a preprocessed image;
in particular, the purpose of the pre-processing is to eliminate extraneous information, recover useful information, and simplify the data to the maximum extent.
S22, carrying out image segmentation on the preprocessed image based on the gray threshold value to obtain a segmented image;
specifically, the image segmentation divides the image into sub-regions which are not overlapped and have respective characteristics, and the pixel set in each region is continuous. In this link, one or more gray threshold values are calculated for the preprocessed image based on the gray features, the gray value of each pixel in the image is compared with the threshold values, and finally the pixels are classified into appropriate categories according to the comparison results.
S23, performing feature extraction on the segmented image to obtain image features;
and S24, classifying the corresponding process parameters in the training flow chart according to the image characteristics to obtain a flow state classification result.
As a further preferred embodiment of the present invention, the step of performing image preprocessing for eliminating irrelevant information on the training flow chart to obtain a preprocessed image specifically includes:
carrying out graying processing on the training streamline graph;
specifically, in the RGB model, if R ═ G ═ B, the color represents a gray color, and the value of R ═ G ═ B, that is, the gray value, in the gray image, only one byte is needed to store the gray value, so that the data amount required to be processed can be greatly reduced.
Carrying out geometric transformation processing on the training streamline graph, wherein the geometric transformation comprises translation, transposition, mirroring, rotation and scaling;
specifically, geometric transformation corrects systematic errors and random errors of image acquisition through spatial transformation such as translation, transposition, mirroring, rotation, scaling and the like.
Carrying out image enhancement processing on the training streamline graph;
in particular, useful information in the image is enhanced, the overall or local characteristics of the image are purposefully emphasized, the identification degree of the streamline in the streamline graph is enhanced, particularly the identification degree of the vortex at the edge of the reaction cavity and above the substrate, and the image interpretation and identification effects are enhanced.
And obtaining a preprocessed image.
As a further preferred embodiment of the present invention, the step of performing feature extraction on the segmented image to obtain the image features specifically includes:
s231, extracting features of the segmented image, wherein the features comprise vortexes between the edge of the reaction cavity and the substrate, vortexes above the substrate and the scale of the vortexes;
and S232, expressing the features into a numerical value and vector form to obtain the image features.
As a further preferred embodiment of the present invention, the flow state includes a piston-rotation transition flow, a rotation flow, a buoyancy flow and a piston flow, and the step of classifying the corresponding process parameters in the training flow line graph according to the image features to obtain the flow state classification result specifically includes:
judging that a vortex exists between the edge of the reaction cavity and the substrate, the scale of the vortex exceeds the edge of the substrate, and the flowing state is classified as a rotational flow;
judging that a vortex exists between the edge of the reaction cavity and the substrate, the scale of the vortex does not exceed the edge of the substrate, and the flowing state is classified into a rotating transition flow;
judging that no vortex exists between the edge of the reaction cavity and the substrate and a vortex exists above the substrate, and classifying the flow state into buoyancy flow;
when no vortex was found between the edge of the reaction chamber and the substrate and no vortex was found above the substrate, the flow state was classified as plug flow.
Specifically, fig. 3 is a schematic view in which the flow state is a rotational flow, fig. 4 is a schematic view in which the flow state is a rotational transition flow, fig. 5 is a schematic view in which the flow state is a buoyancy flow, and fig. 6 is a schematic view in which the flow state is a plug flow.
As a further preferred embodiment of the present invention, the step of training the BP neural network based on the classification result of the flow state to obtain a trained BP neural network specifically includes:
and training the BP neural network by taking four process parameters of pressure, temperature, rotating speed and flow as input and taking a corresponding flow state classification result as output, thereby obtaining an output result of any point on a continuous variable space and the trained BP neural network.
Specifically, the result of the previous section of classification is substituted into a BP neural network for machine learning, wherein four process parameters of pressure, temperature, rotating speed and flow are used as input, the corresponding flow state classification result is used as output, and the output result of any point in a continuous variable space can be obtained through the learning of a finite number of discrete variables.
Further as a preferred embodiment of the present invention, the step of calculating the critical stable pressure and drawing a stability map based on the trained BP neural network specifically includes:
s41, fixing the temperature and the flow, taking a point in a value interval of the rotating speed and calculating the critical stable pressure;
specifically, the value interval of the rotation speed is divided into a plurality of points, the pressure is gradually adjusted from small to large for each rotation speed point, and when the corresponding flow state changes (namely, the flow changes from stable flow to unstable flow), the pressure at the moment is the critical stable pressure.
S42, recording the critical stable pressure of each point in the rotating speed interval and then tracing a point connecting line to obtain a stability curve;
and S43, adjusting the temperature or the flow rate, and returning to the step S41 until the preset cycle number is reached, so as to obtain a stability map under the condition of multiple process parameters.
Specifically, the rotational speed is used as the abscissa and the pressure is used as the ordinate, so the method is also called a P- ω stability map, the P- ω stability map refers to fig. 7 and 8, a stable flow region is shown below a curve, and an unstable flow region is shown above the curve.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (8)
1. The MOCVD (metal organic chemical vapor deposition) intracavity state identification method based on image processing and machine learning is characterized by comprising the following steps of:
s1, constructing an MOCVD reaction cavity model based on computational fluid dynamics software and generating a training streamline diagram;
s2, carrying out image processing and classification on the training streamline graph to obtain a flow state classification result;
s3, training the BP neural network based on the flow state classification result to obtain the trained BP neural network;
and S4, calculating the critical stable pressure based on the trained BP neural network and drawing a stability map.
2. The MOCVD intracavity state recognition method based on image processing and machine learning as claimed in claim 1, wherein the step of constructing an MOCVD reaction chamber model based on computational fluid dynamics software and generating a training streamline diagram specifically comprises:
s11, constructing an MOCVD reaction chamber model based on computational fluid dynamics software;
s12, setting a plurality of groups of process parameters, obtaining a corresponding flow state based on the MOCVD reaction chamber model, and generating a training flow chart;
the process parameters include pressure, temperature, rotational speed, and flow rate.
3. The MOCVD intracavity state recognition method based on image processing and machine learning as claimed in claim 2, wherein the step of performing image processing and classification on the training flow chart to obtain a flow state classification result specifically comprises:
s21, carrying out image preprocessing for eliminating irrelevant information on the training streamline graph to obtain a preprocessed image;
s22, carrying out image segmentation on the preprocessed image based on the gray threshold value to obtain a segmented image;
s23, performing feature extraction on the segmented image to obtain image features;
and S24, classifying the corresponding process parameters in the training flow chart according to the image characteristics to obtain a flow state classification result.
4. The MOCVD intracavity state recognition method based on image processing and machine learning as claimed in claim 3, wherein the step of performing image preprocessing for eliminating irrelevant information on the training flow chart to obtain a preprocessed image specifically comprises:
carrying out graying processing on the training streamline graph;
carrying out geometric transformation processing on the training streamline graph, wherein the geometric transformation comprises translation, transposition, mirroring, rotation and scaling;
carrying out image enhancement processing on the training streamline graph;
and obtaining a preprocessed image.
5. The MOCVD intracavity state recognition method based on image processing and machine learning as claimed in claim 4, wherein the step of performing feature extraction on the segmented image to obtain image features specifically comprises:
s231, extracting features of the segmented image, wherein the features comprise vortexes between the edge of the reaction cavity and the substrate, vortexes above the substrate and the scale of the vortexes;
and S232, expressing the features into a numerical value and vector form to obtain the image features.
6. The MOCVD intracavity state recognition method based on image processing and machine learning as claimed in claim 5, wherein the flow states include piston-rotation transition flow, rotation flow, buoyancy flow and piston flow, and the step of classifying corresponding process parameters in the training flow chart according to image features to obtain a flow state classification result specifically comprises:
judging that a vortex exists between the edge of the reaction cavity and the substrate, the scale of the vortex exceeds the edge of the substrate, and the flowing state is classified as a rotational flow;
judging that a vortex exists between the edge of the reaction cavity and the substrate, the scale of the vortex does not exceed the edge of the substrate, and the flowing state is classified into a rotating transition flow;
judging that no vortex exists between the edge of the reaction cavity and the substrate and a vortex exists above the substrate, and classifying the flow state into buoyancy flow;
when no vortex was found between the edge of the reaction chamber and the substrate and no vortex was found above the substrate, the flow state was classified as plug flow.
7. The MOCVD intracavity state recognition method based on image processing and machine learning as claimed in claim 6, wherein the step of training the BP neural network based on the flow state classification result to obtain the trained BP neural network specifically comprises:
and training the BP neural network by taking four process parameters of pressure, temperature, rotating speed and flow as input and taking a corresponding flow state classification result as output, thereby obtaining an output result of any point on a continuous variable space and the trained BP neural network.
8. The MOCVD intracavity state recognition method based on image processing and machine learning as claimed in claim 7, wherein the step of calculating the critical stable pressure based on the trained BP neural network and drawing a stability map specifically comprises:
s41, fixing the temperature and the flow, taking a point in a value interval of the rotating speed and calculating the critical stable pressure;
s42, recording the critical stable pressure of each point in the rotating speed interval and then tracing a point connecting line to obtain a stability curve;
and S43, adjusting the temperature or the flow rate, and returning to the step S41 until the preset cycle number is reached, so as to obtain a stability map under the condition of multiple process parameters.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111133298.XA CN113792506B (en) | 2021-09-27 | 2021-09-27 | MOCVD intracavity state identification method based on image processing and machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111133298.XA CN113792506B (en) | 2021-09-27 | 2021-09-27 | MOCVD intracavity state identification method based on image processing and machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113792506A true CN113792506A (en) | 2021-12-14 |
| CN113792506B CN113792506B (en) | 2024-04-02 |
Family
ID=78877265
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111133298.XA Active CN113792506B (en) | 2021-09-27 | 2021-09-27 | MOCVD intracavity state identification method based on image processing and machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113792506B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110889260A (en) * | 2018-09-05 | 2020-03-17 | 长鑫存储技术有限公司 | Method and device for detecting process parameters, electronic equipment and computer readable medium |
| CN111046936A (en) * | 2019-12-05 | 2020-04-21 | 湖北民族大学 | Bridge crack classification identification method based on cascade width neural network |
| CN112097830A (en) * | 2020-08-27 | 2020-12-18 | 中山大学 | MOCVD cavity stability detection method and system based on graph model |
| CN112132196A (en) * | 2020-09-14 | 2020-12-25 | 中山大学 | Cigarette case defect identification method combining deep learning and image processing |
| DE102020100565A1 (en) * | 2020-01-13 | 2021-07-15 | Aixtron Se | Process for depositing layers |
-
2021
- 2021-09-27 CN CN202111133298.XA patent/CN113792506B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110889260A (en) * | 2018-09-05 | 2020-03-17 | 长鑫存储技术有限公司 | Method and device for detecting process parameters, electronic equipment and computer readable medium |
| CN111046936A (en) * | 2019-12-05 | 2020-04-21 | 湖北民族大学 | Bridge crack classification identification method based on cascade width neural network |
| DE102020100565A1 (en) * | 2020-01-13 | 2021-07-15 | Aixtron Se | Process for depositing layers |
| CN112097830A (en) * | 2020-08-27 | 2020-12-18 | 中山大学 | MOCVD cavity stability detection method and system based on graph model |
| CN112132196A (en) * | 2020-09-14 | 2020-12-25 | 中山大学 | Cigarette case defect identification method combining deep learning and image processing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113792506B (en) | 2024-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112287978A (en) | Hyperspectral remote sensing image classification method based on self-attention context network | |
| CN111699494A (en) | Object recognition using convolutional neural networks trained by principal component analysis and repetitive spectral clustering | |
| CN111241744B (en) | Low-pressure casting machine time sequence data abnormity detection method based on bidirectional LSTM | |
| CN110580450A (en) | traffic sign identification method based on convolutional neural network | |
| JP2010134957A (en) | Pattern recognition method | |
| CN109726195B (en) | Data enhancement method and device | |
| CN113077388B (en) | Data-augmented deep semi-supervised over-limit learning image classification method and system | |
| CN114752996A (en) | Intelligent adjustment method for monocrystalline silicon shoulder-laying stage forking and storage medium | |
| CN111008649B (en) | Defect detection data set preprocessing method based on three decisions | |
| Suzuki et al. | Adversarial transformations for semi-supervised learning | |
| CN111402208A (en) | Incremental-based weld surface defect feature extraction and classification identification method | |
| CN113421223B (en) | Industrial product surface defect detection method based on deep learning and Gaussian mixture | |
| CN113792506B (en) | MOCVD intracavity state identification method based on image processing and machine learning | |
| CN112270404A (en) | Detection structure and method for bulge defect of fastener product based on ResNet64 network | |
| CN115294424A (en) | Sample data enhancement method based on generation countermeasure network | |
| CN112561949B (en) | Rapid moving object detection algorithm based on RPCA and support vector machine | |
| CN112529081B (en) | Real-time semantic segmentation method based on efficient attention calibration | |
| CN113850274A (en) | Image classification method based on HOG characteristics and DMD | |
| CN114842178A (en) | Online visual interaction system and method based on electronic product | |
| CN113177599A (en) | Enhanced sample generation method based on GAN | |
| JP4509512B2 (en) | Skew detection | |
| Kmiec | New optimal character recognition method based on Hu invariant moments and weighted voting | |
| CN111652311A (en) | Image sparse representation method based on sparse elliptic RBF neural network | |
| CN111882563B (en) | Semantic segmentation method based on directional full convolution network | |
| CN115239657B (en) | Industrial part increment identification method based on deep learning target segmentation |
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 |