CN110310260A - Sub-material decision-making technique, equipment and storage medium based on machine learning model - Google Patents
Sub-material decision-making technique, equipment and storage medium based on machine learning model Download PDFInfo
- Publication number
- CN110310260A CN110310260A CN201910530692.3A CN201910530692A CN110310260A CN 110310260 A CN110310260 A CN 110310260A CN 201910530692 A CN201910530692 A CN 201910530692A CN 110310260 A CN110310260 A CN 110310260A
- Authority
- CN
- China
- Prior art keywords
- sub
- defect
- decision
- material component
- defect type
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30164—Workpiece; Machine component
Abstract
The present invention provides a kind of sub-material decision-making technique, equipment and storage medium based on machine learning model, is related to field of cloud calculation.This method comprises: treating sub-material component carries out multi-angled shooting, the multiple images to sub-material component are obtained；According to described multiple images, determine described at least one defect type existing for sub-material component and the corresponding confidence level of at least one defect type；Using it is described at least one defect type existing for sub-material component and the corresponding confidence level of at least one defect type as the input parameter of the machine learning model, it is described to the corresponding sub-material result of decision of sub-material component to determine.The embodiment of the present invention improves detection sub-material efficiency, can carry out complete detection to product component.
Description
Technical field
The present invention relates to technical field of quality detection more particularly to a kind of sub-material decision-making parties based on machine learning model
Method, equipment and storage medium.
Background technique
It is so-called " 3C Product " if 3C Product components manufacture in traditional industry manufacturing industry production scene, it is exactly computer
(Computer), (Communication) and consumer electronics product (Consumer Electronics) triplicity is communicated,
Also known as " information household appliances ".It generally requires to carry out surface state detection to multiple angles of product.The surface state of product detects
The important link of manufacturer's control delivery quality.
In the related technology, general by manually visualizing quality inspection, i.e. every a certain number of Quality Inspectors of production line training use meat
Eye carries out the defects detection and classification of full angle to the components produced, and detection efficiency is low；And Quality Inspector is often detecting
It just will do it sub-material after to a kind of failure in some face of part, part will not comprehensively be detected.
Summary of the invention
The present invention provides a kind of sub-material decision-making technique, equipment and storage medium based on machine learning model, to improve inspection
Efficiency is surveyed, and complete detection can be carried out to product component.
In a first aspect, the present invention provides a kind of sub-material decision-making technique based on machine learning model, comprising:
It treats sub-material component and carries out multi-angled shooting, obtain the multiple images to sub-material component；
According to described multiple images, determine described at least one defect type existing for sub-material component and described at least one
The kind corresponding confidence level of defect type；
It is respectively corresponded to described at least one defect type existing for sub-material component at least one defect type
Input parameter of the confidence level as the machine learning model, it is described to the corresponding sub-material decision knot of sub-material component to determine
Fruit.
In one possible implementation, described according to described multiple images, it determines described to existing for sub-material component
At least one defect type and the corresponding confidence level of at least one defect type, comprising:
Obtain the multiple defect classification models pre-established；
For each defect classification model, at least one described image is determined, and will at least one described described figure
It is described to the corresponding defect type of sub-material component and the defect to obtain as the input parameter as the defect classification model
The corresponding confidence level of type.
It is in one possible implementation, described to establish multiple defect classification models, comprising:
Obtain the corresponding training data of each defect classification model；The training data includes different Image Acquisition
Under the conditions of the multiple images that acquire and the corresponding defect type of each described image；
According to the corresponding training data of each defect classification model, each defect classification model of training.
In one possible implementation, the general is described at least one defect type existing for sub-material component and institute
State input parameter of the corresponding confidence level of at least one defect type as the machine learning model, with determine it is described to
Before the corresponding sub-material result of decision of sub-material component, further includes:
Obtain the corresponding multiple training datas of the machine learning model；The training data includes: each defect
Defect type, the corresponding confidence level of the defect type and the sub-material result of decision of disaggregated model output；
Pass through the multiple training data, the training machine learning model.
In one possible implementation, further includes:
It, will be described to sub-material component sub-material to material corresponding with the sub-material result of decision according to the sub-material result of decision
Box.
In one possible implementation, described according to the sub-material result of decision, it will be described to sub-material component sub-material
To magazine corresponding with the sub-material result of decision, comprising:
According to the sub-material result of decision, control instruction is generated；
According to the control instruction, control mechanical arm will it is described to sub-material component sub-material to the sub-material result of decision pair
The magazine answered.
In one possible implementation, the machine learning model is following any model: Adaboost model, mind
Through network model, supporting vector machine model, decision-tree model.
Second aspect, the present invention provide a kind of sub-material decision making device based on machine learning model, comprising:
Image capture module carries out multi-angled shooting for treating sub-material component, obtains described to the multiple of sub-material component
Image；
Defect categorization module, for determining described scarce at least one existing for sub-material component according to described multiple images
Fall into type and the corresponding confidence level of at least one defect type；
Processing module is used for by described in at least one defect type existing for sub-material component and at least one defect
Input parameter of the corresponding confidence level of type as the machine learning model, it is described corresponding to sub-material component to determine
The sub-material result of decision.
In one possible implementation, defect categorization module is specifically used for:
Establish multiple defect classification models；
For each defect classification model, at least one described image is determined, and will at least one described described figure
It is described to the corresponding defect type of sub-material component and the defect to obtain as the input parameter as the defect classification model
The corresponding confidence level of type.
In one possible implementation, further includes: model training module is used for:
Obtain the corresponding training data of each defect classification model；The training data includes different Image Acquisition
Under the conditions of the multiple images that acquire and the corresponding defect type of each described image；
According to the corresponding training data of each defect classification model, each defect classification model of training.
In one possible implementation, model training module is also used to:
Obtain the corresponding multiple training datas of the machine learning model；The training data includes: each defect
Defect type, the corresponding confidence level of the defect type and the final sub-material result of decision of disaggregated model output；
Pass through the multiple training data, the training machine learning model.
In one possible implementation, further includes: sub-material module is used for:
It, will be described to sub-material component sub-material to magazine corresponding with the sub-material result of decision according to the sub-material result of decision.
In one possible implementation, sub-material module is specifically used for:
According to the sub-material result of decision, control instruction is generated；
According to the control instruction, controlling mechanical arm will be described extremely corresponding with the sub-material result of decision to sub-material component sub-material
Magazine.
In one possible implementation, the machine learning model is following any model: Adaboost model, mind
Through network model, supporting vector machine model, decision-tree model.
The third aspect, the embodiment of the present invention provide a kind of computer readable storage medium, are stored thereon with computer program,
Method described in any one of first aspect is realized when the computer program is executed by processor.
Fourth aspect, the embodiment of the present invention provide a kind of electronic equipment, comprising:
Processor；And
Memory, for storing the executable instruction of the processor；
Wherein, the processor is configured to execute described in any one of first aspect via the executable instruction is executed
Method.
Sub-material decision-making technique, equipment and storage medium provided in an embodiment of the present invention based on machine learning model, is treated
Sub-material component carries out multi-angled shooting, obtains the multiple images to sub-material component；According to described multiple images, determine described in
To at least one defect type existing for sub-material component and the corresponding confidence level of at least one defect type；It will be described
To at least one defect type existing for sub-material component and at least one corresponding confidence level of defect type as institute
The input parameter of machine learning model is stated, it is described to the corresponding sub-material result of decision of sub-material component to determine, improve detection effect
Rate, and according to at least one defect type existing for sub-material component and described at least one defect type is corresponding sets
Reliability more can comprehensively treat sub-material component and carry out defects detection using machine learning model, and can be determined according to final
Plan result carries out sub-material, and the accuracy of sub-material is higher.
Detailed description of the invention
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the disclosure
Example, and together with specification for explaining the principles of this disclosure.
Fig. 1 is the flow diagram of sub-material decision-making technique one embodiment provided by the invention based on machine learning model；
Fig. 2 is the schematic illustration of one embodiment of method provided by the invention；
Fig. 3 is the realization process schematic of one embodiment of method provided by the invention；
Fig. 4 is the structural schematic diagram of sub-material decision making device one embodiment provided by the invention based on machine learning model；
Fig. 5 is the structural schematic diagram of electronic equipment embodiment provided by the invention.
Through the above attached drawings, it has been shown that the specific embodiment of the disclosure will be hereinafter described in more detail.These attached drawings
It is not intended to limit the scope of this disclosure concept by any means with verbal description, but is by referring to specific embodiments
Those skilled in the art illustrate the concept of the disclosure.
Specific embodiment
Example embodiments are described in detail here, and the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent all implementations consistent with this disclosure.On the contrary, they be only with it is such as appended
The example of the consistent device and method of some aspects be described in detail in claims, the disclosure.
Term " includes " in description and claims of this specification and the attached drawing and " having " and they appoint
What is deformed, it is intended that is covered and non-exclusive is included.Such as contain the process, method, system, production of a series of steps or units
Product or equipment are not limited to listed step or unit, but optionally further comprising the step of not listing or unit, or
Optionally further comprising the other step or units intrinsic for these process, methods, product or equipment.
Application scenarios according to the present invention are introduced first:
Distribution method provided in an embodiment of the present invention based on machine learning model, applied to based on product component surface
Image data carries out quality testing to product component, and product component is carried out to the scene of sub-material, i.e., final according to quality testing
Product component is carried out sub-material by obtained defect type.
The product component being directed in the embodiment of the present invention is, for example, 3C Product components etc., for example, mobile phone, tablet computer, can
Dress the components of smart machine, such as the charging mouthpiece of mobile phone.
Technical solution of the present invention is described in detail with specific embodiment below.These specific implementations below
Example can be combined with each other, and the same or similar concept or process may be repeated no more in some embodiments.
Fig. 1 is the flow diagram of sub-material decision-making technique one embodiment provided by the invention based on machine learning model.
As shown in Figure 1, method provided in this embodiment, comprising:
Step 101 treats sub-material component progress multi-angled shooting, obtains the multiple images to sub-material component.
Specifically, by image collecting device, such as camera, each face for treating sub-material component are shot, and are acquired
To multiple images.Such as multiple images are collected under different shooting angles, illumination condition.
Collected training data under different images acquisition condition can be instructed according to deep learning algorithm in advance
Practice, obtains the multiple defect classification models for predicting different image datas；It is lacked in the training data including what artificial quality inspection obtained
Fall into type.Image Acquisition condition is for example including at least one of following: shooting angle, illumination condition.
Step 102, according to described multiple images, determine described at least one defect type existing for sub-material component and institute
State the corresponding confidence level of at least one defect type.
Specifically, multiple images can be inputted in the defect classification model pre-established, obtains this and deposited to sub-material component
At least one defect type and the corresponding confidence level of various defect types, confidence level is bigger, then should to sub-material component exist
A possibility that the type defect, is bigger.
To at least one defect type existing for sub-material component and at least one defect type described in step 103, general
Input parameter of corresponding confidence level as the machine learning model, it is described to the corresponding sub-material of sub-material component to determine
The result of decision.
Specifically, at least one defect type and at least one corresponding confidence level of defect type are inputted pre-
In the machine learning model first established, the sub-material result of decision of machine learning model output includes that should deposit to sub-material component
Defect type, further can also include the corresponding final confidence level of the defect type, or include this to sub-material component
For qualified products, the corresponding confidence level of qualified products.
Further, after obtaining the sub-material result of decision, further includes:
It, will be described to sub-material component sub-material to magazine corresponding with the sub-material result of decision according to the sub-material result of decision.
Specifically it can realize in the following way sub-material:
According to the sub-material result of decision, control instruction is generated；
According to the control instruction, controlling mechanical arm will be described extremely corresponding with the sub-material result of decision to sub-material component sub-material
Magazine.
The method of the present embodiment treats sub-material component and carries out multi-angled shooting, obtains multiple figures to sub-material component
Picture；According to described multiple images, determine described at least one defect type existing for sub-material component and described at least one scarce
Fall into the corresponding confidence level of type；It will be described at least one defect type existing for sub-material component and described at least one scarce
Input parameter of the corresponding confidence level of type as the machine learning model is fallen into, it is described corresponding to sub-material component to determine
The sub-material result of decision, improve detection efficiency, and according to at least one defect type existing for sub-material component and described
At least one corresponding confidence level of defect type more can comprehensively treat sub-material component using machine learning model
Defects detection is carried out, and sub-material can be carried out according to final decision result, the accuracy of sub-material is higher.
On the basis of the above embodiments, further, step 102 can be specifically accomplished in that
Obtain the multiple defect classification models pre-established；
For each defect classification model, at least one described image is determined, and will at least one described described figure
It is described to the corresponding defect type of sub-material component and the defect to obtain as the input parameter as the defect classification model
The corresponding confidence level of type.
Further, establishing multiple defect classification models can specifically be accomplished in that
Obtain the corresponding training data of each defect classification model；The training data includes different Image Acquisition
Under the conditions of the multiple images that acquire and the corresponding defect type of each described image；
According to the corresponding training data of each defect classification model, each defect classification model of training.
Specifically, multiple defect classification models can be established using deep learning algorithm previously according to training data.Training
Data include collected training data under the conditions of different shooting angles, illumination condition etc..
The input parameter for determining each defect classification model respectively determines at least one image conduct that is, from multiple images
The input parameter of each defect classification model obtains the defect classification results of each model output, i.e., to existing for sub-material component
Defect type and the corresponding confidence level of each defect type.
Different defect classification models can correspond to different Image Acquisition conditions, such as shooting angle, illumination condition.
The input parameter of different defect classification models is determined according to Image Acquisition condition, i.e., by collected multiple images
The defect classification model that each image should input is determined respectively according to the Image Acquisition condition of the image.
On the basis of the above embodiments, further, it before step 103, can also establish and its learning model, tool
Body can be accomplished in that
Obtain the corresponding multiple training datas of the machine learning model；The training data includes: each defect
Defect type, the corresponding confidence level of the defect type and the final sub-material result of decision of disaggregated model output；
Pass through the multiple training data, the training machine learning model.
Further, the machine learning model is following any model: Adaboost model, neural network model, branch
Hold vector machine model, decision-tree model.
The image of multiple components is inputted into defect classification model, obtains the different components of each defect classification model output
Defect type and the corresponding confidence level of the defect type, and corresponding final point of all parts is determined according to artificial experience
The material result of decision is trained machine learning model using above-mentioned data as training data.
As shown in Fig. 2, the obtained machine learning model of training is by the defect type answered to sub-material component 3 and each
The corresponding confidence level of defect type obtains the final sub-material result of decision as input, i.e., should be to sub-material component existing probability most
Big defect type, or should be qualified products (further can also include the corresponding confidence level of qualified products) to sub-material component,
Sub-material decision is carried out according to the sub-material result of decision, i.e., this is waited for that sub-material component is put into magazine corresponding with the sub-material result of decision；
The input of machine learning model can be the vector that the corresponding confidence level of each defect type is formed.
It in above scheme, is predicted, is obtained pre- by the defect type that multiple defect classification models treat sub-material component
The defect type of survey and the corresponding confidence level of each defect type, and according to each defect type and each defect type
Corresponding confidence level judges the final sub-material result of decision by machine learning model, carries out sub-material according to the sub-material result of decision,
Accuracy rate is higher.
The sub-material decision-making technique of the embodiment of the present invention can be executed by one or more electronic equipments, the place of electronic equipment
Reason device control image collecting device treats sub-material component acquisition image, and image collecting device can be one on electronic equipment
Point, or separately positioned with electronic equipment, image collecting device is by collected multiple images and should information to sub-material component
It generates predictions request and is sent to processor, for processor according to predictions request by defect classification model, output should be to sub-material component
Corresponding defect type and the corresponding confidence level of each defect type.This is finally waited for that sub-material component is corresponding and lacked by processor
The input of type and the corresponding confidence level of each defect type as machine learning model is fallen into, the machine learning model is run
It is exported as a result, the i.e. sub-material result of decision, such as (should can also further be wrapped to the corresponding final defect type of sub-material component
Include the corresponding final confidence level of defect type), or should be qualified products to sub-material component.Finally according to the sub-material result of decision to this
Sub-material is carried out to sub-material component, that is, is put into corresponding magazine, such as be put into the magazine of faulty item or the magazine of qualified products,
The magazine of faulty item can be arranged multiple according to different defect types.The electronic equipment can also export intermediate result
Or storage, such as the corresponding defect type of sub-material component and the corresponding confidence level of each defect type, which export, to be waited for this
Or storage, it is used convenient for subsequent analysis.The electronic equipment can be sent out the prompt informations such as alarm, such as some corresponds to defect class
The corresponding confidence level of type is directly alarmed after being greater than a certain threshold value.
Further, the electronic equipment can also to each computer hardware resource of entire product line, operating system, CPU with
And Heterogeneous Computing chip service condition, memory service condition are monitored.
In conclusion the method for the embodiment of the present invention, for the mutually independent prediction result of multiple defect classification models, benefit
With machine learning algorithm such as Adaboost, the decision classified based on more defect classification model prediction results greatly improved
Accuracy rate can effectively inhibit the model of small probability to cross and kill, while can adapt to produce in real time by the adjustment of manual oversight data
Line examination criteria ensure that larger yields, significantly reduce the manufacturing cost of factory, and the method Shandong of the embodiment of the present invention
Stick is strong, being capable of sustainable study optimization.
Fig. 4 is the structure chart of sub-material decision making device one embodiment provided by the invention based on machine learning model, such as Fig. 4
It is shown, the sub-material decision making device based on machine learning model of the present embodiment, comprising:
Image capture module 401 carries out multi-angled shooting for treating sub-material component, obtains described to the more of sub-material component
A image；
Defect categorization module 402, for determining described at least one existing for sub-material component according to described multiple images
Defect type and the corresponding confidence level of at least one defect type；
Processing module 403, being used for will be described at least one defect type existing for sub-material component and at least one
Input parameter of the corresponding confidence level of defect type as the machine learning model, it is described to sub-material component pair to determine
The sub-material result of decision answered.
In one possible implementation, defect categorization module 402, is specifically used for:
Establish multiple defect classification models；
For each defect classification model, at least one described image is determined, and will at least one described described figure
It is described to the corresponding defect type of sub-material component and the defect to obtain as the input parameter as the defect classification model
The corresponding confidence level of type.
In one possible implementation, further includes: model training module is used for:
Obtain the corresponding training data of each defect classification model；The training data includes different Image Acquisition
Under the conditions of the multiple images that acquire and the corresponding defect type of each described image；
According to the corresponding training data of each defect classification model, each defect classification model of training.
In one possible implementation, model training module is also used to:
Obtain the corresponding multiple training datas of the machine learning model；The training data includes: each defect
Defect type, the corresponding confidence level of the defect type and the final sub-material result of decision of disaggregated model output；
Pass through the multiple training data, the training machine learning model.
In one possible implementation, further includes: sub-material module 404 is used for:
It, will be described to sub-material component sub-material to magazine corresponding with the sub-material result of decision according to the sub-material result of decision.
In one possible implementation, sub-material module 404, is specifically used for:
According to the sub-material result of decision, control instruction is generated；
According to the control instruction, controlling mechanical arm will be described extremely corresponding with the sub-material result of decision to sub-material component sub-material
Magazine.
In one possible implementation, the machine learning model is following any model: Adaboost model, mind
Through network model, supporting vector machine model, decision-tree model.
As shown in figure 3, image capture module is by collected multiple images and should generate to the information of sub-material component pre-
It surveys request and is sent to processor, for processing module according to predictions request by defect classification model, output should be corresponding to sub-material component
Defect classification results, i.e., should to defect type existing for sub-material component and the corresponding confidence level of each defect type.Processing
This is waited for the corresponding defect type of sub-material component and the corresponding confidence level of each defect type as machine learning model by module
Input, run the machine learning model and exported as a result, the i.e. sub-material result of decision, such as should be corresponding most to sub-material component
Whole defect type (further can also include the corresponding final confidence level of defect type), or should be qualified produce to sub-material component
Product.Sub-material module waits for that sub-material component carries out sub-material to this according to the sub-material result of decision, that is, is put into corresponding magazine, such as be put into
The magazine of faulty item or the magazine of qualified products, the magazine of faulty item can be arranged multiple according to different defect types.
The device of the present embodiment can be used for executing the technical solution of above method embodiment, realization principle and technology
Effect is similar, and details are not described herein again.
Fig. 5 is the structure chart of electronic equipment embodiment provided by the invention, as shown in figure 5, the electronic equipment includes:
Processor 501, and, the memory 302 of the executable instruction for storage processor 501.
It optionally, can also include: image acquisition component 503, for acquiring the image to sub-material component.
Above-mentioned component can be communicated by one or more bus.
Wherein, processor 501 is configured to execute via the executable instruction is executed corresponding in preceding method embodiment
Method, specific implementation process may refer to preceding method embodiment, and details are not described herein again.
A kind of computer readable storage medium is also provided in the embodiment of the present invention, is stored thereon with computer program, it is described
Realize that corresponding method in preceding method embodiment, specific implementation process may refer to when computer program is executed by processor
Preceding method embodiment, it is similar that the realization principle and technical effect are similar, and details are not described herein again.
Those skilled in the art after considering the specification and implementing the invention disclosed here, will readily occur to its of the disclosure
Its embodiment.The present invention is directed to cover any variations, uses, or adaptations of the disclosure, these modifications, purposes or
Person's adaptive change follows the general principles of this disclosure and including the undocumented common knowledge in the art of the disclosure
Or conventional techniques.The description and examples are only to be considered as illustrative, and the true scope and spirit of the disclosure are by following
Claims are pointed out.
It should be understood that the present disclosure is not limited to the precise structures that have been described above and shown in the drawings, and
And various modifications and changes may be made without departing from the scope thereof.The scope of the present disclosure is only limited by appended claims
System.
Claims (11)
1. a kind of sub-material decision-making technique based on machine learning model characterized by comprising
It treats sub-material component and carries out multi-angled shooting, obtain the multiple images to sub-material component；
According to described multiple images, determine described at least one defect type existing for sub-material component and described at least one scarce
Fall into the corresponding confidence level of type；
By described at least one defect type existing for sub-material component and described at least one defect type is corresponding sets
Input parameter of the reliability as the machine learning model, it is described to the corresponding sub-material result of decision of sub-material component to determine.
2. determination is described to sub-material the method according to claim 1, wherein described according to described multiple images
At least one defect type existing for component and the corresponding confidence level of at least one defect type, comprising:
Obtain the multiple defect classification models pre-established；
For each defect classification model, at least one described image is determined, and at least one described described image is made
It is described to the corresponding defect type of sub-material component and the defect type to obtain for the input parameter of the defect classification model
Corresponding confidence level.
3. according to the method described in claim 2, it is characterized in that, described establish multiple defect classification models, comprising:
Obtain the corresponding training data of each defect classification model；The training data includes different Image Acquisition condition
The multiple images and the corresponding defect type of each described image of lower acquisition；
According to the corresponding training data of each defect classification model, each defect classification model of training.
4. method according to claim 1-3, which is characterized in that it is described will it is described to existing for sub-material component extremely
A kind of few defect type and the corresponding confidence level of at least one defect type are as the defeated of the machine learning model
Enter parameter, before described in determination to the corresponding sub-material result of decision of sub-material component, further includes:
Obtain the corresponding multiple training datas of the machine learning model；The training data includes: each defect classification
Defect type, the corresponding confidence level of the defect type and the sub-material result of decision of model output；
Pass through the multiple training data, the training machine learning model.
5. method according to claim 1-3, which is characterized in that further include:
It, will be described to sub-material component sub-material to magazine corresponding with the sub-material result of decision according to the sub-material result of decision.
6. according to the method described in claim 5, general is described wait divide it is characterized in that, described according to the sub-material result of decision
Expect component sub-material to magazine corresponding with the sub-material result of decision, comprising:
According to the sub-material result of decision, control instruction is generated；
According to the control instruction, controlling mechanical arm will be described extremely corresponding with the sub-material result of decision to sub-material component sub-material
Magazine.
7. method according to claim 1-3, which is characterized in that the machine learning model is following any mould
Type: Adaboost model, neural network model, supporting vector machine model, decision-tree model.
8. a kind of sub-material decision making device based on machine learning model characterized by comprising
Image capture module carries out multi-angled shooting for treating sub-material component, obtains the multiple images to sub-material component；
Defect categorization module, for determining described at least one defect class existing for sub-material component according to described multiple images
Type and the corresponding confidence level of at least one defect type；
Processing module is used for by described in at least one defect type existing for sub-material component and at least one defect type
Input parameter of corresponding confidence level as the machine learning model, it is described to the corresponding sub-material of sub-material component to determine
The result of decision.
9. device according to claim 8, which is characterized in that defect categorization module is specifically used for:
Obtain the multiple defect classification models pre-established；
For each defect classification model, at least one described image is determined, and at least one described described image is made
It is described to the corresponding defect type of sub-material component and the defect type to obtain for the input parameter of the defect classification model
Corresponding confidence level.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
Claim 1-8 described in any item methods are realized when being executed by processor.
11. a kind of electronic equipment characterized by comprising
Processor；And
Memory, for storing the executable instruction of the processor；
Wherein, the processor is configured to require 1-8 described in any item via executing the executable instruction and carry out perform claim
Method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910530692.3A CN110310260B (en) | 2019-06-19 | 2019-06-19 | Material distribution decision method, equipment and storage medium based on machine learning model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910530692.3A CN110310260B (en) | 2019-06-19 | 2019-06-19 | Material distribution decision method, equipment and storage medium based on machine learning model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110310260A true CN110310260A (en) | 2019-10-08 |
| CN110310260B CN110310260B (en) | 2021-10-15 |
Family
ID=68076950
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910530692.3A Active CN110310260B (en) | 2019-06-19 | 2019-06-19 | Material distribution decision method, equipment and storage medium based on machine learning model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110310260B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113781485A (en) * | 2021-11-12 | 2021-12-10 | 成都数联云算科技有限公司 | Intelligent detection method and device for PCB defect types, electronic equipment and medium |
| WO2024050700A1 (en) * | 2022-09-06 | 2024-03-14 | 宁德时代新能源科技股份有限公司 | Detection method, detection apparatus and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107167716A (en) * | 2017-07-11 | 2017-09-15 | 国网福建省电力有限公司泉州供电公司 | A kind of shelf depreciation default kind identification method and device |
| CN107657603A (en) * | 2017-08-21 | 2018-02-02 | 北京精密机电控制设备研究所 | A kind of industrial appearance detecting method based on intelligent vision |
| CN108672316A (en) * | 2018-03-27 | 2018-10-19 | 哈尔滨理工大学 | A kind of micro parts quality detecting system based on convolutional neural networks |
| CN109118482A (en) * | 2018-08-07 | 2019-01-01 | 腾讯科技（深圳）有限公司 | A kind of panel defect analysis method, device and storage medium |
| CN109344905A (en) * | 2018-10-22 | 2019-02-15 | 王子蕴 | A kind of transmission facility automatic fault recognition methods based on integrated study |
| CN109408389A (en) * | 2018-10-30 | 2019-03-01 | 北京理工大学 | A kind of aacode defect detection method and device based on deep learning |
-
2019
- 2019-06-19 CN CN201910530692.3A patent/CN110310260B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107167716A (en) * | 2017-07-11 | 2017-09-15 | 国网福建省电力有限公司泉州供电公司 | A kind of shelf depreciation default kind identification method and device |
| CN107657603A (en) * | 2017-08-21 | 2018-02-02 | 北京精密机电控制设备研究所 | A kind of industrial appearance detecting method based on intelligent vision |
| CN108672316A (en) * | 2018-03-27 | 2018-10-19 | 哈尔滨理工大学 | A kind of micro parts quality detecting system based on convolutional neural networks |
| CN109118482A (en) * | 2018-08-07 | 2019-01-01 | 腾讯科技（深圳）有限公司 | A kind of panel defect analysis method, device and storage medium |
| CN109344905A (en) * | 2018-10-22 | 2019-02-15 | 王子蕴 | A kind of transmission facility automatic fault recognition methods based on integrated study |
| CN109408389A (en) * | 2018-10-30 | 2019-03-01 | 北京理工大学 | A kind of aacode defect detection method and device based on deep learning |
Non-Patent Citations (2)
| Title |
|---|
| 符瑞盛: "基于多分类器集成的工业品缺陷分析方法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 魏秀参: "《解析深度学习 卷积神经网络原理与视觉实践》", 30 November 2018 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113781485A (en) * | 2021-11-12 | 2021-12-10 | 成都数联云算科技有限公司 | Intelligent detection method and device for PCB defect types, electronic equipment and medium |
| WO2024050700A1 (en) * | 2022-09-06 | 2024-03-14 | 宁德时代新能源科技股份有限公司 | Detection method, detection apparatus and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110310260B (en) | 2021-10-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102641116B1 (en) | Method and device to recognize image and method and device to train recognition model based on data augmentation | |
| CN109297974A (en) | Information processing unit | |
| CN107392091B (en) | Agricultural artificial intelligence crop detection method, mobile terminal and computer readable medium | |
| CN108242054A (en) | A kind of steel plate defect detection method, device, equipment and server | |
| CN109060817B (en) | Artificial intelligence reinspection system and method thereof | |
| CN109685097B (en) | Image detection method and device based on GAN | |
| CN108108745A (en) | Classification method, classification module and computer program product | |
| CN106898017B (en) | The method, apparatus and terminal device of image local area for identification | |
| CN110009614A (en) | Method and apparatus for output information | |
| CN108009497A (en) | Image recognition monitoring method, system, computing device and readable storage medium storing program for executing | |
| CN109284735B (en) | Mouse feelings monitoring method, device and storage medium | |
| CN109840503B (en) | Method and device for determining category information | |
| EP3796115A1 (en) | Anomaly detection for robotic arms using vibration data | |
| CN110310260A (en) | Sub-material decision-making technique, equipment and storage medium based on machine learning model | |
| US20180197287A1 (en) | Process of using machine learning for cannabis plant health diagnostics | |
| CN112668410B (en) | Sorting behavior detection method, system, electronic device and storage medium | |
| CN111709371A (en) | Artificial intelligence based classification method, device, server and storage medium | |
| US20220366244A1 (en) | Modeling Human Behavior in Work Environment Using Neural Networks | |
| JP2023528688A (en) | Method and system for training automatic defect classification inspection equipment | |
| CN112766040A (en) | Method, device and apparatus for detecting residual bait and readable storage medium | |
| CN111310647A (en) | Generation method and device for automatic identification falling model | |
| CN110895716A (en) | Inspection apparatus and machine learning method | |
| CN113557546A (en) | Method, device, equipment and storage medium for detecting associated object in image | |
| CN113541985B (en) | Internet of things fault diagnosis method, model training method and related devices | |
| CN108509876A (en) | For the object detecting method of video, device, equipment, storage medium and program |
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 |