CN109272497A - Method for detecting surface defects of products, device and computer equipment - Google Patents
Method for detecting surface defects of products, device and computer equipment Download PDFInfo
- Publication number
- CN109272497A CN109272497A CN201811037591.4A CN201811037591A CN109272497A CN 109272497 A CN109272497 A CN 109272497A CN 201811037591 A CN201811037591 A CN 201811037591A CN 109272497 A CN109272497 A CN 109272497A
- Authority
- CN
- China
- Prior art keywords
- chromosome
- network model
- fitness value
- nervus opticus
- detected
- 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
- 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
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Genetics & Genomics (AREA)
- Physiology (AREA)
- Signal Processing (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
This application discloses method for detecting surface defects of products and device, computer equipment and computer-readable medium, this method includes obtaining image to be detected comprising product surface；Described image to be detected is detected using first nerves network model, the product surface for being included with the described image to be detected of determination is with the presence or absence of defect；Wherein, the first nerves network model carries out compression processing acquisition to trained nervus opticus network model using genetic algorithm, the nervus opticus network model is obtained using preset training sample training, and the first nerves network model is not less than default precision based on the precision of the training sample.This method and device, to product surface progress defects detection, have the advantages that calculation amount and memory space are lower, can be applied to the equipment stored and computing resource is all limited using by the neural network model after genetic algorithm compression processing.
Description
Technical field
This application involves computer application technology more particularly to a kind of method for detecting surface defects of products and device,
Computer equipment and computer-readable medium.
Background technique
In recent years, with the development of artificial intelligence, neural network (Neural Network, NN) algorithm is widely used in
Surface defects of products detection field, such as fabric defects detection, electronic component surface defects detection etc..However, detection effect
Preferable deep neural network often has the biggish node of quantity (neuron) and a model parameter, it is not only computationally intensive and also
Model occupies larger a part of space in actual deployment, limits its application the equipment all limited in storage and computing resource.
Summary of the invention
In view of problem above, the embodiment of the present invention provides a kind of method for detecting surface defects of products and device, computer
Equipment and computer-readable medium, be used for surface defects of products detection neural network model calculation amount and memory space compared with
It is low, it can be applied to the equipment stored and computing resource is all limited.
The method for detecting surface defects of products of embodiment according to the invention, comprising: obtain to be checked comprising product surface
Altimetric image；Described image to be detected is detected using first nerves network model, is wrapped with the described image to be detected of determination
The product surface contained whether there is defect；Wherein, the first nerves network model is using genetic algorithm to trained
Nervus opticus network model carries out compression processing acquisition, and the nervus opticus network model is instructed using preset training sample
It gets, the first nerves network model is not less than default precision based on the precision of the training sample.
The surface defects of products detection device of embodiment according to the invention, comprising: module is obtained, for obtaining comprising producing
The image to be detected on product surface；Detection module, for being detected using first nerves network model to described image to be detected,
The product surface for being included with the described image to be detected of determination is with the presence or absence of defect；Wherein, the first nerves network model is
Compression processing acquisition, the nervus opticus network mould are carried out to trained nervus opticus network model using genetic algorithm
Type is obtained using preset training sample training, and the precision of the first nerves network model based on the training sample is not
Lower than default precision.
The computer equipment of embodiment according to the invention, comprising: processor；And memory, being stored thereon with can hold
Row instruction, wherein the executable instruction makes the processor execute method above-mentioned upon being performed.
The computer-readable medium of embodiment according to the invention, is stored thereon with executable instruction, wherein described to hold
Row instruction makes computer execute method above-mentioned upon being performed.
It can be seen from the above that the scheme of the embodiment of the present invention is utilized by after genetic algorithm compression processing
First nerves network model carries out defects detection to product surface, has the advantages that calculation amount and memory space are lower, can apply
In all limited equipment of storage and computing resource.Meanwhile the scheme of the embodiment of the present invention can combine accuracy in detection
With two aspects of compression.
Detailed description of the invention
Fig. 1 is the flow chart of the method for detecting surface defects of products of one embodiment according to the invention；
Fig. 2 is the utilization genetic algorithm of one embodiment according to the invention to trained nervus opticus network model
Carry out the flow chart of the method for compression processing；
Fig. 2 a is the exemplary diagram of a neural network structure；
Fig. 3 is the schematic diagram of the surface defects of products detection device of one embodiment according to the invention；
Fig. 4 is the schematic diagram of the computer equipment of one embodiment according to the invention；
Fig. 5 is the illustrative computer for being suitable for being used to realize embodiment of the present invention of one embodiment according to the invention
The block diagram of equipment.
Specific embodiment
Theme described herein is discussed referring now to example embodiment.It should be understood that discussing these embodiments only
It is in order to enable those skilled in the art can better understand that being not to claim to realize theme described herein
Protection scope, applicability or the exemplary limitation illustrated in book.It can be in the protection scope for not departing from present disclosure
In the case of, the function and arrangement of the element discussed are changed.Each example can according to need, omit, substitute or
Add various processes or component.For example, described method can be executed according to described order in a different order, with
And each step can be added, omits or combine.In addition, feature described in relatively some examples is in other examples
It can be combined.
As used in this article, term " includes " and its modification indicate open term, are meant that " including but not limited to ".
Term "based" indicates " being based at least partially on ".Term " one embodiment " and " embodiment " expression " at least one implementation
Example ".Term " another embodiment " expression " at least one other embodiment ".Term " first ", " second " etc. may refer to not
Same or identical object.Here may include other definition, either specific or implicit.Unless bright in context
It really indicates, otherwise the definition of a term is consistent throughout the specification.
The embodiment of the present invention is used through the first nerves network model after genetic algorithm compression processing come to product
Surface is detected, and does briefly introduction to genetic algorithm and neural network below.
Genetic algorithm (Genetic Algorithm, i.e. GA) be it is a kind of use for reference living nature evolution laws (survival of the fittest,
Select the superior and eliminate the inferior genetic mechanism) develop randomization searching method.It is to be taught by the J.Holland in the U.S. in head in 1975
It first proposes, is mainly characterized by directly operating structure objects, there is no the restrictions of derivation and function continuity；With interior
Implicit Parallelism and better global optimizing ability；Using the optimization method of randomization, optimization can be obtained and instructed automatically
Search space is adaptively adjusted the direction of search, does not need determining rule.These properties of genetic algorithm are wide by people
It is applied to the fields such as Combinatorial Optimization, machine learning, signal processing, self adaptive control and artificial life generally.It is modern related
Key technology in intelligence computation.
Neural network (NeuralNetwork, i.e. NN), the research that artificial intelligence field rises since being the 1980s
Hot spot.It is abstracted human brain neuroid from information processing angle, establishes certain naive model, by different connection sides
Formula forms different networks.Neural network is a kind of operational model, by being coupled to each other between a large amount of node (or neuron)
It constitutes.A kind of each specific output function of node on behalf, referred to as excitation function (activation function).Every two
Connection between node all represents a weighted value for passing through the connection signal, referred to as connection weight.The output of network then according to
The difference of the connection type of network, connection weight and excitation function and it is different.The structural information of neural network includes node and connection
The information such as power.
Fig. 1 shows the flow chart of the method for detecting surface defects of products of one embodiment according to the invention.Shown in Fig. 1
Method 100 can by computer or other suitably there is the electronic equipment of computing capability to realize.In addition, art technology
Personnel will be understood that, execute any system of method 100 all in the scope of embodiments of the invention and spirit.
As shown in Figure 1, obtaining image to be detected comprising product surface in step S102.In the present embodiment, specific implementation
When, product surface to be detected can be shot by CCD industrial camera, and transmit to image data by interchanger,
To obtain above-mentioned image to be detected.Wherein, above-mentioned CCD (Charge Coupled Device, photosensitive coupling component) is digital phase
For recording the semiconductor subassembly of light variation in machine.
In the present embodiment, when it is implemented, the size of above-mentioned image to be detected can specifically require as 256*256.
It is of course also possible to carry out image procossing after obtaining above-mentioned image to be checked to the image to be checked of acquisition, obtain the size that meets the requirements
Image to be checked.
In one embodiment, the said goods specifically can be fabric, steel plate, glass, magnetic shoe, electronic product, workpiece,
The products such as plastic products, timber.Certainly, it should be noted that above-mentioned cited product type is intended merely to be better described
Embodiments of the present invention.When it is implemented, can also select as the case may be with detection demand to above-mentioned cited production
Other products other than product carry out the detection of corresponding surface defect.
In step S104, described image to be detected is detected using first nerves network model, with determine it is described to
The product surface that detection image is included whether there is defect；Wherein, the first nerves network model is to utilize genetic algorithm
Compression processing acquisition is carried out to trained nervus opticus network model, the nervus opticus network model is using preset
Training sample training obtain, the first nerves network model is based on the precision of the training sample not less than default essence
Degree.
In the present embodiment, when it is implemented, can be executed to image to be detected pre- when to be detected to image to be detected
Processing, such as, but not limited to, is converted to gray level image etc. for image to be detected.Certainly, the present invention is not limited thereto, in this hair
In bright some other embodiment, in the case where image to be detected has been adapted in the initial state using model to detect,
Pretreatment can not be executed to image to be detected.
In the present embodiment, when it is implemented, when to detect to image to be detected knowledge can be executed to image to be detected
Ding Wei not be with image dividing processing, the one or more for obtaining image to be detected by identification positioning and image dividing processing is candidate
Region, by being detected to candidate region to determine that the product surface that candidate region is included whether there is defect.Identification is fixed
Position and image dividing processing are known technologies, omit descriptions thereof herein.
The present embodiment is using genetic algorithm to having trained for the first nerves network model of testing product surface defect
It is obtained after good nervus opticus network model progress compression processing.Wherein, the present embodiment is using genetic algorithm to nervus opticus
The principle that network model carries out compression processing is to take into account neural network model essence according to the principle of genetic algorithm " survival of the fittest "
In the case where degree, using " compression neural network model " as criterion, trained nervus opticus network model is executed various
Genetic manipulation finally obtains the first nerves network model that structure simplifies.When it is implemented, can be by setting one based on preset
The default precision of training sample constrains the compression process to nervus opticus network model, wherein nervus opticus network mould
Type is obtained using the preset training sample training, which can be the original essence of nervus opticus network model
Degree, or the numerical value of the slightly below original precision.The default precision, which can be, to be artificially arranged, and above-mentioned electronic equipment base is also possible to
It is arranged in preset algorithm, and the default precision can be adjusted according to actual needs, the present embodiment is not to this
Aspect does any restriction.When it is implemented, using the preset training sample to the neural network model after compression processing into
Row training, to obtain first nerves network model.
In some optional implementations of the present embodiment, compression processing includes deleting neural network model to be compressed extremely
A few node and its corresponding connection, and/or, delete at least one connection of neural network model to be compressed, with reduce to
The network complexity for compressing neural network model, that is, improve the network reduction degree of neural network model to be compressed.Preferably, it compresses
At least one node and its connection accordingly that processing includes deletion neural network model hidden layer to be compressed, and/or, delete institute
State at least one connection of hidden layer.Wherein, hidden layer (hidden layer, be also hidden layer) refers to except input layer (input
) and other each layers other than output layer (output layer) layer.
In some optional implementations of the present embodiment, training sample may include multiple normal pictures of product and more
A problem image, wherein normal picture refers to that the image of defect is not present in product surface, and problem image refers to that product surface exists
The image of defect.Being trained using training sample to neural network model is known technology, omits descriptions thereof herein.
It is described to utilize genetic algorithm to trained nervus opticus network model in an embodiment of method 100
Carry out compression processing, comprising: using the fitness value based on compression as standard, to dye corresponding to the nervus opticus network model
Colour solid individual executes genetic manipulation, to generate the optimal chromosome of fitness value；Using the training sample to described suitable
Neural network model corresponding to the chromosome for answering angle value optimal is trained, to obtain the first nerves network mould
Type.
In the present embodiment, the fitness value based on compression, which refers to, can reflect the suitable of network reduction degree (or network complexity)
Angle value is answered, such as can be that fitness value is bigger, and network reduction degree is higher, that is, realizes and is effectively compressed；Fitness value is smaller, net
Network simplification degree is lower, that is, is not carried out and is effectively compressed.It is optional then when being compressed using genetic algorithm to neural network model
It selects the big chromosome of fitness value and executes genetic manipulation, finally the fitness in the chromosome that N-Generation group generates
Being worth maximum chromosome is optimal chromosome.It should be noted that in other embodiments of the invention, it can also
With bigger using fitness value, network complexity is higher, that is, is not carried out and is effectively compressed；Fitness value is smaller, network complexity
It is lower, that is, it realizes and is effectively compressed, then when compressing using genetic algorithm to neural network model, fitness value may be selected
Small chromosome executes genetic manipulation, finally the smallest dye of fitness value in the chromosome that N-Generation group generates
Colour solid individual is optimal chromosome.
In some optional implementation of the present embodiment, following fitness function can be used and calculate fitness value:
Or
Wherein, f (i, t) indicates the fitness of i-th of body in t generation；E (i, t) indicates that i-th of body in t generation is corresponding
The network error of neural network model；H (i, t) indicates the network reduction degree of i-th of body in t generation.
In some optional implementations of the present embodiment, following formula calculating is can be used in E (i, t):
Wherein,Respectively the corresponding neural network model of i-th of body in t generation is based on q-th preset of instruction
Practice the desired output and real output value of sample.
Following formula calculating can be used in H (i, t):
Wherein, m (i, t) is the node number of i-th of body in t generation.Network structure is more simplified, and network reduction angle value is got over
Greatly.
In the implementation, the compression processing mistake to neural network model to be compressed is constrained using network error E (i, t)
Journey can combine precision and compression.Network error E (i, t) is smaller, then the precision of the neural network model after compression processing is got over
It is high.Network reduction angle value is bigger, then the structure of the neural network model after compression processing is more simplified.Therefore, in present embodiment
In, the chromosome that network error is smaller, network reduction degree is bigger, fitness value is bigger.
In the present embodiment, decoding operate is executed to optimal chromosome, the optimal net of neural network model can be obtained
Network structure.In some optional implementations of the present embodiment, the first nerves network model that is obtained after to compression processing into
When row training, first nerves network model can be finely adjusted (fine-tuning).Precision will can be slightly below preset in this way
Neural network model finely tune to meeting default required precision.
It can be seen from the above that scheme provided by the embodiment of the present invention, which utilizes, passes through genetic algorithm compression processing
First nerves network model afterwards carries out defects detection to product surface, has the advantages that calculation amount and memory space are lower, can
The equipment all limited applied to storage and computing resource.Meanwhile the scheme of the embodiment of the present invention can combine detection standard
Two aspects of exactness and compression.
Fig. 2 shows one embodiment according to the invention to chromosome corresponding to the nervus opticus network model
Individual executes genetic manipulation, to generate the flow chart of the method for the optimal chromosome of fitness value, method shown in Fig. 2
200 can by computer or other suitably there is the electronic equipment of computing capability to realize.In addition, those skilled in the art will
Understand, executes any system of method 200 all in the scope of embodiments of the invention and spirit.
As shown in Fig. 2, obtaining the structural information of trained nervus opticus network model in step S202.Wherein, institute
Stating nervus opticus network model is to be trained on previously preset training sample, and training obtains making nervus opticus network
The precision of model meets default precision.The nervus opticus network model of the embodiment of the present invention can be convolutional neural networks (CNN:
Convolutional Neural Network) model, the convolutional neural networks (RCNN:Region based on area information
Based Convolutional Neural Network) model, Recognition with Recurrent Neural Network (RNN:Recurrent Neural
Network) model, shot and long term memory models (LSTM:Long Short-Term Memory) or gating cycle unit (GRU:
Gated Recurrent Unit), further, it is also possible to be other kinds of neural network model or combined by a variety of neural networks
Cascaded neural network model.The structural information of neural network model includes nodal information and connection weight information, and network structure can
It is indicated by a connection matrix, such as a N N matrix C=(cij) N × N indicates the network structure for having N number of node, wherein cij
Value indicate from node i to the connection weight of node j；cij=0 indicates from node i to connectionless node j；ciiIndicate node i
Biasing.
The nervus opticus network model is encoded according to the structural information in step S204, to obtain a dye
Colour solid.The structure of neural network model needs to be expressed as a genetic algorithm individual chromosome coding, could enough genetic algorithms
To be calculated.In one embodiment, if nervus opticus network model has N number of neuron, serial number is the defeated of the arrangement from 1 to N
Enter layer (input layer), hidden layer (hidden layer, be also hidden layer), output layer (output layer) node, it can
Neural network structure is indicated with a N N matrix.Now with shown in Fig. 2 a with 7 nodes neural network structure as an example,
To illustrate coding method of the present embodiment to nervus opticus network model.Table 1 is the node connection relationship of the neural network structure,
In table 1, (i, j) corresponding element representation is from i-th of node to the connection relationship of j-th of node in matrix.Due to the present invention
Embodiment will not relate to the change to nervus opticus network model connection weight when compressing to nervus opticus network model, because
The connection relationship of node is expressed as 0,1, -1 form by this present embodiment, wherein " 0 " indicates not connect；" 1 " indicates connection
Weight is 1, has excitation (excitory) effect, is indicated in Fig. 2 a with solid line；" -1 " indicates that connection weight is -1, has and inhibits
(inhibitory) it acts on, is represented by dotted lines in Fig. 2 a.It can be seen that structural equivalence shown in table 1 and Fig. 2 a.
Table 1, the present embodiment exemplary neural network structure connection relationship
It can be the number of 0,1, -1 composition by the coded representation of the neural network according to node connection relationship shown in table 1
Element (3,1) to element (7,6) are linked in sequence, form following chromosome by word string form from left to right, from top to bottom
Coding:
It carries out Population Initialization according to the chromosome in step S206 and generates initial population.When it is implemented, can be right
The chromosome that neural network model to be compressed encodes executes duplication operation, generates the chromosome of predetermined quantity at random,
Using the set of these chromosomes as initial population.The size of initial population determines by population size M, population size M
It can be such as, but not limited to 10~100.After executing duplication operation to this chromosome, a plurality of identical chromosome is obtained, every
Chromosome completes the initialization of group using the set of this multiple chromosome as initial population for a chromosome.
In step S208, the fitness value of chromosome in group is calculated.In the present embodiment, fitness function can be adopted
It is calculated with following formula:
Wherein, f (i, t) indicates the fitness of i-th of body in t generation；E (i, t) indicates that i-th of body in t generation is corresponding
The network error of neural network model；H (i, t) indicates the network reduction degree of i-th of body in t generation.
The calculating of E (i, t) and H (i, t) can refer to the calculation formula that embodiment illustrated in fig. 1 provides.
In the present embodiment, fitness function include formula 1. with formula 2..Wherein, formula is 1. based on the adaptation of network error
Function is spent, what is reflected is the precision of neural network model；2. formula is the fitness function based on network reduction degree, reflection
Be neural network model compression.Thus fitness value of the chromosome based on precision can be calculated separately and based on compression
Fitness value.
In step S210, judge whether to reach termination condition.Wherein, termination condition may include preset in advance changes
For frequency threshold value or the condition of convergence of setting.The number of iterations such as, but not limited to can be set to 500 times, but the number of iterations reaches
It is judged as at 500 times and reaches termination condition.The condition of convergence such as, but not limited to can be set to meet centainly when fitness value
Condition when, be judged as and reach termination condition.In one embodiment, the condition of convergence may be configured as fitness value and meanwhile meet with
Lower condition:
Wherein, E0To preset network error value, H0To preset network reduction angle value.
In step S212, if step S210 judging result is to be not up to the termination condition, using fitness value as standard,
The chromosome that selected section fitness value is met the requirements executes the genetic manipulations such as duplication, intersection or variation, to generate new
Generation group, then return step S208.The selection criteria of the present embodiment can use following steps: (1) 1. being calculated with formula
Then fitness value of each chromosome based on precision in group calculates the selected first choice probability of individual, according to
The first choice probability selection goes out the first chromosome individual；(2) each chromosome in group is 2. calculated with formula to be based on
Then the fitness value of compression calculates the second selected select probability of individual, according to second select probability from step (1)
The second chromosome is selected in the first chromosome individual selected.Optionally, according to select probability selective staining body
Before body, the highest and lowest chromosome of fitness value in current group can be first found out, optimum dyeing body individual is retained straight
It taps into the next generation, eliminates worst chromosome, can guarantee in this way by excellent gene genetic to the next generation.It changes for the first time
Dai Shi, since the chromosome of initial population is all the same, optimized individual that when first time iteration is retained and eliminate
Worst individual is identical.Roulette method can be used to the selection of chromosome, the select probability of chromosome can use
Following formula calculates:
Wherein, p (i, t) is the select probability of t i-th of body of generation, and f (i, t) is the fitness of t i-th of body of generation, f
(sum, t) is t for the total fitness of group.The precision and pressure of neural network model can be taken into account using the selection criteria of the present embodiment
Contracting.
Duplication, intersection or mutation operation are executed to the chromosome being selected, such as contaminated using above-mentioned selection criteria
The selection of colour solid individual is then that duplication, intersection or mutation operation are executed to the second chromosome selected.Wherein, it replicates
Operation refers to the parent chromosome individual that will be selected under conditions of without any variation from when former generation is copied directly to new one
In generation individual.Crossover operation, which refers to, randomly chooses two parent chromosome individuals by above-mentioned selection method from group, by two
The constituent part of a parent chromosome individual is substituted for each other, and forms new child chromosome individual.Mutation operation refers to from group
In by above-mentioned selection method randomly choose a parent chromosome individual, then at random selected one in the expression formula of the individual
A node forms new child chromosome by the way that the value of the variation point gene is become another virtual value as change point
Body.
Whether crossover operation occurs can be according to crossover probability PcIt determines, method is to be randomly generated between one 0~1
Random number P, as P≤Pc, crossover operation occurs, as P > Pc, intersect and do not occur.Equally, whether mutation operation occur can also root
According to mutation probability PmIt determines, due to omitting descriptions thereof herein for the prior art.
In the present embodiment, when executing crossover operation, it can be selected at random in each parent chromosome individual according to certain probability
A crosspoint is selected, is partially known as transposition section below crosspoint.After first parent chromosome individual deletes its transposition section,
The transposition section of two parent chromosome individuals is inserted into his intersection, thus generates first child chromosome
Body.Equally, after second parent chromosome individual deletes its transposition section, the transposition section of first parent chromosome individual is inserted into
Second filial generation chromosome is formed after to his intersection.In this case, if two parents dyeing of selection
Body individual is identical, but due to its crosspoint difference, generated child chromosome individual is not also identical, and it is numerous to effectively prevent close relative
It grows, improves ability of searching optimum.
In the present embodiment, when executing mutation operation, it can be at random using one of following operation: (a) deleting neural network
At least one node and its corresponding connection in model hidden layer；(b) at least one in neural network model hidden layer is deleted
A connection；(c) deleted node or connection are repaired at random with certain probability；(d) increase hidden layer node, at random
Generate corresponding connection weight.Wherein, deletion of node is always prior to increasing node, and increased number of adding some points should not exceed deletion
Number of nodes, meanwhile, only when deletion of node cannot generate a good filial generation, just increase node, such mutation operation energy
Ensuring method is carried out toward the direction of compression neural network model always.
Optimal chromosome is exported if step S210 judging result is to reach the termination condition in step S214
To obtain the first nerves network model after compression processing.In the present embodiment, optimal chromosome may be configured as meeting following
Condition:
f0=max [H (i, t)]
Alternatively, optimal chromosome may be configured as meeting the following conditions simultaneously:
Wherein, f0For the fitness of optimal chromosome, E0To preset network error value, H (i, t) is the i-th of t generation
The network reduction degree of individual.
Fig. 3 shows the schematic diagram of the surface defects of products detection device of one embodiment according to the invention.Shown in Fig. 3
Device 300 it is corresponding with the said goods detection method of surface flaw, since the embodiment of device 300 is substantially similar to method
Embodiment, so describing fairly simple, the relevent part can refer to the partial explaination of embodiments of method.Device 300 can benefit
It is realized with the mode of software, hardware or software and hardware combining, may be mounted at computer or other suitably has computing capability
Electronic equipment in.
As shown in figure 3, device 300 may include obtaining module 302 and detection module 304.Module 302 is obtained for obtaining
Image to be detected comprising product surface.Detection module 304 is used for using first nerves network model to described image to be detected
It is detected, the surface for being included with the described image to be detected of determination is with the presence or absence of defect；Wherein, the first nerves network mould
Type carries out compression processing acquisition, the nervus opticus net to trained nervus opticus network model using genetic algorithm
Network model is obtained using preset training sample training, the essence of the first nerves network model based on the training sample
Degree is not less than default precision.
In an embodiment of device 300, device 300 further includes genetic manipulation module, for suitable based on compression
Answering angle value is standard, executes genetic manipulation to chromosome corresponding to the nervus opticus network model, is adapted to generating
The optimal chromosome of angle value；Training module, for the chromosome optimal to the fitness value using the training sample
Neural network model corresponding to individual is trained, to obtain the first nerves network model.
In another embodiment of device 300, the genetic manipulation module includes: acquiring unit, described for obtaining
The structural information of nervus opticus network model；Coding unit is used for according to the structural information, to the nervus opticus network mould
Type is encoded, to obtain a chromosome；Initialization unit, for carrying out Population Initialization and generating just according to the chromosome
Beginning group；Computing unit, for calculating the fitness value of chromosome in group；Judging unit reaches for judging whether
Termination condition；Genetic manipulation unit, if being used for the not up to described termination condition, using fitness value as standard, selected section is suitable
The chromosome that angle value is met the requirements is answered, duplication, intersection or mutation operation are executed, to generate group of new generation；Output is single
Member, if exporting the optimal chromosome of the fitness value for reaching the termination condition.
In the another embodiment of device 300, the computing unit is further used for: calculating separately the chromosome
Body is based on precision and based on the fitness value of compression；Correspondingly, the genetic manipulation unit is further used for: being based on according to described
The fitness value of precision obtains the first choice probability of chromosome in the group, is selected according to the first choice probability
The first chromosome individual is selected, and, according to the fitness value based on compression, obtain the of chromosome in the group
Two select probabilities select the second chromosome from the first chromosome individual according to second select probability；To institute
It states the second chromosome and executes duplication, intersection or mutation operation, to generate group of new generation.
Fig. 4 shows the schematic diagram of the computer equipment of one embodiment according to the invention.As shown in figure 4, computer
Equipment 400 may include processor 402 and memory 404, wherein be stored with executable instruction on memory 402, wherein institute
State executable instruction makes processor 402 execute method 100 or method shown in Fig. 2 200 shown in FIG. 1 upon being performed.
Fig. 5 shows the block diagram for being suitable for the exemplary computer device for being used to realize embodiment of the present invention.It is shown in fig. 5
Computer equipment 500 is only an example, should not function to the embodiment of the present invention and use scope bring any restrictions.
As shown in figure 5, computer equipment 500 is realized in the form of universal computing device.The component of computer equipment 500 can
To include but is not limited to: processor 502, system storage 504, connecting different system components, (including processor 502 and system are deposited
Reservoir 504) bus 506.
Bus 506 indicates one of a few class bus structures or a variety of, including memory bus or Memory Controller,
Peripheral bus, graphics acceleration port, processor or the local bus using any bus structures in a variety of bus structures.It lifts
For example, these architectures include but is not limited to industry standard architecture (ISA) bus, microchannel architecture (MAC)
Bus, enhanced isa bus, Video Electronics Standards Association (VESA) local bus and peripheral component interconnection (PCI) bus.
Computer equipment 500 typically comprises a variety of computer system readable media.These media can be it is any can
The usable medium accessed by computer equipment 500, including volatile and non-volatile media, moveable and immovable Jie
Matter.
System storage 504 may include the computer system readable media of form of volatile memory, such as deposit at random
Access to memory (RAM) 508 and and/or cache memory 510.Computer equipment 500 may further include other removable
Dynamic/immovable, volatile/non-volatile computer system storage medium.Only as an example, storage system 512 can be used
In reading and writing immovable, non-volatile magnetic media (Fig. 5 do not show, commonly referred to as " hard disk drive ").Although not showing in Fig. 5
Out, the disc driver for reading and writing to removable non-volatile magnetic disk (such as " floppy disk ") can be provided, and to removable
The CD drive of anonvolatile optical disk (such as CD-ROM, DVD-ROM or other optical mediums) read-write.In these cases,
Each driver can be connected by one or more data media interfaces with bus 506.System storage 504 may include
At least one program product, the program product have one group of (for example, at least one) program module, these program modules are configured
To execute the function of the above-mentioned Fig. 1 or Fig. 2 embodiment of the present invention.
Program/utility 514 with one group of (at least one) program module 516, can store and deposit in such as system
In reservoir 504, such program module 516 includes but is not limited to operating system, one or more application program, other programs
It may include the realization of network environment in module and program data, each of these examples or certain combination.Program mould
Block 516 usually executes the function and/or method in above-mentioned Fig. 1 or Fig. 2 embodiment described in the invention.
Computer equipment 500 can also be with one or more external equipments 600 (such as keyboard, sensing equipment, display
700 etc.) it communicates, the equipment interacted with the computer equipment 500 communication can be also enabled a user to one or more, and/or
(such as network interface card is adjusted with any equipment for enabling the computer equipment 500 to be communicated with one or more of the other calculating equipment
Modulator-demodulator etc.) communication.This communication can be carried out by input/output (I/0) interface 518.Also, computer equipment
500 can also by network adapter 520 and one or more network (such as local area network (LAN), wide area network (WAN) and/or
Public network, such as internet) communication.As shown, network adapter 520 passes through its of bus 506 and computer equipment 500
The communication of its module.It should be understood that although not shown in the drawings, other hardware and/or software can be used in conjunction with computer equipment 500
Module, including but not limited to: microcode, device driver, redundant processor, external disk drive array, RAID system, tape
Driver and data backup storage system etc..
Processor 502 by the program that is stored in system storage 504 of operation, thereby executing various function application and
Data processing, such as realize method for detecting surface defects of products shown in above-described embodiment.
The embodiment of the present invention also provides a kind of computer-readable medium, is stored thereon with executable instruction, wherein described
Executable instruction makes computer execute method 100 or method shown in Fig. 2 200 shown in FIG. 1 upon being performed.
The computer-readable medium of the present embodiment may include in the system storage 504 in above-mentioned embodiment illustrated in fig. 5
RAM508, and/or cache memory 510, and/or storage system 512.
With the development of science and technology, the route of transmission of computer program is no longer limited by tangible medium, it can also be directly from net
Network downloading, or obtained using other modes.Therefore, the computer-readable medium in the present embodiment not only may include tangible
Medium can also include invisible medium.
The computer-readable medium of the present embodiment can be using any combination of one or more computer-readable media.
Computer-readable medium can be computer-readable signal media or computer readable storage medium.Computer-readable storage medium
Matter can for example be but not limited to system, device or the device of electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor, or arbitrarily with
On combination.The more specific example (non exhaustive list) of computer readable storage medium includes: to lead with one or more
The electrical connection of line, portable computer diskette, hard disk, random access memory (RAM), read-only memory (ROM), erasable type can
Program read-only memory (EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light storage device,
Magnetic memory device or above-mentioned any appropriate combination.In this document, computer readable storage medium can be any packet
Contain or store the tangible medium of program, which can be commanded execution system, device or device use or in connection
It uses.
Computer-readable signal media may include in a base band or as carrier wave a part propagate data-signal,
Wherein carry computer-readable program code.The data-signal of this propagation can take various forms, including but unlimited
In electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media can also be that computer can
Any computer-readable medium other than storage medium is read, which can send, propagates or transmit and be used for
By the use of instruction execution system, device or device or program in connection.
The program code for including on computer-readable medium can transmit with any suitable medium, including but not limited to without
Line, electric wire, optical cable, RF etc. or above-mentioned any appropriate combination.
The computer for executing operation of the present invention can be write with one or more programming languages or combinations thereof
Program code, described program design language include object oriented program language, such as Java, Smalltalk, C++, also
Including conventional procedural programming language, such as " C " language or similar programming language.Program code can be complete
Ground executes on the user computer, partly executes on the user computer, executing as an independent software package, partially existing
Part executes on the remote computer or executes on a remote computer or server completely on subscriber computer.It is being related to
In the situation of remote computer, remote computer can pass through the network of any kind, including local area network (LAN) or wide area network
(WAN), it is connected to subscriber computer, or, it may be connected to outer computer (such as led to using ISP
Cross internet connection).
It will be understood by those skilled in the art that the embodiment of the present invention can provide as method, apparatus or computer program production
Product.Therefore, in terms of the embodiment of the present invention can be used complete hardware embodiment, complete software embodiment or combine software and hardware
Embodiment form.Moreover, it wherein includes computer available programs generation that the embodiment of the present invention, which can be used in one or more,
The meter implemented in the computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) of code
The form of calculation machine program product.
The embodiment of the present invention be referring to according to the method for the embodiment of the present invention, the process of device and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing terminal devices
To generate a machine, so that being produced by the instruction that computer or the processor of other programmable data processing terminal devices execute
Life is for realizing the function of specifying in one or more flows of the flowchart and/or one or more blocks of the block diagram
Device.
The specific embodiment illustrated above in conjunction with attached drawing describes exemplary embodiment, it is not intended that may be implemented
Or fall into all embodiments of the protection scope of claims." exemplary " meaning of the term used in entire this specification
Taste " be used as example, example or illustration ", be not meant to than other embodiments " preferably " or " there is advantage ".For offer pair
The purpose of the understanding of described technology, specific embodiment include detail.However, it is possible in these no details
In the case of implement these technologies.In some instances, public in order to avoid the concept to described embodiment causes indigestion
The construction and device known is shown in block diagram form.
The foregoing description of present disclosure is provided so that any those of ordinary skill in this field can be realized or make
Use present disclosure.To those skilled in the art, the various modifications carried out to present disclosure are apparent
, also, can also answer generic principles defined herein in the case where not departing from the protection scope of present disclosure
For other modifications.Therefore, present disclosure is not limited to examples described herein and design, but disclosed herein with meeting
Principle and novel features widest scope it is consistent.
Claims (10)
1. method for detecting surface defects of products, comprising:
Obtain image to be detected comprising product surface；
Described image to be detected is detected using first nerves network model, is included with the described image to be detected of determination
Product surface whether there is defect；
Wherein, the first nerves network model is to be pressed using genetic algorithm trained nervus opticus network model
Contracting processing obtains, and the nervus opticus network model is obtained using preset training sample training, the first nerves
Network model is not less than default precision based on the precision of the training sample.
2. according to the method described in claim 1, wherein, the utilization genetic algorithm is to trained nervus opticus network mould
Type carries out compression processing, comprising:
Using the fitness value based on compression as standard, something lost is executed to chromosome corresponding to the nervus opticus network model
Operation is passed, to generate the optimal chromosome of fitness value；
It is instructed using neural network model corresponding to the training sample chromosome optimal to the fitness value
Practice, to obtain the first nerves network model.
3. described to chromosome corresponding to the nervus opticus network model according to the method described in claim 2, wherein
Body executes genetic manipulation, to generate the optimal chromosome of fitness value, comprising:
Obtain the structural information of the nervus opticus network model；
According to the structural information, the nervus opticus network model is encoded, to obtain a chromosome；
According to the chromosome, carries out Population Initialization and generate initial population；
Calculate the fitness value of chromosome in group；
Judge whether to reach termination condition；
If the not up to described termination condition, using fitness value as standard, chromosome that selected section fitness value is met the requirements
Individual executes duplication, intersection or mutation operation to generate group of new generation and then returns to chromosome in the calculating group
The fitness value step of individual；
If reaching the termination condition, the optimal chromosome of the fitness value is exported.
4. according to the method described in claim 3, wherein, the fitness value for calculating chromosome in group, comprising:
The chromosome is calculated separately based on precision and based on the fitness value of compression；
Correspondingly, described using fitness value as standard, the chromosome that selected section fitness value is met the requirements, executes multiple
System, intersection or mutation operation, to generate group of new generation, comprising:
According to the fitness value based on precision, the first choice probability of chromosome in the group is obtained, according to institute
First choice probability selection the first chromosome individual is stated, and, according to the fitness value based on compression, obtain the group
Second select probability of middle chromosome selects second from the first chromosome individual according to second select probability
Chromosome；Duplication, intersection or mutation operation are executed to second chromosome, to generate group of new generation.
5. surface defects of products detection device, comprising:
Module is obtained, for obtaining image to be detected comprising product surface；
Detection module, it is described to be checked with determination for being detected using first nerves network model to described image to be detected
The product surface that altimetric image is included whether there is defect；
Wherein, the first nerves network model is to be pressed using genetic algorithm trained nervus opticus network model
Contracting processing obtains, and the nervus opticus network model is obtained using preset training sample training, the first nerves
Network model is not less than default precision based on the precision of the training sample.
6. device according to claim 5, wherein described device further include:
Genetic manipulation module is right to the neural network model institute to be compressed for using the fitness value based on compression as standard
The chromosome answered executes genetic manipulation, to generate the optimal chromosome of fitness value；
Training module, for utilizing nerve net corresponding to the training sample chromosome optimal to the fitness value
Network model is trained, to obtain the first nerves network model.
7. device according to claim 6, wherein the genetic manipulation module includes:
Acquiring unit, for obtaining the structural information of the nervus opticus network model；
Coding unit, for being encoded to the nervus opticus network model, to obtain a dyeing according to the structural information
Body；
Initialization unit, for carrying out Population Initialization and generating initial population according to the chromosome；
Computing unit, for calculating the fitness value of chromosome in group；
Judging unit reaches termination condition for judging whether；
Genetic manipulation unit, if being used for the not up to described termination condition, using fitness value as standard, selected section fitness value
The chromosome met the requirements executes duplication, intersection or mutation operation, to generate group of new generation；
Output unit, if exporting the optimal chromosome of the fitness value for reaching the termination condition.
8. device according to claim 7, wherein the computing unit is further used for:
The chromosome is calculated separately based on precision and based on the fitness value of compression；
Correspondingly, the genetic manipulation unit is further used for:
According to the fitness value based on precision, the first choice probability of chromosome in the group is obtained, according to institute
First choice probability selection the first chromosome individual is stated, and, according to the fitness value based on compression, obtain the group
Second select probability of middle chromosome selects second from the first chromosome individual according to second select probability
Chromosome；Duplication, intersection or mutation operation are executed to second chromosome, to generate group of new generation.
9. a kind of computer equipment, comprising:
Processor；And
Memory is stored thereon with executable instruction, wherein the executable instruction holds the processor
The described in any item methods of row claim 1-4.
10. a kind of computer-readable medium, is stored thereon with executable instruction, wherein the executable instruction is upon being performed
So that computer perform claim requires the described in any item methods of 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811037591.4A CN109272497A (en) | 2018-09-05 | 2018-09-05 | Method for detecting surface defects of products, device and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811037591.4A CN109272497A (en) | 2018-09-05 | 2018-09-05 | Method for detecting surface defects of products, device and computer equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109272497A true CN109272497A (en) | 2019-01-25 |
Family
ID=65188046
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811037591.4A Pending CN109272497A (en) | 2018-09-05 | 2018-09-05 | Method for detecting surface defects of products, device and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109272497A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109919939A (en) * | 2019-03-27 | 2019-06-21 | 王合山 | A kind of yarn faults detection method and device based on genetic algorithm |
| CN110211121A (en) * | 2019-06-10 | 2019-09-06 | 北京百度网讯科技有限公司 | Method and apparatus for pushing model |
| CN110297041A (en) * | 2019-07-01 | 2019-10-01 | 天津工业大学 | A kind of 3D woven composite defect inspection method based on FCN and GRU |
| CN110910360A (en) * | 2019-11-14 | 2020-03-24 | 腾讯云计算（北京）有限责任公司 | Power grid image positioning method and image positioning model training method |
| CN112580639A (en) * | 2021-03-01 | 2021-03-30 | 四川大学 | Early gastric cancer image identification method based on evolutionary neural network model compression |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101246200A (en) * | 2008-03-10 | 2008-08-20 | 湖南大学 | Analog PCB intelligent test system based on neural network |
| CN101315557A (en) * | 2008-06-25 | 2008-12-03 | 浙江大学 | Propylene polymerization production process optimal soft survey instrument and method based on genetic algorithm optimization BP neural network |
| CN101599138A (en) * | 2009-07-07 | 2009-12-09 | 武汉大学 | Land evaluation method based on artificial neural network |
| CN106295795A (en) * | 2016-08-09 | 2017-01-04 | 衢州学院 | A kind of copper pipe interior whorl forming qualitative forecasting method based on BP neural network algorithm |
-
2018
- 2018-09-05 CN CN201811037591.4A patent/CN109272497A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101246200A (en) * | 2008-03-10 | 2008-08-20 | 湖南大学 | Analog PCB intelligent test system based on neural network |
| CN101315557A (en) * | 2008-06-25 | 2008-12-03 | 浙江大学 | Propylene polymerization production process optimal soft survey instrument and method based on genetic algorithm optimization BP neural network |
| CN101599138A (en) * | 2009-07-07 | 2009-12-09 | 武汉大学 | Land evaluation method based on artificial neural network |
| CN106295795A (en) * | 2016-08-09 | 2017-01-04 | 衢州学院 | A kind of copper pipe interior whorl forming qualitative forecasting method based on BP neural network algorithm |
Non-Patent Citations (2)
| Title |
|---|
| 席军强: "《车辆信息技术》", 31 December 2013 * |
| 罗超 等: "基于机器视觉的缺陷在线检测系统的研究", 《数字技术与应用》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109919939A (en) * | 2019-03-27 | 2019-06-21 | 王合山 | A kind of yarn faults detection method and device based on genetic algorithm |
| CN109919939B (en) * | 2019-03-27 | 2023-09-29 | 王合山 | Yarn defect detection method and device based on genetic algorithm |
| CN110211121A (en) * | 2019-06-10 | 2019-09-06 | 北京百度网讯科技有限公司 | Method and apparatus for pushing model |
| CN110297041A (en) * | 2019-07-01 | 2019-10-01 | 天津工业大学 | A kind of 3D woven composite defect inspection method based on FCN and GRU |
| CN110910360A (en) * | 2019-11-14 | 2020-03-24 | 腾讯云计算（北京）有限责任公司 | Power grid image positioning method and image positioning model training method |
| CN110910360B (en) * | 2019-11-14 | 2023-06-13 | 腾讯云计算（北京）有限责任公司 | Positioning method of power grid image and training method of image positioning model |
| CN112580639A (en) * | 2021-03-01 | 2021-03-30 | 四川大学 | Early gastric cancer image identification method based on evolutionary neural network model compression |
| CN112580639B (en) * | 2021-03-01 | 2021-08-13 | 四川大学 | Early gastric cancer image identification method based on evolutionary neural network model compression |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109165720A (en) | Neural network model compression method, device and computer equipment | |
| CN109272497A (en) | Method for detecting surface defects of products, device and computer equipment | |
| RU2666631C2 (en) | Training of dnn-student by means of output distribution | |
| CN111723937A (en) | Method, device, equipment and medium for generating description information of multimedia data | |
| CN113538506A (en) | Pedestrian trajectory prediction method based on global dynamic scene information depth modeling | |
| WO2019205319A1 (en) | Commodity information format processing method and apparatus, and computer device and storage medium | |
| CN112418292B (en) | Image quality evaluation method, device, computer equipment and storage medium | |
| CN111429885A (en) | Method for mapping audio clip to human face-mouth type key point | |
| WO2023109714A1 (en) | Multi-mode information fusion method and system for protein representative learning, and terminal and storage medium | |
| CN111352965A (en) | Training method of sequence mining model, and processing method and equipment of sequence data | |
| KR20190125029A (en) | Methods and apparatuses for generating text to video based on time series adversarial neural network | |
| CN109299780A (en) | Neural network model compression method, device and computer equipment | |
| CN113128432B (en) | Machine vision multitask neural network architecture searching method based on evolution calculation | |
| CN108959388A (en) | information generating method and device | |
| CN115131849A (en) | Image generation method and related device | |
| CN115937369A (en) | Expression animation generation method and system, electronic equipment and storage medium | |
| CN109166118A (en) | Fabric surface attribute detection method, device and computer equipment | |
| CN114821248B (en) | Point cloud understanding-oriented data active screening and labeling method and device | |
| CN116361643A (en) | Model training method for realizing object recommendation, object recommendation method and related device | |
| WO2022127603A1 (en) | Model processing method and related device | |
| CN115905518A (en) | Emotion classification method, device and equipment based on knowledge graph and storage medium | |
| CN113822790B (en) | Image processing method, device, equipment and computer readable storage medium | |
| CN108596334B (en) | Data corresponding relation judging and generating method and system based on bidirectional deep learning | |
| JP7230324B2 (en) | Neural network learning method, computer program and computer device | |
| WO2020054402A1 (en) | Neural network processing device, computer program, neural network manufacturing method, neural network data manufacturing method, neural network use device, and neural network downscaling method |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190125 |