CN109086790A - A kind of alternative manner of disaggregated model, device and electronic equipment - Google Patents
A kind of alternative manner of disaggregated model, device and electronic equipment Download PDFInfo
- Publication number
- CN109086790A CN109086790A CN201810628785.5A CN201810628785A CN109086790A CN 109086790 A CN109086790 A CN 109086790A CN 201810628785 A CN201810628785 A CN 201810628785A CN 109086790 A CN109086790 A CN 109086790A
- Authority
- CN
- China
- Prior art keywords
- disaggregated model
- classification results
- testing image
- model
- testing
- 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.)
- Pending
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The invention discloses a kind of alternative manner of disaggregated model, device and electronic equipment, which includes: the classification results for obtaining testing image and corresponding each testing image；All testing images are stored respectively according to pre-set ratio into training set, verifying collection and test set；Reference disaggregated model is trained according to the testing image corresponding classification results in training set and training set；According to verifying collection classification results corresponding with the testing image that verifying is concentrated, the accuracy rate with reference to the reference disaggregated model after disaggregated model and training is verified, and selects accuracy rate higher as optimal with reference to disaggregated model；According to the corresponding classification results of testing image in test set and test set, the accuracy rate of the optimal current class model used with reference to disaggregated model and producing line is tested；In the case where the optimal accuracy rate with reference to disaggregated model is higher than the accuracy rate of current class model, control optimal with reference to disaggregated model replacement current class model.
Description
Technical field
The present invention relates to disaggregated model iterative technique fields, more particularly it relates to a kind of iteration of disaggregated model
Method, apparatus and electronic equipment.
Background technique
With the enhancing of deep learning image-capable, in Industrial Image Detecting field, the application of deep learning is increasingly
It is more, it is also increasingly common that image classification detection is done using trained disaggregated model.
Disaggregated model detection technique achieves breakthrough in terms of Image Information Processing in recent years, in image classification
Accuracy rate progressively reach the requirement of industrial Defect Detection so that being greatly shortened from scientific research to the practical application period.
But the process that existing disaggregated model requires manual intervention during being iterated is more, it can not be according to adopting
The image of collection is iterated automatically.
Summary of the invention
One purpose of the embodiment of the present invention is to provide a kind of new technical solution of disaggregated model automatic Iterative.
According to the first aspect of the invention, a kind of alternative manner of disaggregated model is provided, comprising:
Obtain the classification results of testing image and corresponding each testing image；
All testing images are stored respectively according to pre-set ratio to training set, verifying collection and test set
In；
According to the corresponding classification results of testing image in the training set and the training set to reference disaggregated model into
Row training；
According to verifying collection classification results corresponding with the testing image that the verifying is concentrated, verify described with reference to classification
The accuracy rate of reference disaggregated model after model and training, and select accuracy rate higher as optimal with reference to disaggregated model；
According to the corresponding classification results of testing image in the test set and the test set, the optimal reference is tested
The accuracy rate for the current class model that disaggregated model and producing line use；
In the case where the optimal accuracy rate with reference to disaggregated model is higher than the accuracy rate of the current class model, control
It makes described optimal with reference to the disaggregated model replacement current class model.
Optionally, the step of classification results for obtaining testing image and corresponding each testing image include:
It controls camera and acquires testing image；
It controls the reference disaggregated model and class tests is carried out to all testing images, obtain each described to mapping
The classification results of picture.
Optionally, the alternative manner further include:
It controls the reference disaggregated model and class tests is carried out to all testing images, obtain each described to mapping
The confidence level of the classification results of picture and corresponding each classification results；
Judge whether the confidence level is less than pre-set confidence threshold value, if so, then:
The confirmation classification for obtaining user's input is determined as the corresponding classification results of the confidence level.
Optionally, the alternative manner further include:
The classification results of testing image and corresponding each testing image are obtained according to the preset period.
Optionally, the alternative manner further include:
If the current class model is identical as the reference disaggregated model, the reference disaggregated model after training is made
Disaggregated model is referred to be optimal.
According to the second aspect of the invention, a kind of iteration means of disaggregated model are provided, comprising:
Module is obtained, for obtaining the classification results of testing image and corresponding each testing image；
Memory module, for being stored respectively according to pre-set ratio to training set, being tested by all testing images
In card collection and test set；
Training module, for according to the corresponding classification results of testing image in the training set and the training set to ginseng
Examination mark class model is trained；
Authentication module, for testing according to verifying collection classification results corresponding with the testing image that the verifying is concentrated
The accuracy rate with reference to the reference disaggregated model after disaggregated model and training is demonstrate,proved, and selects accuracy rate higher as optimal ginseng
Examination mark class model；
Test module, for surveying according to the corresponding classification results of testing image in the test set and the test set
Try the accuracy rate of the optimal current class model used with reference to disaggregated model and producing line；
Replacement module, for being higher than the accurate of the current class model in the optimal accuracy rate with reference to disaggregated model
In the case where rate, control described optimal with reference to the disaggregated model replacement current class model.
Optionally, the acquisition module includes:
Acquisition unit, for controlling camera acquisition testing image；
Test cell carries out class tests to all testing images for controlling the reference disaggregated model, obtains
The classification results of each testing image.
Optionally, the test cell is also used to control the reference disaggregated model and carries out to all testing images
Class test obtains the classification results of each testing image and the confidence level of corresponding each classification results；
The acquisition module further include:
Judging unit, for judging whether the confidence level is less than pre-set confidence threshold value；
Acquiring unit, for obtaining the confirmation of user's input in the case where the judging result of the judgment module, which is, is
Classification is determined as the corresponding classification results of the confidence level.
Optionally, the acquisition module be also used to obtain according to the preset period testing image and it is corresponding it is each to
The classification results of altimetric image.
Optionally, the iteration means further include:
Selecting module, for that will train in current class model situation identical with the reference disaggregated model
Reference disaggregated model afterwards refers to disaggregated model as optimal.
According to the third aspect of the invention we, a kind of electronic equipment is provided, including described according to a second aspect of the present invention
Iteration means.
According to the fourth aspect of the invention, a kind of electronic equipment, including memory and processor, the memory are provided
For storing instruction, described instruction is used to control the processor and executes the alternative manner described according to a first aspect of the present invention.
A beneficial effect of the invention is that disaggregated model can be realized automatically in alternative manner through the invention
Iteration reduces manual intervention, promotes user experience.
By referring to the drawings to the detailed description of exemplary embodiment of the present invention, other feature of the invention and its
Advantage will become apparent.
Detailed description of the invention
It is combined in the description and the attached drawing for constituting part of specification shows the embodiment of the present invention, and even
With its explanation together principle for explaining the present invention.
Fig. 1 is the flow diagram according to the alternative manner of the disaggregated model of an embodiment of the present invention；
Fig. 2 is the flow diagram according to the alternative manner of the disaggregated model of an embodiment of the present invention；
Fig. 3 is the flow diagram according to the alternative manner of the disaggregated model of an embodiment of the present invention；
Fig. 4 is the frame principle figure according to the iteration means of the disaggregated model of an embodiment of the present invention；
Fig. 5 is the frame principle figure according to the iteration means of the disaggregated model of an embodiment of the present invention；
Fig. 6 is the frame principle figure according to the electronic equipment of an embodiment of the present invention.
Specific embodiment
Carry out the various exemplary embodiments of detailed description of the present invention now with reference to attached drawing.It should also be noted that unless in addition having
Body explanation, the unlimited system of component and the positioned opposite of step, numerical expression and the numerical value otherwise illustrated in these embodiments is originally
The range of invention.
Be to the description only actually of at least one exemplary embodiment below it is illustrative, never as to the present invention
And its application or any restrictions used.
Technology, method and apparatus known to person of ordinary skill in the relevant may be not discussed in detail, but suitable
In the case of, the technology, method and apparatus should be considered as part of specification.
It is shown here and discuss all examples in, any occurrence should be construed as merely illustratively, without
It is as limitation.Therefore, other examples of exemplary embodiment can have different values.
It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, then in subsequent attached drawing does not need that it is further discussed.
Fig. 1 is the flow diagram according to the alternative manner of the disaggregated model of an embodiment of the present invention.
According to Fig. 1, the alternative manner the following steps are included:
Step S110 obtains the classification results of testing image and corresponding each testing image.
Wherein, testing image can be multiple.The classification results of each testing image, which can be, manually to be determined, can also be with
It is to be determined by disaggregated model.
Further, the step of executing step S110 may include step S111 and S112 as shown in Figure 2.
Step S111, control camera acquire testing image.
Step S112, control carry out class test to all testing images with reference to disaggregated model, obtain corresponding to each to be measured
The classification results of image.
Wherein it can be one or more with reference to disaggregated model.If it is multiple disaggregated models, then, this multiple classification
Model can be series connection, be also possible to parallel connection.And this multiple disaggregated model can export final point of corresponding each testing image
Class result.
Still further, step S112 can be with are as follows: control classifies to all testing images with reference to disaggregated model
Test obtains the confidence level of the classification results for corresponding to each testing image and corresponding each classification results.
On this basis, which can also include the steps that S113-S114 as shown in Figure 3.
Step S113, judges whether confidence level is less than preset confidence threshold value, if so, thening follow the steps S114；
If not, thening follow the steps S120.
Step S114 obtains classification results of the confirmation classification decision of user's input as the testing image.
For example, the classification results of a testing image are X, corresponding confidence level is 90%, and preset confidence level threshold
Value is 80%, and since the confidence level is more than confidence threshold value, then the classification results X tested reference disaggregated model is as this
The classification results of testing image.
For another example the classification results of a testing image are X, corresponding confidence level is 20%, and preset confidence level
Threshold value is 80%, since the confidence level is less than confidence threshold value, is then illustrated with reference to classification to the classification results X of testing image not
Be it is very determining, require manual intervention.So, if the confirmation classification of user's input is determined as Y, by the confirmation point of user's input
Class determines classification results of the Y as the testing image.
All testing images are stored respectively according to pre-set ratio to training set, verifying and collect and survey by step S120
Examination is concentrated.
Specifically, the allocation proportion of testing image can be preset, and the allocation proportion is stored in and executes the present invention
Alternative manner electronic equipment in.
For example, it is 1:2:3 that allocation proportion, which can be set, then, it, can in the case where the sum of testing image is 600
To be to store 100 testing images therein into training set, 200 testing images therein are stored to verifying and are concentrated,
300 testing images therein are stored into test set.
It is possible to further being to be randomly assigned testing image according to allocation proportion.
Still further, if there remains several not enough after distributing testing image according to allocation proportion to the maximum extent
The testing image of distribution then can distribute remaining testing image according to the priority of training set, verifying collection and test set.
For example, if presetting allocation proportion is 1:2:3, and in the case that the sum of testing image is 604, it can be with
It is to store 100 testing images therein into training set, 200 testing images therein is stored to verifying and are concentrated, it will
300 testing images therein are stored into test set, remaining 4 testing images.
If the priority of pre-set training set is higher than the priority of test set, the priority of test set is higher than verifying
Wherein 1 testing image then can be stored first into training set, remaining 3 testing image is stored to survey by the priority of collection
Examination is concentrated.
If the priority of pre-set test set is higher than the priority of verifying collection, the priority for verifying collection is higher than training
The priority of collection then can first store 3 testing images therein into test set, and a remaining testing image is stored
It is concentrated to verifying.
Step S130 carries out reference disaggregated model according to the corresponding classification results of testing image in training set and training set
Training.
Specifically, according to the testing image and the corresponding classification results of each testing image for including in training set to reference point
Class model is trained, the reference disaggregated model after being trained.
If with reference to disaggregated model being the series connection of disaggregated model A and disaggregated model B, need to disaggregated model A and classification mould
Type B connects again after being trained respectively, the reference disaggregated model after being trained.
If with reference to disaggregated model being the parallel connection of disaggregated model A and disaggregated model B, need to disaggregated model A and classification mould
Type B carries out parallel connection after being trained respectively again, the reference disaggregated model after being trained.
Step S140, according to verifying collection classification results corresponding with verifying concentration testing image, verifying refers to disaggregated model
With the accuracy rate of the reference disaggregated model after training, and select accuracy rate higher as optimal with reference to disaggregated model.
Specifically, concentrating all testing images and the corresponding classification results of each testing image for including according to verifying, test
Card refers to the accuracy rate of disaggregated model and the accuracy rate of the reference disaggregated model after training.If with reference to the accurate of disaggregated model
Rate is higher than the accuracy rate of the reference disaggregated model after training, then will refer to disaggregated model as optimal with reference to disaggregated model.If
The accuracy rate of reference disaggregated model after training is higher than the accuracy rate with reference to disaggregated model, then by the reference disaggregated model after training
Disaggregated model is referred to as optimal.
Step S150 is tested optimal with reference to classification according to the corresponding classification results of testing image in test set and test set
The accuracy rate for the current class model that model and producing line use.
Specifically, the current class model that producing line uses is possible different from reference to disaggregated model.Include according to test set
All testing images and the corresponding classification results of each testing image test the accuracy rate of current class model and with reference to classification mould
The accuracy rate of type.
Step S160, in the case where the optimal accuracy rate with reference to disaggregated model is higher than the accuracy rate of current class model,
It controls optimal with reference to disaggregated model replacement current class model.
If the optimal accuracy rate with reference to disaggregated model is higher than the accuracy rate of current class model, illustrate optimal reference point
Class model is better than current class model, current class model can be replaced with reference to disaggregated model with optimal.
Specifically, in the test for executing the electronic equipment of alternative manner of the present invention and being tested using current class model
In the case that equipment is different, controlling the optimal mode with reference to disaggregated model replacement current class model be may is that optimal reference
Disaggregated model is issued in the test equipment tested using current class model, to replace current class model, to realize
The iteration of disaggregated model.
In the case where the test equipment tested using current class model executes alternative manner of the present invention, control is most
It is excellent with reference to disaggregated model replacement current class model mode may is that control test equipment using it is optimal with reference to disaggregated model into
Row follow-up test.
If optimal classification is less than or equal to the accuracy rate of current class model with reference to the accuracy rate of disaggregated model, illustrate
It is optimal to be not better than current class model with reference to disaggregated model, current class model can be continued to use and tested, without pair
Current class model is replaced.
In this way, alternative manner through the invention, so that it may realize that the automatic of the current class model used producing line changes
In generation, can be further reduced manual intervention.
In one example, the current class model that producing line uses can be identical with reference disaggregated model.It is possible to take
Disappear verifying collection, and refers to disaggregated model using the reference disaggregated model after training as optimal.
Specifically, can be in above-mentioned steps S120 and deposit all testing images respectively according to preset ratio
It puts into training set and test set.Wherein, preset ratio can be, but not limited to as 1:1.
Above-mentioned steps S140 can be with are as follows: refers to disaggregated model using the reference disaggregated model after training as optimal.
Alternatively, above-mentioned steps S140 and step S150 can be merged into a step, it may be assumed that according to the institute for including in test set
There are testing image and the corresponding classification results of each testing image, the accuracy rate and current class model of test reference disaggregated model
Accuracy rate.Above-mentioned steps S160 then can be with are as follows: if the accuracy rate with reference to disaggregated model is higher than the accurate of current class model
Rate then controls and replaces current class model with reference to disaggregated model.
In one example, it can be and execute above-mentioned alternative manner according to the preset period.The preset period
Such as it can be, but not limited to be one day or one week.
In this way, alternative manner through the invention, so that it may be carried out automatically to disaggregated model according to the preset period
Iteration.
It corresponds to the above method, the present invention also provides a kind of iteration means of disaggregated model.Fig. 4 is according to this hair
A kind of frame principle figure of the iteration means of the disaggregated model of bright embodiment.
According to Fig.4, which includes obtaining module 410, memory module 420, training module 430, verifying mould
Block 440, test module 450 and replacement module 460.
The acquisition module 410 is used to obtain the classification results of testing image and corresponding each testing image.
The memory module 420 by all testing images according to pre-set ratio for storing respectively to training set, testing
In card collection and test set.
The training module 430 is used for according to the corresponding classification results of testing image in training set and training set to reference point
Class model is trained.
The authentication module 440 is used for according to verifying collection classification results corresponding with the testing image that verifying is concentrated, verifying ginseng
The accuracy rate of reference disaggregated model after examination mark class model and training, and select accuracy rate higher as optimal with reference to classification mould
Type.
The test module 450 is used for according to the corresponding classification results of testing image in test set and test set, and test is most
The accuracy rate of the excellent current class model used with reference to disaggregated model and producing line.
The replacement module 460 is used to be higher than the accuracy rate of current class model in the optimal accuracy rate with reference to disaggregated model
In the case of, it controls optimal with reference to disaggregated model replacement current class model.
Further, as shown in figure 5, obtaining module 410 may include acquisition unit 411 and test cell 412.The acquisition
Unit 411 is for controlling camera acquisition testing image；The test cell 412 is for controlling with reference to disaggregated model to all to mapping
As carrying out class test, the classification results of each testing image are obtained.
Yet further test cell 411, which is also used to control, carries out classification survey to all testing images with reference to disaggregated model
Examination obtains the classification results of each testing image and the confidence level of corresponding each classification results.
On this basis, as shown in figure 5, the acquisition module 410 can also include judging unit 413 and acquisition module 414.
The judging unit 413 is for judging whether confidence level is less than pre-set confidence threshold value.The acquisition module 414 is for sentencing
The judging result of disconnected module is that the confirmation for obtaining user's input, which is classified, to be determined as the corresponding classification knot of confidence level in the case where being
Fruit.
In one example, module 410 is obtained to be also used to obtain testing image according to the preset period and correspond to every
The classification results of one testing image.
In one example, iteration means further include selecting module 510 as shown in Figure 5, in current class model
In situation identical with reference disaggregated model, disaggregated model is referred to using the reference disaggregated model after training as optimal.
The present invention also provides a kind of electronic equipment, and on the one hand, which includes changing for disaggregated model above-mentioned
For device.
Fig. 6 is the frame principle figure according to the implementation structure of the electronic equipment of another aspect of the present invention.
According to Fig.6, which includes memory 601 and processor 602, and the memory 601 is for storing
Instruction, the instruction are operated for control processor 602 to execute the alternative manner of above-mentioned disaggregated model.
The processor 602 is such as can be central processor CPU, Micro-processor MCV.The memory 601 for example including
ROM (read-only memory), RAM (random access memory), nonvolatile memory of hard disk etc..
In addition to this, according to Fig.6, which further includes interface arrangement 603, input unit 604, display
Device 605, communication device 606, loudspeaker 607, camera 608 etc..Although multiple devices are shown in FIG. 6, this hair
Bright electronic equipment can only relate to partial devices therein, for example, processor 601, memory 602 and camera 608 etc..
Above-mentioned communication device 606 has for example been able to carry out wired or wireless communication.
Above-mentioned interface arrangement 603 is for example including earphone jack, USB interface etc..
Above-mentioned input unit 604 is such as may include touch screen, key.
Above-mentioned display device 605 is, for example, liquid crystal display, touch display screen etc..
The difference of the various embodiments described above primary focus description and other embodiments, but those skilled in the art should be clear
Chu, the various embodiments described above can according to need exclusive use or are combined with each other.
All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment
Divide cross-reference, each embodiment focuses on the differences from other embodiments, but those skilled in the art
Member is it should be understood that the various embodiments described above can according to need exclusive use or be combined with each other.In addition, for device
For embodiment, since it is corresponding with embodiment of the method, so describing fairly simple, related place is implemented referring to method
The explanation of the corresponding part of example.System embodiment described above is only schematical, wherein being used as separation unit
The module of explanation may or may not be physically separated.
The present invention can be system, method and/or computer program product.Computer program product may include computer
Readable storage medium storing program for executing, containing for making processor realize the computer-readable program instructions of various aspects of the invention.
Computer readable storage medium, which can be, can keep and store the tangible of the instruction used by instruction execution equipment
Equipment.Computer readable storage medium for example can be-- but it is not limited to-- storage device electric, magnetic storage apparatus, optical storage
Equipment, electric magnetic storage apparatus, semiconductor memory apparatus or above-mentioned any appropriate combination.Computer readable storage medium
More specific example (non exhaustive list) includes: portable computer diskette, hard disk, random access memory (RAM), read-only deposits
It is reservoir (ROM), erasable programmable read only memory (EPROM or flash memory), static random access memory (SRAM), portable
Compact disk read-only memory (CD-ROM), digital versatile disc (DVD), memory stick, floppy disk, mechanical coding equipment, for example thereon
It is stored with punch card or groove internal projection structure and the above-mentioned any appropriate combination of instruction.Calculating used herein above
Machine readable storage medium storing program for executing is not interpreted that instantaneous signal itself, the electromagnetic wave of such as radio wave or other Free propagations lead to
It crosses the electromagnetic wave (for example, the light pulse for passing through fiber optic cables) of waveguide or the propagation of other transmission mediums or is transmitted by electric wire
Electric signal.
Computer-readable program instructions as described herein can be downloaded to from computer readable storage medium it is each calculate/
Processing equipment, or outer computer or outer is downloaded to by network, such as internet, local area network, wide area network and/or wireless network
Portion stores equipment.Network may include copper transmission cable, optical fiber transmission, wireless transmission, router, firewall, interchanger, gateway
Computer and/or Edge Server.Adapter or network interface in each calculating/processing equipment are received from network to be counted
Calculation machine readable program instructions, and the computer-readable program instructions are forwarded, for the meter being stored in each calculating/processing equipment
In calculation machine readable storage medium storing program for executing.
Computer program instructions for executing operation of the present invention can be assembly instruction, instruction set architecture (ISA) instructs,
Machine instruction, machine-dependent instructions, microcode, firmware instructions, condition setup data or with one or more programming languages
The source code or object code that any combination is write, the programming language include the programming language-of object-oriented such as
Python, java, C++ etc., and conventional procedural programming languages-such as " C " language or similar programming language.It calculates
Machine readable program instructions can be executed fully on the user computer, partly be executed on the user computer, as one
Independent software package executes, part executes on the remote computer or completely in remote computation on the user computer for part
It is executed on machine or server.In situations involving remote computers, remote computer can pass through the network-of any kind
It is connected to subscriber computer including local area network (LAN) or wide area network (WAN)-, or, it may be connected to outer computer (such as
It is connected using ISP by internet).In some embodiments, by being referred to using computer-readable program
The status information of order carrys out personalized customization electronic circuit, for example, programmable logic circuit, field programmable gate array (FPGA) or
Programmable logic array (PLA), the electronic circuit can execute computer-readable program instructions, to realize of the invention each
Aspect.
Referring herein to according to the method for the embodiment of the present invention, the flow chart of device (system) and computer program product and/
Or block diagram describes various aspects of the invention.It should be appreciated that flowchart and or block diagram each box and flow chart and/
Or in block diagram each box combination, can be realized by computer-readable program instructions.
These computer-readable program instructions can be supplied to general purpose computer, special purpose computer or other programmable datas
The processor of processing unit, so that a kind of machine is produced, so that these instructions are passing through computer or other programmable datas
When the processor of processing unit executes, function specified in one or more boxes in implementation flow chart and/or block diagram is produced
The device of energy/movement.These computer-readable program instructions can also be stored in a computer-readable storage medium, these refer to
It enables so that computer, programmable data processing unit and/or other equipment work in a specific way, thus, it is stored with instruction
Computer-readable medium then includes a manufacture comprising in one or more boxes in implementation flow chart and/or block diagram
The instruction of the various aspects of defined function action.
Computer-readable program instructions can also be loaded into computer, other programmable data processing units or other
In equipment, so that series of operation steps are executed in computer, other programmable data processing units or other equipment, to produce
Raw computer implemented process, so that executed in computer, other programmable data processing units or other equipment
Instruct function action specified in one or more boxes in implementation flow chart and/or block diagram.
The flow chart and block diagram in the drawings show the system of multiple embodiments according to the present invention, method and computer journeys
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
One module of table, program segment or a part of instruction, the module, program segment or a part of instruction include one or more use
The executable instruction of the logic function as defined in realizing.In some implementations as replacements, function marked in the box
It can occur in a different order than that indicated in the drawings.For example, two continuous boxes can actually be held substantially in parallel
Row, they can also be executed in the opposite order sometimes, and this depends on the function involved.It is also noted that block diagram and/or
The combination of each box in flow chart and the box in block diagram and or flow chart, can the function as defined in executing or dynamic
The dedicated hardware based system made is realized, or can be realized using a combination of dedicated hardware and computer instructions.It is right
For those skilled in the art it is well known that, by hardware mode realize, by software mode realize and pass through software and
It is all of equal value that the mode of combination of hardware, which is realized,.
Various embodiments of the present invention are described above, above description is exemplary, and non-exclusive, and
It is not limited to disclosed each embodiment.Without departing from the scope and spirit of illustrated each embodiment, for this skill
Many modifications and changes are obvious for the those of ordinary skill in art field.The selection of term used herein, purport
In principle, the practical application or to the technological improvement in market for best explaining each embodiment, or make the art its
Its those of ordinary skill can understand each embodiment disclosed herein.The scope of the present invention is defined by the appended claims.
Claims (10)
1. a kind of alternative manner of disaggregated model characterized by comprising
Obtain the classification results of testing image and corresponding each testing image；
All testing images are stored respectively according to pre-set ratio into training set, verifying collection and test set；
Reference disaggregated model is instructed according to the testing image corresponding classification results in the training set and the training set
Practice；
According to verifying collection classification results corresponding with the testing image that the verifying is concentrated, verifying is described to refer to disaggregated model
With the accuracy rate of the reference disaggregated model after training, and select accuracy rate higher as optimal with reference to disaggregated model；
According to the corresponding classification results of testing image in the test set and the test set, test described optimal with reference to classification
The accuracy rate for the current class model that model and producing line use；
In the case where the optimal accuracy rate with reference to disaggregated model is higher than the accuracy rate of the current class model, institute is controlled
It states optimal with reference to the disaggregated model replacement current class model.
2. alternative manner according to claim 1, which is characterized in that the acquisition testing image and correspondence are each to mapping
The step of classification results of picture includes:
It controls camera and acquires testing image；
It controls the reference disaggregated model and class tests is carried out to all testing images, obtain each testing image
Classification results.
3. alternative manner according to claim 2, which is characterized in that the alternative manner further include:
It controls the reference disaggregated model and class tests is carried out to all testing images, obtain each testing image
The confidence level of classification results and corresponding each classification results；
Judge whether the confidence level is less than pre-set confidence threshold value, if so, then:
The confirmation classification for obtaining user's input is determined as the corresponding classification results of the confidence level.
4. alternative manner according to claim 1, which is characterized in that the alternative manner further include:
The classification results of testing image and corresponding each testing image are obtained according to the preset period.
5. alternative manner according to claim 1, which is characterized in that the alternative manner further include:
If the current class model is identical as the reference disaggregated model, using the reference disaggregated model after training as most
It is excellent to refer to disaggregated model.
6. a kind of iteration means of disaggregated model characterized by comprising
Module is obtained, for obtaining the classification results of testing image and corresponding each testing image；
Memory module collects for being stored all testing images respectively according to pre-set ratio to training set, verifying
In test set；
Training module, for being divided according to the corresponding classification results of testing image in the training set and the training set reference
Class model is trained；
Authentication module, for verifying institute according to verifying collection classification results corresponding with the testing image that the verifying is concentrated
The accuracy rate with reference to the reference disaggregated model after disaggregated model and training is stated, and selects accuracy rate higher as optimal reference point
Class model；
Test module, for testing institute according to the corresponding classification results of testing image in the test set and the test set
State the accuracy rate of the optimal current class model used with reference to disaggregated model and producing line；
Replacement module, for being higher than the accuracy rate of the current class model in the optimal accuracy rate with reference to disaggregated model
In the case of, it controls described optimal with reference to the disaggregated model replacement current class model.
7. iteration means according to claim 6, which is characterized in that the acquisition module includes:
Acquisition unit, for controlling camera acquisition testing image；
Test cell carries out class tests to all testing images for controlling the reference disaggregated model, obtains each
The classification results of the testing image.
8. iteration means according to claim 7, which is characterized in that the test cell is also used to control the reference point
Class model carries out class test to all testing images, and classification results and the correspondence for obtaining each testing image are each
The confidence level of classification results；
The acquisition module further include:
Judging unit, for judging whether the confidence level is less than pre-set confidence threshold value；
Acquiring unit, in the case where the judging result of the judgment module, which is, is, obtaining the confirmation classification of user's input
It determines as the corresponding classification results of the confidence level.
9. a kind of electronic equipment, which is characterized in that including iteration means a method according to any one of claims 6-8.
10. a kind of electronic equipment, which is characterized in that including memory and processor, the memory is for storing instruction, described
Instruction executes alternative manner according to any one of claims 1-5 for controlling the processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810628785.5A CN109086790A (en) | 2018-06-19 | 2018-06-19 | A kind of alternative manner of disaggregated model, device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810628785.5A CN109086790A (en) | 2018-06-19 | 2018-06-19 | A kind of alternative manner of disaggregated model, device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109086790A true CN109086790A (en) | 2018-12-25 |
Family
ID=64840021
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810628785.5A Pending CN109086790A (en) | 2018-06-19 | 2018-06-19 | A kind of alternative manner of disaggregated model, device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109086790A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111046929A (en) * | 2019-11-28 | 2020-04-21 | 北京金山云网络技术有限公司 | Method and device for analyzing model error cases and electronic equipment |
| CN111382757A (en) * | 2018-12-28 | 2020-07-07 | 深圳市优必选科技有限公司 | Method for dynamically adjusting training samples in deep learning classification algorithm and terminal equipment |
| CN112232417A (en) * | 2020-10-16 | 2021-01-15 | 北京紫光展锐通信技术有限公司 | Classification method and device, storage medium and terminal |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107633255A (en) * | 2017-08-11 | 2018-01-26 | 天津大学 | A kind of rock lithology automatic recognition classification method under deep learning pattern |
| CN108009525A (en) * | 2017-12-25 | 2018-05-08 | 北京航空航天大学 | A kind of specific objective recognition methods over the ground of the unmanned plane based on convolutional neural networks |
-
2018
- 2018-06-19 CN CN201810628785.5A patent/CN109086790A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107633255A (en) * | 2017-08-11 | 2018-01-26 | 天津大学 | A kind of rock lithology automatic recognition classification method under deep learning pattern |
| CN108009525A (en) * | 2017-12-25 | 2018-05-08 | 北京航空航天大学 | A kind of specific objective recognition methods over the ground of the unmanned plane based on convolutional neural networks |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111382757A (en) * | 2018-12-28 | 2020-07-07 | 深圳市优必选科技有限公司 | Method for dynamically adjusting training samples in deep learning classification algorithm and terminal equipment |
| CN111382757B (en) * | 2018-12-28 | 2023-09-08 | 深圳市优必选科技有限公司 | Method and terminal equipment for dynamically adjusting training samples in deep learning classification algorithm |
| CN111046929A (en) * | 2019-11-28 | 2020-04-21 | 北京金山云网络技术有限公司 | Method and device for analyzing model error cases and electronic equipment |
| CN111046929B (en) * | 2019-11-28 | 2023-09-26 | 北京金山云网络技术有限公司 | Analysis method and device for model error cases and electronic equipment |
| CN112232417A (en) * | 2020-10-16 | 2021-01-15 | 北京紫光展锐通信技术有限公司 | Classification method and device, storage medium and terminal |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106290378B (en) | Defect classification method and defect inspecting system | |
| US11983394B2 (en) | System and method for generating photorealistic synthetic images based on semantic information | |
| CN109086790A (en) | A kind of alternative manner of disaggregated model, device and electronic equipment | |
| CN109949286A (en) | Method and apparatus for output information | |
| CN105554782B (en) | The prediction technique and device of user's perception index | |
| TW201947463A (en) | Model test method and device | |
| CN105893253A (en) | Application software testing method, device and equipment | |
| US20210004723A1 (en) | Learning device, learning method, and learning program | |
| JP6780769B2 (en) | Learning equipment, learning methods and learning programs | |
| US10424059B2 (en) | Quality evaluation | |
| CN113868102A (en) | Method, electronic device and computer program product for information display | |
| US20230048386A1 (en) | Method for detecting defect and method for training model | |
| CN112181698A (en) | Method, device and equipment for testing automatic driving limit performance and storage medium | |
| CN110348522A (en) | A kind of image detection recognition methods and system, electronic equipment, image classification network optimized approach and system | |
| CN108734296A (en) | Optimize method, apparatus, electronic equipment and the medium of the training data of supervised learning | |
| KR20200018411A (en) | Method and apparatus for detecting burr of electrode piece | |
| CN112017245B (en) | Certificate positioning method | |
| CN113688957A (en) | Target detection method, device, equipment and medium based on multi-model fusion | |
| CN109979531A (en) | A kind of genetic mutation recognition methods, device and storage medium | |
| CN110443824A (en) | Method and apparatus for generating information | |
| CN106842224A (en) | For the mounting structure detection method and device of the range sensor of electronic equipment | |
| CN108717520A (en) | A kind of pedestrian recognition methods and device again | |
| CN109743286A (en) | A kind of IP type mark method and apparatus based on figure convolutional neural networks | |
| CN110339567A (en) | System resource configuration, scene prediction model training method and device | |
| AU2017208235A1 (en) | Relative position encoding based networks for action recognition |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181225 |