CN108230318A - Ladle defects detection sorting technique, device, equipment and computer-readable medium - Google Patents
Ladle defects detection sorting technique, device, equipment and computer-readable medium Download PDFInfo
- Publication number
- CN108230318A CN108230318A CN201810020267.5A CN201810020267A CN108230318A CN 108230318 A CN108230318 A CN 108230318A CN 201810020267 A CN201810020267 A CN 201810020267A CN 108230318 A CN108230318 A CN 108230318A
- Authority
- CN
- China
- Prior art keywords
- ladle
- detection
- defect
- detection model
- image data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- 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
- G01N2021/8854—Grading and classifying of flaws
-
- 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
- G01N2021/8883—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 involving the calculation of gauges, generating models
-
- 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
- G06T2207/30136—Metal
Landscapes
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The present invention proposes a kind of ladle defects detection sorting technique, includes the following steps：Ladle image data to be detected is received, and generates the detection request of ladle defect；According to load balancing and scheduling strategy, the ladle image data and detection request are sent on the best server for carrying detection model；The prediction result for ladle image data export after prediction calculating by detection model is received, the prediction result includes the classification of ladle defect；According to the prediction result that detection model exports, corresponding response action is performed.The embodiment of the present invention to the ladle image acquired in real time by being detected judgement, position and its affiliated classification where final acquisition ladle defect.Further, the embodiment of the present invention is detected the iteration update of model, and detection model is made to can adapt to the newest demand of production environment, brings and is obviously improved for industrial production line in nicety of grading, scalability, standardization etc..
Description
Technical field
The present invention relates to field of artificial intelligence more particularly to a kind of ladle defects detection sorting technique and device, set
Standby and computer-readable medium.
Background technology
In steel and iron manufacturing industry, ladle is the important equipment of steel-making, and state is not only related to the quality of steel production, and
And it is related to the safety of production environment.Therefore it is the key link in Steel Production Flow Chart for the quality inspection of ladle state.Tradition
It is the state progress to ladle inner wall surface to a kind of important means that ladle state is monitored in iron and steel enterprise's production environment
Detection, with judge ladle whether defect, and corresponding processing is done to ladle according to testing result.In on steel sheet production line, this
Quality inspection of the kind based on ladle inner wall surface state is mostly manual inspection or semi-automatic optical instrument auxiliary quality inspection, and not only efficiency is low
Under, and erroneous judgement is susceptible to, in addition, the industrial data that this mode generates is not easy to store, manage and mining again recycles.
And existing quality inspection system in defects detection and positioning application there are mainly two types of mode.First way is pure people
Working medium procuratorial organ formula, i.e., the photo visually observed dependent on industry specialists in production environment provide judgement.The second way is machine
The artificial quality inspection mode of auxiliary, mainly by having the filtering of the quality inspection system of certain judgement not have defective photo, by industry
Expert is detected judgement to the photo of doubtful existing defects.Wherein, the second way is mostly expert system and Feature Engineering system
System develops, and experience is solidificated in quality inspection system by expert, has certain automatic capability.
However, for the first quality inspection mode, due in existing industrial production line quality inspection system, the detection of defect picture and
Positioning relies primarily on the detecting system of pure hand inspection or feature based engineering.In the case of artificial quality inspection, need business special
Family carries out walkaround inspection in production scene, and manual record gets off to do subsequent processing again after finding defect.This method is not only imitated
Rate is low, erroneous judgement of easily failing to judge, and data are difficult to carry out secondary use excavation, and industrial production environment is often relatively more severe, to people
The health and safety of member can adversely affect.
For second of quality inspection mode, in the quality inspection system based on traditional expert system or Feature Engineering, feature and sentence
Set pattern is then all based on experience and is cured in machine, it is difficult to the development iteration of business, lead to the development with production technology,
The accuracy of detection of system is lower and lower or even is reduced to complete unusable state.In addition, the feature of traditional quality inspection system all by
Within hardware, when upgrading, not only needs to carry out key technological transformation to production line, but also expensive third-party vendor's curing in advance.
All there is apparent deficiencies in safety, standardization, scalability etc. for traditional quality inspection system, are unfavorable for traditional industry production
The optimization and upgrading of line.
Invention content
The embodiment of the present invention provides a kind of ladle defects detection sorting technique, device, equipment and computer-readable medium, with
It solves or alleviates Yi Shang technical problem of the prior art.
In a first aspect, an embodiment of the present invention provides a kind of ladle defects detection sorting technique, include the following steps：
Ladle image data to be detected is received, and generates the detection request of ladle defect；
According to load balancing and scheduling strategy, the ladle image data and detection request are sent to carrying detection model
Best server on；
Receive the prediction result for ladle image data export after prediction calculating by detection model, the prediction result
Include the classification of ladle defect；
According to the prediction result of detection model, corresponding response action is performed.
With reference to first aspect, the present invention is described to be exported according to detection model in the first realization method of first aspect
Prediction result, after the step of performing corresponding response action, further include：More new command is sent to best server, by most
Good server is trained update according to the ladle image data and prediction result of reception to the detection model.
Second aspect, an embodiment of the present invention provides a kind of ladle defects detection sorting techniques, include the following steps：
Receive the detection request of ladle image data and ladle defect；
The prediction for carrying out ladle defect to the image data of ladle by detection model calculates and exports prediction result, described
Prediction result includes the classification of ladle defect.
With reference to second aspect, in the first realization method of second aspect, the detection model includes the present invention：Depth
Convolutional neural networks and defect location sorter network；
The depth convolutional neural networks are used to extract the feature of ladle picture, and the feature is input to the defect
It positions in sorter network；
The defect location sorter network be used to judging feature that depth convolutional neural networks are extracted whether existing defects,
And obtain the classification of the defect.
With reference to the first realization method of second aspect, the present invention is described in second of realization method of second aspect
Depth convolutional neural networks include：
Convolutional layer for being scanned convolution to ladle image using the different convolution kernel of weights, therefrom extracts various meanings
The feature of justice, and export into characteristic pattern；
Pond layer, for carrying out dimensionality reduction operation to characteristic pattern；
Full articulamentum, for by the Feature Mapping extracted to defect location sorter network.
With reference to second aspect, the present invention further includes step in the third realization method of second aspect：According to reception
Ladle image data and prediction result are trained update to the detection model.
With reference to the third realization method of second aspect, the present invention is described in the 4th kind of realization method of second aspect
The step of being trained update to the detection model according to the ladle image data of reception and prediction result includes：
Ladle image data and prediction result are stored, and the training data of detection model is updated；
Detection model is trained according to updated training data, updates detection model.
The third aspect, an embodiment of the present invention provides a kind of ladle defects detection sorter, including：
Receiving module for receiving ladle image data to be detected, and generates the detection request of ladle defect；
Scheduler module, for according to load balancing and scheduling strategy, the ladle image data and detection request to be sent
To the best server for carrying detection model；
Result of calculation receiving module by detection model carries out ladle image data what is exported after prediction calculating for receiving
Prediction result, the prediction result include the classification of ladle defect；
Respond module for the prediction result exported according to detection model, performs corresponding response action.
With reference to the third aspect, the present invention further includes more new command and sends mould in the first realization method of the third aspect
Block, for sending more new command to best server, by best server according to the ladle image data and prediction result of reception
Update is trained to the detection model.
Fourth aspect, an embodiment of the present invention provides a kind of ladle defects detection sorter, including：
Request receiving module is detected, for receiving the detection of ladle image data and ladle defect request；
Computing module, the prediction that ladle defect is carried out for passing through detection model to the image data of ladle are calculated and are exported
Prediction result, the prediction result include the classification of ladle defect.
With reference to fourth aspect, in the first realization method of fourth aspect, the detection model includes the present invention：Depth
Convolutional neural networks and defect location sorter network；
The depth convolutional neural networks are used to extract the feature of ladle picture, and the feature is input to the defect
It positions in sorter network；
The defect location sorter network be used to judging feature that depth convolutional neural networks are extracted whether existing defects,
And obtain the classification of the defect.
With reference to the first realization method of fourth aspect, the present invention is described in second of realization method of fourth aspect
Depth convolutional neural networks include：
Convolutional layer for being scanned convolution to ladle image using the different convolution kernel of weights, therefrom extracts various meanings
The feature of justice, and export into characteristic pattern；
Pond layer, for carrying out dimensionality reduction operation to characteristic pattern；
Full articulamentum, for by the Feature Mapping in characteristic pattern to defect location sorter network.
With reference to fourth aspect, the present invention further includes update module, for root in the third realization method of second aspect
Update is trained to the detection model according to the ladle image data and prediction result of reception.
With reference to the third realization method of fourth aspect, the present invention is described in the 4th kind of realization method of fourth aspect
Update module includes：
Sub-module stored for storing ladle image data and prediction result, and carries out the training data of detection model
Update；
Training submodule for being trained according to updated training data to detection model, updates detection model.
The function of described device by hardware can also be performed corresponding software and be realized by hardware realization.It is described
Hardware or software include the one or more and corresponding module of above-mentioned function.
In a possible design, the structure of ladle defects detection sorter includes processor and memory, institute
State memory for store support ladle defects detection classification side in the above-mentioned first aspect of ladle defects detection sorter execution
The program of method, the processor are configurable for performing the program stored in the memory.The ladle defects detection point
Class device can also include communication interface, for ladle defects detection sorter and other equipment or communication.
5th aspect, an embodiment of the present invention provides a kind of computer-readable medium, for storing ladle defects detection point
Computer software instructions used in class device, including for performing the ladle defects detection of above-mentioned first aspect and second aspect
Program involved by sorting technique.
A technical solution in above-mentioned technical proposal has the following advantages that or advantageous effect：It is right that the embodiment of the present invention passes through
The ladle image acquired in real time is detected judgement, position and its affiliated classification where final acquisition ladle defect.Into one
Step, the embodiment of the present invention are detected the iteration update of model, detection model are made to can adapt to the newest demand of production environment,
Nicety of grading, scalability, standardization etc. are brought for industrial production line to be obviously improved.
Above-mentioned general introduction is merely to illustrate that the purpose of book, it is not intended to be limited in any way.Except foregoing description
Schematical aspect, except embodiment and feature, by reference to attached drawing and the following detailed description, the present invention is further
Aspect, embodiment and feature will be what is be readily apparent that.
Description of the drawings
In the accompanying drawings, unless specified otherwise herein, otherwise represent the same or similar through the identical reference numeral of multiple attached drawings
Component or element.What these attached drawings were not necessarily to scale.It should be understood that these attached drawings are depicted only according to the present invention
Some disclosed embodiments, and should not serve to limit the scope of the present invention.
Fig. 1 is the step flow chart of the ladle defects detection sorting technique of embodiment one；
Fig. 2 is the principle schematic of the detection model of embodiment one；
Fig. 3 is the principle schematic of the depth convolutional neural networks of embodiment one；
Fig. 4 is the step flow chart of the ladle defects detection sorting technique of embodiment two；
Fig. 5 is the step flow chart of the ladle defects detection sorting technique of embodiment three；
Fig. 6 is the specific steps flow chart of the step S330 of embodiment three；
Fig. 7 is the connection block diagram of the ladle defects detection sorter of example IV；
Fig. 8 is the connection block diagram of the ladle defects detection sorter of embodiment five；
Fig. 9 is the connection block diagram of the ladle defects detection sorter of embodiment six；
Figure 10 is the application schematic diagram of embodiment six.
Figure 11 is that the ladle defects detection sorting device of embodiment seven connects block diagram.
Specific embodiment
Hereinafter, certain exemplary embodiments are simply just described.As one skilled in the art will recognize that
Like that, without departing from the spirit or scope of the present invention, described embodiment can be changed by various different modes.
Therefore, attached drawing and description are considered essentially illustrative rather than restrictive.
The embodiment of the present invention aims to solve the problem that the prior art when carrying out quality inspection the technical issues of inefficiency.The present invention is implemented
Example is automatically detected the defects of ladle mainly by using detection model, and exports the position of corresponding ladle defect and divide
Class.The expansion for carrying out technical solution by following embodiment separately below describes.
Embodiment one
Referring to Fig. 1, the step flow chart of its ladle defects detection sorting technique for the embodiment of the present invention one.This implementation
Example one provides a kind of ladle defects detection sorting technique, includes the following steps：
S110：Ladle image data to be detected is received, and generates the detection request of ladle defect.
When the surface defect for ladle is needed to be detected, the acquisition of image can be first carried out to ladle surface, than
It can be such as acquired by video camera equipment.Then, after the image information for having acquired corresponding ladle, generation detection please
It asks, to be detected to ladle surface defect.
S120：According to load balancing and scheduling strategy, the ladle image data and detection request are sent to carrying inspection
It surveys on the best server of model.
Before it will detect request and send, first the deployment scenario of detection model is monitored, is disposed according to detection model
Situation, will detection request be sent on corresponding server.According to load balancing and scheduling strategy, by current request task
It is sent on current best server, to realize balance dispatching, improves the whole speed of service.Such as：According to different equal
Weigh scheduling strategy, can be sent to the smaller server of load, then load smaller server at this time as best server.Separately
Outside, the higher server of processing speed can also be sent to, then the higher server of processing speed is best server at this time.
S130：The prediction result for ladle image data export after prediction calculating by detection model is received, it is described pre-
Survey the classification that result includes ladle defect.
As shown in Fig. 2, the principle schematic of its detection model for the present embodiment.In the present embodiment, the detection mould
Type can include depth convolutional neural networks and defect location sorter network.
Wherein, the depth convolutional neural networks are used to extract the feature of ladle picture, and the feature is input to institute
It states in defect location sorter network.The depth convolutional neural networks can extract the characteristic information of ladle picture, such as may
The parts of images of existing defects extracts.
The defect location sorter network be used to judging feature that depth convolutional neural networks are extracted whether existing defects,
And obtain the classification of the defect.The defect location sorter network may be used the object detections such as RCNN, SSD, Mask RCNN and
Image Segmentation Model.Judge current feature whether existing defects, if in the presence of the relative position of current defect and affiliated is judged
Classification.If there are many places defects in current feature, the relative position coordinates and classification of each defect are provided.
As shown in figure 3, the principle schematic of its depth convolutional neural networks for the embodiment of the present invention one.It will be on production line
Input of the original image as model, and the output of model is then the label for representing defect classification, such as 0-n represents n+ respectively
1 class defect.Wherein, the structure of depth convolutional neural networks mainly by convolutional layer, pond layer, connect layer entirely and connect etc. and form.Wherein,
Convolution is scanned to original image or characteristic pattern (feature map) using weights different convolution kernel in convolutional layer, therefrom
The feature of various meanings is extracted, and is exported into characteristic pattern.Pond layer then carries out dimensionality reduction operation to characteristic pattern, in keeping characteristics figure
Main feature.It, can be to the change of photo on production line using this deep neural network model operated with convolution, pondization
Shape, fuzzy, illumination variation etc. have higher robustness, for classification task have it is higher can generalization.Full articulamentum is then
By the Feature Mapping of extraction to defect location sorter network.It the characteristics of for different production scenes and data, can be by setting
The deep neural network model that different depth, different number neuron, different convolution pond organizational forms are formed is counted, is then adopted
With the historical data marked, model is trained by the way of signal propagated forward, error back propagation.When model
Export label between error amount be less than it is preset meet business need threshold value when, deconditioning.
When being trained to convolutional neural networks, mainly by using a large amount of defect image data and non-defective image
Data are trained convolutional neural networks.Wherein, when using defect image data as input when, using probability value as " 1 " as
The benchmark of output.And when using non-defective image data as input when, using probability value " 0 " as export benchmark.By a large amount of
Data training after, the training of complete convolutional neural networks.
S140：According to the prediction result that detection model exports, corresponding response action is performed.
When the prediction result of detection model output is existing defects, then corresponding response action can be performed, such as：It can be with
Alarm operation is performed, for staff to be reminded to check etc..Furthermore it is also possible to real-time detection daily record is stored.
Embodiment two
With embodiment one difference lies in：The present embodiment two carries out on the basis of embodiment one also directed to detection model
Update, specific scheme are as follows：
Referring to Fig. 4, the step flow chart of its ladle defects detection sorting technique for the present embodiment two.The present embodiment two
A kind of ladle defects detection sorting technique is provided, is included the following steps：
S210：Ladle image data to be detected is received, and generates the detection request of ladle defect.
S220：According to load balancing and scheduling strategy, the ladle image data and detection request are sent to carrying inspection
It surveys on the best server of model.
S230：The prediction result for ladle image data export after prediction calculating by detection model is received, it is described pre-
Survey the classification that result includes ladle defect.
S240：According to the prediction result that detection model exports, corresponding response action is performed.
S250：More new command is sent to best server, by best server according to the ladle image data of reception and in advance
It surveys result and update is trained to the detection model.
In the present embodiment two, according to the image data and prediction result of newest reception, detection model is trained, from
And detection model is updated.
Step S210-S240 in the present embodiment two is identical with one principle of embodiment, and so it will not be repeated.
Embodiment three
Referring to Fig. 5, the step flow chart of its ladle defects detection sorting technique for the present embodiment three.The present embodiment three
A kind of ladle defects detection sorting technique is provided, is included the following steps：
S310：Receive the detection request of ladle image data and ladle defect.
S320：The prediction for carrying out ladle defect to the image data of ladle by detection model calculates and exports prediction knot
Fruit, the prediction result include the classification of ladle defect.
Wherein, the detection model includes：Depth convolutional neural networks and defect location sorter network.
The depth convolutional neural networks are used to extract the feature of ladle picture, and the feature is input to the defect
It positions in sorter network.
The defect location sorter network be used to judging feature that depth convolutional neural networks are extracted whether existing defects,
And obtain the classification of the defect.
The structure of depth convolutional neural networks mainly by convolutional layer, pond layer, connect layer entirely and connect etc. and form.Wherein, convolution
Convolution is scanned to original image or characteristic pattern (feature map) using weights different convolution kernel in layer, is therefrom extracted
The feature of various meanings, and export into characteristic pattern.Pond layer then carries out dimensionality reduction operation to characteristic pattern, the master in keeping characteristics figure
Want feature.Using this deep neural network model with convolution, pondization operation, can to the deformation of photo on production line,
Fuzzy, illumination variation etc. has higher robustness, for classification task have it is higher can generalization.Full articulamentum will then carry
The Feature Mapping taken is to defect location sorter network.It the characteristics of for different production scenes and data, can be by designing not
The deep neural network model that same depth, different number neuron, different convolution pond organizational forms are formed, then using mark
The historical data being poured in is trained model by the way of signal propagated forward, error back propagation.When the output of model
Error amount between label be less than it is preset meet business need threshold value when, deconditioning.
S330：Update is trained to the detection model according to the ladle image data of reception and prediction result.
As shown in fig. 6, the step S330 includes：
S331：Ladle image data and prediction result are stored, and the training data of detection model is updated.
S332：Detection model is trained according to updated training data, updates detection model.
The present embodiment three is identical with the principle of embodiment two, and so it will not be repeated.
Example IV
The present embodiment four corresponds to embodiment one, provides a kind of ladle defects detection sorter.Referring to Fig. 7, its
The connection block diagram of ladle defects detection sorter for the present embodiment four.
The embodiment of the present invention four provides a kind of ladle defects detection sorter, including：
Receiving module 110 for receiving ladle image data to be detected, and generates the detection request of ladle defect.
Scheduler module 120, for according to load balancing and scheduling strategy, the ladle image data and detection request to be sent out
It send to the best server for carrying detection model.
Result of calculation receiving module 130, for receive by detection model ladle image data is carried out it is defeated after prediction calculating
The prediction result gone out, the prediction result include the classification of ladle defect.
Respond module 140 for the prediction result exported according to detection model, performs corresponding response action.
The present embodiment four is identical with the principle of embodiment one, and so it will not be repeated.
Embodiment five
The present embodiment five corresponds to embodiment two, provides a kind of ladle defects detection sorter.Referring to Fig. 8, its
The connection block diagram of ladle defects detection sorter for the present embodiment five.
The embodiment of the present invention five provides a kind of ladle defects detection sorter, including：
Receiving module 210 for receiving ladle image data to be detected, and generates the detection request of ladle defect.
Scheduler module 220, for according to load balancing and scheduling strategy, the ladle image data and detection request to be sent out
It send to the best server for carrying detection model.
Result of calculation receiving module 230, for receive by detection model ladle image data is carried out it is defeated after prediction calculating
The prediction result gone out, the prediction result include the classification of ladle defect.
Respond module 240 for the prediction result exported according to detection model, performs corresponding response action.
Instruction sending module 250 is updated, for sending more new command to best server, by best server according to reception
Ladle image data and prediction result update is trained to the detection model.
The present embodiment four is identical with the principle of embodiment two, and so it will not be repeated.
Embodiment six
The present embodiment six corresponds to embodiment three, provides a kind of ladle defects detection sorter.Referring to Fig. 9, its
The connection block diagram of ladle defects detection sorter for the present embodiment six.The present embodiment six provides a kind of ladle defects detection
Sorter, including：
Request receiving module 310 is detected, for receiving the detection of ladle image data and ladle defect request；
Computing module 320, the prediction that ladle defect is carried out for passing through detection model to the image data of ladle calculate simultaneously
Prediction result is exported, the prediction result includes the classification of ladle defect.
The detection model includes：Depth convolutional neural networks and defect location sorter network.
The depth convolutional neural networks are used to extract the feature of ladle picture, and the feature is input to the defect
It positions in sorter network.
The depth convolutional neural networks include：
Convolutional layer for being scanned convolution to ladle image using the different convolution kernel of weights, therefrom extracts various meanings
The feature of justice, and export into characteristic pattern.
Pond layer, for carrying out dimensionality reduction operation to characteristic pattern.
Full articulamentum, for by the Feature Mapping in characteristic pattern to defect location sorter network.
The defect location sorter network be used to judging feature that depth convolutional neural networks are extracted whether existing defects,
And obtain the classification of the defect.
Update module 330 instructs the detection model for the ladle image data according to reception and prediction result
Practice update.
The update module 330 includes：
Sub-module stored 331, for storing ladle image data and prediction result, and to the training data of detection model into
Row update.
Training submodule 332, for being trained according to updated training data to detection model, update detection mould
Type.
The application mode of the present embodiment four is introduced in detail below：As shown in Figure 10, it is the application signal of the present embodiment six
Figure.In practice, camera may be used and the image of ladle is acquired, and be sent to prediction engine and creation data
Library.After wherein prediction engine receives ladle image, according to the loading condition of detection model, detection request is sent to best
On server.After the calculating of detection model, testing result is exported, and be sent to controller.Controller according to testing result,
Control alarm is alarmed, while detection daily record is stored into Production database.
Then, then for detection model be updated operation, can by by the data update in Production database to instruction
Practice in database, then detection model is trained by training engine, updates detection model.
Embodiment seven
The embodiment of the present invention seven provides a kind of ladle defects detection sorting device, and as shown in figure 11, which includes：Storage
Device 410 and processor 420,410 memory of memory contain the computer program that can be run on processor 420.The processor
The ladle defects detection sorting technique in above-described embodiment is realized during the 420 execution computer program.410 He of memory
The quantity of processor 420 can be one or more.
The equipment further includes：
Communication interface 430 for communicating with external device, carries out data interaction.
Memory 410 may include high-speed RAM memory, it is also possible to further include nonvolatile memory (non-
Volatile memory), a for example, at least magnetic disk storage.
If memory 410, processor 420 and the independent realization of communication interface 430, memory 410,420 and of processor
Communication interface 430 can be connected with each other by bus and complete mutual communication.The bus can be Industry Standard Architecture
Structure (ISA, Industry Standard Architecture) bus, external equipment interconnection (PCI, Peripheral
Component) bus or extended industry-standard architecture (EISA, Extended Industry Standard
Component) bus etc..The bus can be divided into address bus, data/address bus, controlling bus etc..For ease of representing, Figure 11
In only represented with a thick line, it is not intended that an only bus or a type of bus.
Optionally, in specific implementation, if memory 410, processor 420 and communication interface 430 are integrated in one piece of core
On piece, then memory 410, processor 420 and communication interface 430 can complete mutual communication by internal interface.
In the description of this specification, reference term " one embodiment ", " example ", " is specifically shown " some embodiments "
The description of example " or " some examples " etc. means specific features, structure, material or the spy for combining the embodiment or example description
Point is contained at least one embodiment of the present invention or example.Moreover, particular features, structures, materials, or characteristics described
It may be combined in any suitable manner in any one or more of the embodiments or examples.In addition, without conflicting with each other, this
The technical staff in field can be by the different embodiments or examples described in this specification and the spy of different embodiments or examples
Sign is combined.
In addition, term " first ", " second " are only used for description purpose, and it is not intended that instruction or hint relative importance
Or the implicit quantity for indicating indicated technical characteristic." first " is defined as a result, the feature of " second " can be expressed or hidden
Include at least one this feature containing ground.In the description of the present invention, " multiple " are meant that two or more, unless otherwise
It is clearly specific to limit.
Any process described otherwise above or method description are construed as in flow chart or herein, represent to include
Module, segment or the portion of the code of the executable instruction of one or more the step of being used to implement specific logical function or process
Point, and the range of the preferred embodiment of the present invention includes other realization, wherein can not press shown or discuss suitable
Sequence, including according to involved function by it is basic simultaneously in the way of or in the opposite order, to perform function, this should be of the invention
Embodiment person of ordinary skill in the field understood.
Expression or logic and/or step described otherwise above herein in flow charts, for example, being considered use
In the order list for the executable instruction for realizing logic function, may be embodied in any computer-readable medium, for
Instruction execution system, device or equipment (such as computer based system, including the system of processor or other can be held from instruction
The system of row system, device or equipment instruction fetch and execute instruction) it uses or combines these instruction execution systems, device or set
It is standby and use.For the purpose of this specification, " computer-readable medium " can any can be included, store, communicate, propagate or pass
Defeated program is for instruction execution system, device or equipment or the dress used with reference to these instruction execution systems, device or equipment
It puts.
Computer-readable medium described in the embodiment of the present invention can be that computer-readable signal media or computer can
Storage medium either the two is read arbitrarily to combine.The more specific example of computer readable storage medium is at least (non-poor
Property list to the greatest extent) including following：Electrical connection section (electronic device) with one or more wiring, portable computer diskette box (magnetic
Device), random access memory (RAM), read-only memory (ROM), erasable edit read-only storage (EPROM or flash
Memory), fiber device and portable read-only memory (CDROM).In addition, computer readable storage medium even can be with
It is the paper or other suitable media that can print described program on it, because can be for example by being carried out to paper or other media
Optical scanner is then handled described electronically to obtain into edlin, interpretation or when necessary with other suitable methods
Program is then stored in computer storage.
In embodiments of the present invention, computer-readable signal media can be included in a base band or as a carrier wave part
The data-signal of propagation, wherein carrying computer-readable program code.The data-signal of this propagation may be used a variety of
Form, including but not limited to electromagnetic signal, optical signal or above-mentioned any appropriate combination.Computer-readable signal media is also
Can be any computer-readable medium other than computer readable storage medium, which can send, pass
Either transmission is broadcast for instruction execution system, input method or device use or program in connection.Computer can
Reading the program code included on medium can be transmitted with any appropriate medium, including but not limited to：Wirelessly, electric wire, optical cable, penetrate
Frequently (Radio Frequency, RF) etc. or above-mentioned any appropriate combination.
It should be appreciated that each section of the present invention can be realized with hardware, software, firmware or combination thereof.Above-mentioned
In embodiment, software that multiple steps or method can in memory and by suitable instruction execution system be performed with storage
Or firmware is realized.If for example, with hardware come realize in another embodiment, can be under well known in the art
Any one of row technology or their combination are realized：With for the logic gates to data-signal realization logic function
Discrete logic, have suitable combinational logic gate circuit application-specific integrated circuit, programmable gate array (PGA), scene
Programmable gate array (FPGA) etc..
Those skilled in the art are appreciated that realize all or part of step that above-described embodiment method carries
Suddenly it is that relevant hardware can be instructed to complete by program, the program can be stored in a kind of computer-readable storage medium
In matter, the program when being executed, one or a combination set of the step of including embodiment of the method.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing module, it can also
That each unit is individually physically present, can also two or more units be integrated in a module.Above-mentioned integrated mould
The form that hardware had both may be used in block is realized, can also be realized in the form of software function module.The integrated module is such as
Fruit is realized in the form of software function module and is independent product sale or in use, can also be stored in a computer
In readable storage medium storing program for executing.The storage medium can be read-only memory, disk or CD etc..
In conclusion the embodiment of the present invention finally obtains steel by being detected judgement to the ladle image acquired in real time
Position and its affiliated classification where packet defect.Further, the embodiment of the present invention is detected the iteration update of model, makes inspection
The newest demand that model can adapt to production environment is surveyed, is industrial production line in nicety of grading, scalability, standardization etc.
It brings and is obviously improved.
The above description is merely a specific embodiment, but protection scope of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in its various change or replacement,
These should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with the guarantor of the claim
It protects subject to range.
Claims (16)
1. a kind of ladle defects detection sorting technique, which is characterized in that include the following steps：
Ladle image data to be detected is received, and generates the detection request of ladle defect；
According to load balancing and scheduling strategy, the ladle image data and detection request are sent to and carry detection model most
On good server；
The prediction result for ladle image data export after prediction calculating by detection model is received, the prediction result includes
The classification of ladle defect；
According to the prediction result of detection model, corresponding response action is performed.
2. ladle defects detection sorting technique according to claim 1, which is characterized in that it is described according to detection model export
Prediction result after the step of performing corresponding response action, further includes：More new command is sent to best server, by best
Server is trained update according to the ladle image data and prediction result of reception to the detection model.
3. a kind of ladle defects detection sorting technique, which is characterized in that include the following steps：
Receive the detection request of ladle image data and ladle defect；
The prediction for carrying out ladle defect to the image data of ladle by detection model calculates and exports prediction result, the prediction
As a result include the classification of ladle defect.
4. ladle defects detection sorting technique according to claim 3, which is characterized in that the detection model includes：Depth
Convolutional neural networks and defect location sorter network；
The depth convolutional neural networks are used to extract the feature of ladle picture, and the feature is input to the defect location
In sorter network；
The defect location sorter network be used to judging feature that depth convolutional neural networks are extracted whether existing defects, and obtain
Obtain the classification of the defect.
5. ladle defects detection sorting technique according to claim 4, which is characterized in that the depth convolutional neural networks packet
It includes：
Convolutional layer for being scanned convolution to ladle image using the different convolution kernel of weights, therefrom extracts various meanings
Feature, and export into characteristic pattern；
Pond layer, for carrying out dimensionality reduction operation to characteristic pattern；
Full articulamentum, for by the Feature Mapping extracted to defect location sorter network.
6. ladle defects detection sorting technique according to claim 3, which is characterized in that further include step：According to reception
Ladle image data and prediction result are trained update to the detection model.
7. ladle defects detection sorting technique according to claim 6, which is characterized in that the ladle picture according to reception
The step of data and prediction result are trained update to the detection model includes：
Ladle image data and prediction result are stored, and the training data of detection model is updated；
Detection model is trained according to updated training data, updates detection model.
8. a kind of ladle defects detection sorter, which is characterized in that including：
Receiving module for receiving ladle image data to be detected, and generates the detection request of ladle defect；
Scheduler module, for according to load balancing and scheduling strategy, the ladle image data and detection request being sent to and being taken
On the best server for carrying detection model；
Result of calculation receiving module, for receiving the prediction for ladle image data export after prediction calculating by detection model
As a result, the prediction result includes the classification of ladle defect；
Respond module for the prediction result exported according to detection model, performs corresponding response action.
9. ladle defects detection sorter according to claim 8, which is characterized in that further include more new command and send mould
Block, for sending more new command to best server, by best server according to the ladle image data and prediction result of reception
Update is trained to the detection model.
10. a kind of ladle defects detection sorter, which is characterized in that including：
Request receiving module is detected, for receiving the detection of ladle image data and ladle defect request；
Computing module, the prediction that ladle defect is carried out for passing through detection model to the image data of ladle calculate and export prediction
As a result, the prediction result includes the classification of ladle defect.
11. ladle defects detection sorter according to claim 10, which is characterized in that the detection model includes：It is deep
Spend convolutional neural networks and defect location sorter network；
The depth convolutional neural networks are used to extract the feature of ladle picture, and the feature is input to the defect location
In sorter network；
The defect location sorter network be used to judging feature that depth convolutional neural networks are extracted whether existing defects, and obtain
Obtain the classification of the defect.
12. the ladle defects detection sorter according to claim 11, which is characterized in that the depth convolutional neural networks
Including：
Convolutional layer for being scanned convolution to ladle image using the different convolution kernel of weights, therefrom extracts various meanings
Feature, and export into characteristic pattern；
Pond layer, for carrying out dimensionality reduction operation to characteristic pattern；
Full articulamentum, for by the Feature Mapping in characteristic pattern to defect location sorter network.
13. ladle defects detection sorter according to claim 10, which is characterized in that further include update module, be used for
Update is trained to the detection model according to the ladle image data of reception and prediction result.
14. the ladle defects detection sorter according to claim 13, which is characterized in that the update module includes：
Sub-module stored for storing ladle image data and prediction result, and is updated the training data of detection model；
Training submodule for being trained according to updated training data to detection model, updates detection model.
15. a kind of ladle defects detection sorting device, which is characterized in that the equipment includes：
One or more processors；
Storage device, for storing one or more programs；
When one or more of programs are performed by one or more of processors so that one or more of processors
Realize the ladle defects detection sorting technique as described in any in claim 1-7.
16. a kind of computer-readable medium, is stored with computer program, which is characterized in that when the program is executed by processor
Realize the ladle defects detection sorting technique as described in any in claim 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810020267.5A CN108230318B (en) | 2018-01-09 | 2018-01-09 | Ladle defects detection classification method, device, equipment and computer-readable medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810020267.5A CN108230318B (en) | 2018-01-09 | 2018-01-09 | Ladle defects detection classification method, device, equipment and computer-readable medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108230318A true CN108230318A (en) | 2018-06-29 |
| CN108230318B CN108230318B (en) | 2019-04-16 |
Family
ID=62640771
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810020267.5A Active CN108230318B (en) | 2018-01-09 | 2018-01-09 | Ladle defects detection classification method, device, equipment and computer-readable medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108230318B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108846841A (en) * | 2018-07-02 | 2018-11-20 | 北京百度网讯科技有限公司 | Display screen quality determining method, device, electronic equipment and storage medium |
| CN108921841A (en) * | 2018-07-02 | 2018-11-30 | 北京百度网讯科技有限公司 | Continuous casting of iron and steel base quality determining method, device, equipment and storage medium |
| CN108921839A (en) * | 2018-07-02 | 2018-11-30 | 北京百度网讯科技有限公司 | Continuous casting billet quality detection method, device, electronic equipment and storage medium |
| CN108961239A (en) * | 2018-07-02 | 2018-12-07 | 北京百度网讯科技有限公司 | Continuous casting billet quality detection method, device, electronic equipment and storage medium |
| CN108982522A (en) * | 2018-08-09 | 2018-12-11 | 北京百度网讯科技有限公司 | Method and apparatus for detecting defect of pipeline |
| CN109087282A (en) * | 2018-07-02 | 2018-12-25 | 北京百度网讯科技有限公司 | Display screen peripheral circuit detection method, device, electronic equipment and storage medium |
| CN109087281A (en) * | 2018-07-02 | 2018-12-25 | 北京百度网讯科技有限公司 | Display screen peripheral circuit detection method, device, electronic equipment and storage medium |
| CN109087280A (en) * | 2018-07-02 | 2018-12-25 | 北京百度网讯科技有限公司 | Continuous casting billet quality detection method, device, control equipment, server and medium |
| CN109584235A (en) * | 2018-11-30 | 2019-04-05 | 北京百度网讯科技有限公司 | Quality determining method, device, server and the storage medium of plastic foam cutlery box |
| CN109598712A (en) * | 2018-11-30 | 2019-04-09 | 北京百度网讯科技有限公司 | Quality determining method, device, server and the storage medium of plastic foam cutlery box |
| CN109682820A (en) * | 2018-11-05 | 2019-04-26 | 苏州佳智彩光电科技有限公司 | A kind of defect automatic optical detection method of display screen |
| CN110120044A (en) * | 2019-05-20 | 2019-08-13 | 北京百度网讯科技有限公司 | Article defects detection method and device, edge device, server |
| WO2020007119A1 (en) * | 2018-07-02 | 2020-01-09 | 北京百度网讯科技有限公司 | Display screen peripheral circuit detection method and device, electronic device and storage medium |
| WO2020007095A1 (en) * | 2018-07-02 | 2020-01-09 | 北京百度网讯科技有限公司 | Display screen quality inspection method and apparatus, electronic device, and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104850845A (en) * | 2015-05-30 | 2015-08-19 | 大连理工大学 | Traffic sign recognition method based on asymmetric convolution neural network |
| CN105118044A (en) * | 2015-06-16 | 2015-12-02 | 华南理工大学 | Method for automatically detecting defects of wheel-shaped cast product |
| WO2016168235A1 (en) * | 2015-04-17 | 2016-10-20 | Nec Laboratories America, Inc. | Fine-grained image classification by exploring bipartite-graph labels |
| CN107392896A (en) * | 2017-07-14 | 2017-11-24 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of Wood Defects Testing method and system based on deep learning |
-
2018
- 2018-01-09 CN CN201810020267.5A patent/CN108230318B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016168235A1 (en) * | 2015-04-17 | 2016-10-20 | Nec Laboratories America, Inc. | Fine-grained image classification by exploring bipartite-graph labels |
| CN104850845A (en) * | 2015-05-30 | 2015-08-19 | 大连理工大学 | Traffic sign recognition method based on asymmetric convolution neural network |
| CN105118044A (en) * | 2015-06-16 | 2015-12-02 | 华南理工大学 | Method for automatically detecting defects of wheel-shaped cast product |
| CN107392896A (en) * | 2017-07-14 | 2017-11-24 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of Wood Defects Testing method and system based on deep learning |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020007119A1 (en) * | 2018-07-02 | 2020-01-09 | 北京百度网讯科技有限公司 | Display screen peripheral circuit detection method and device, electronic device and storage medium |
| CN108921841A (en) * | 2018-07-02 | 2018-11-30 | 北京百度网讯科技有限公司 | Continuous casting of iron and steel base quality determining method, device, equipment and storage medium |
| CN108921839A (en) * | 2018-07-02 | 2018-11-30 | 北京百度网讯科技有限公司 | Continuous casting billet quality detection method, device, electronic equipment and storage medium |
| CN108961239A (en) * | 2018-07-02 | 2018-12-07 | 北京百度网讯科技有限公司 | Continuous casting billet quality detection method, device, electronic equipment and storage medium |
| CN108846841A (en) * | 2018-07-02 | 2018-11-20 | 北京百度网讯科技有限公司 | Display screen quality determining method, device, electronic equipment and storage medium |
| CN109087282A (en) * | 2018-07-02 | 2018-12-25 | 北京百度网讯科技有限公司 | Display screen peripheral circuit detection method, device, electronic equipment and storage medium |
| CN109087281A (en) * | 2018-07-02 | 2018-12-25 | 北京百度网讯科技有限公司 | Display screen peripheral circuit detection method, device, electronic equipment and storage medium |
| CN109087280A (en) * | 2018-07-02 | 2018-12-25 | 北京百度网讯科技有限公司 | Continuous casting billet quality detection method, device, control equipment, server and medium |
| US11488294B2 (en) | 2018-07-02 | 2022-11-01 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method for detecting display screen quality, apparatus, electronic device and storage medium |
| US11380232B2 (en) | 2018-07-02 | 2022-07-05 | Beijing Baidu Netcom Science Technology Co., Ltd. | Display screen quality detection method, apparatus, electronic device and storage medium |
| JP2020529056A (en) * | 2018-07-02 | 2020-10-01 | 北京百度網訊科技有限公司 | Display quality detection methods, devices, electronic devices and storage media |
| WO2020007095A1 (en) * | 2018-07-02 | 2020-01-09 | 北京百度网讯科技有限公司 | Display screen quality inspection method and apparatus, electronic device, and storage medium |
| CN108982522A (en) * | 2018-08-09 | 2018-12-11 | 北京百度网讯科技有限公司 | Method and apparatus for detecting defect of pipeline |
| CN109682820A (en) * | 2018-11-05 | 2019-04-26 | 苏州佳智彩光电科技有限公司 | A kind of defect automatic optical detection method of display screen |
| CN109598712A (en) * | 2018-11-30 | 2019-04-09 | 北京百度网讯科技有限公司 | Quality determining method, device, server and the storage medium of plastic foam cutlery box |
| CN109584235A (en) * | 2018-11-30 | 2019-04-05 | 北京百度网讯科技有限公司 | Quality determining method, device, server and the storage medium of plastic foam cutlery box |
| CN110120044A (en) * | 2019-05-20 | 2019-08-13 | 北京百度网讯科技有限公司 | Article defects detection method and device, edge device, server |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108230318B (en) | 2019-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108230317A (en) | Steel plate defect detection sorting technique, device, equipment and computer-readable medium | |
| CN108230318B (en) | Ladle defects detection classification method, device, equipment and computer-readable medium | |
| CN108320278A (en) | Product defects detect localization method, device, equipment and computer-readable medium | |
| CN108154508B (en) | Method, apparatus, storage medium and the terminal device of product defects detection positioning | |
| CN108257121B (en) | Method, apparatus, storage medium and the terminal device that product defects detection model updates | |
| CN108564104A (en) | Product defects detection method, device, system, server and storage medium | |
| CN105574550A (en) | Vehicle identification method and device | |
| CN108242054A (en) | A kind of steel plate defect detection method, device, equipment and server | |
| CN107844417A (en) | Method for generating test case and device | |
| CN108257122A (en) | Paper sheet defect detection method, device and server based on machine vision | |
| CN111612002A (en) | Multi-target object motion tracking method based on neural network | |
| CN109635825A (en) | Vehicle attribute detection method, device and storage medium | |
| US11748975B2 (en) | Method and device for optimizing object-class model based on neural network | |
| CN110197147A (en) | Building Cass collection method, apparatus, storage medium and the equipment of remote sensing image | |
| CN109903242A (en) | A kind of image generating method and device | |
| CN113887608A (en) | Model training method, image detection method and device | |
| CN108229658A (en) | The implementation method and device of object detector based on finite sample | |
| CN109903287A (en) | Quality determining method and device | |
| CN116977257A (en) | Defect detection method, device, electronic apparatus, storage medium, and program product | |
| CN108182444A (en) | The method and device of video quality diagnosis based on scene classification | |
| CN108228863B (en) | Method, apparatus, storage medium and the terminal device of information inquiry | |
| CN116843983A (en) | Pavement disease recognition method, model training method, electronic equipment and medium | |
| CN116522565A (en) | BIM-based power engineering design power distribution network planning method and computer equipment | |
| CN115690204A (en) | Hole site measuring method, device, equipment and medium for manufactured part | |
| CN112001388B (en) | Method for detecting circular target in PCB based on YOLOv3 improved model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |