CN106952250A - A kind of metal plate and belt detection method of surface flaw and device based on Faster R CNN networks - Google Patents
A kind of metal plate and belt detection method of surface flaw and device based on Faster R CNN networks Download PDFInfo
- Publication number
- CN106952250A CN106952250A CN201710110525.4A CN201710110525A CN106952250A CN 106952250 A CN106952250 A CN 106952250A CN 201710110525 A CN201710110525 A CN 201710110525A CN 106952250 A CN106952250 A CN 106952250A
- Authority
- CN
- China
- Prior art keywords
- image
- metal plate
- defect
- rectangle frame
- faster
- 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
- G06T7/0008—Industrial image inspection checking presence/absence
-
- 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
- G01N2021/8887—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
- G06T2207/20—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)
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The present invention provides a kind of metal plate and belt detection method of surface flaw and device based on Faster R CNN networks, it is possible to increase the accuracy rate of metal plate and belt surface defects detection.Method includes：Gather the image on the metal plate and belt surface for training Faster R CNN networks；The image on the metal plate and belt surface to collecting carries out data enhancing；Defect part in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing is labeled with rectangle frame, the species that the coordinate and rectangle frame for obtaining rectangle frame include defect；Faster R CNN networks are trained according to the species that the coordinate of the image obtained after the image on the metal plate and belt surface collected and data enhancing, and the rectangle frame obtained includes defect with the rectangle frame；Metal strip surface image to be detected is gathered in real time, and according to the Faster R CNN networks trained, whether detection metal plate and belt surface is defective in real time, and the defect detected is classified and positioned.The present invention is applied to technical field of machine vision.
Description
Technical field
The present invention relates to technical field of machine vision, a kind of metal plate and belt based on Faster R-CNN networks is particularly related to
Detection method of surface flaw and device.
Background technology
Metal plate and belt is the indispensable raw material of the industries such as automobile, machine-building, chemical industry, Aero-Space and shipbuilding.Gold
Category strip surface defect refers to metal plate and belt during production and processing, because technique or other a variety of causes cause metallic plate
Belt surface regional area is physically or chemically uneven.Common metal plate and belt surface defect has roll marks, spot, cut, hole
Hole, holiday, depression, bubble, foreign matter, peeling etc..Surface defect is the higher position of atom active, usually as metal erosion
Place is originated, the presence of surface defect can substantially reduce the fatigue resistance of part, damage the quality of piece surface, influence machine,
The performance of instrument and life-span, influence the performance and quality of its final products.Therefore metal plate and belt surface defect is detected in time,
Order of severity evaluation is carried out to defect, for improving surface quality and product economy benefit important in inhibiting.
With continuing to develop for China's industrialized level, requirement to metal plate and belt surface quality also more and more higher, how
Rapidly and accurately detect surface defect as a link very crucial during rolling metal plate and tape.The table of current main flow
Planar defect detection technique is the machine vision technique based on image procossing, and two steps are divided into substantially：
(1) first the image obtained by industrial camera is handled, then extracts feature, the method for this step mainly has
The class of statistic law, Structure Method, Spectrum Method, modelling etc. four, wherein using it is relatively broad be statistic law and Spectrum Method [2]；
(2) by the grader trained, the feature that (1) is extracted is classified, had at present using more grader
BP neural network, SVM etc..
But these methods have a low contrast between defect and non-defective region, noise and fine defects it is similar
Property, detection speed is slow and the low problem of accuracy of identification, it is impossible to meet industrial accuracy and property is required.
The content of the invention
Lack the technical problem to be solved in the present invention is to provide a kind of metal plate and belt surface based on Faster R-CNN networks
Detection method and device are fallen into, to solve that present in prior art industrial accuracy and requirement of real-time can not be met
Problem.
In order to solve the above technical problems, the embodiment of the present invention provides a kind of metal plate and belt based on Faster R-CNN networks
Detection method of surface flaw, including：
Gather the image on the metal plate and belt surface for training Faster R-CNN networks；
The image on the metal plate and belt surface to collecting carries out data enhancing；
Defect part in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing is used
Rectangle frame is labeled, the species that the coordinate and the rectangle frame for obtaining the rectangle frame include defect；
The image obtained after being strengthened according to the image on the metal plate and belt surface collected and data, and acquisition are described
The species that the coordinate of rectangle frame includes defect with the rectangle frame is trained to Faster R-CNN networks；
Metal strip surface image to be detected is gathered in real time, is adopted according to the Faster R-CNN networks trained and in real time
The metal strip surface image to be detected collected, whether detection metal plate and belt surface is defective in real time, and to lacking for detecting
It is trapped into row classification and positions.
Further, the described pair of image on the metal plate and belt surface collected, which carries out data enhancing, includes：
The image on the metal plate and belt surface by upset, translation strategy to collecting carries out data enhancing.
Further, in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
Defect part be labeled with rectangle frame, the species bag that the coordinate and the rectangle frame for obtaining the rectangle frame include defect
Include：
Defect part in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing is used
Rectangle frame is labeled, and obtains the center point coordinate of the rectangle frame, the width of the rectangle frame, height, and the rectangle frame
The species of included defect；
By the image obtained after the image on the metal plate and belt surface collected and data enhancing, the rectangle obtained
The center point coordinate of frame, the width of the rectangle frame include the species generation data set of defect with height and the rectangle frame.
Further, the image obtained after the image according to the metal plate and belt surface collected and data enhancing,
And the species that the coordinate of the rectangle frame obtained includes defect with the rectangle frame is trained to FasterR-CNN networks
Before, methods described also includes：
According to the defect classification to be detected, total defect classification that modification Faster R-CNN networks can be detected, and be
The fixed numbering of each defect classification to be detected distribution.
Further, the image on the metal plate and belt surface that the basis is collected in real time and data are obtained after strengthening
Image, and the coordinate species that includes defect with the rectangle frame of the rectangle frame obtained are entered to Faster R-CNN networks
Row training includes：
Faster R-CNN networks are trained according to the data set of generation.
The embodiment of the present invention also provides a kind of metal plate and belt surface defect detection apparatus based on Faster R-CNN networks,
Including：
Acquisition module, the image for gathering the metal plate and belt surface for being used to train Faster R-CNN networks；
Strengthen module, the image for the metal plate and belt surface to collecting carries out data enhancing；
Acquisition module, for will be after the image on the metal plate and belt surface collected and data enhancing in obtained image
Defect part be labeled with rectangle frame, the species that the coordinate and the rectangle frame for obtaining the rectangle frame include defect；
Training module, for the figure obtained after the image according to the metal plate and belt surface collected and data enhancing
Picture, and the coordinate of the rectangle frame obtained include the species of defect to the progress of FasterR-CNN networks with the rectangle frame
Training；
Detection module, for gathering metal strip surface image to be detected in real time, according to the FasterR-CNN trained
Network and the metal strip surface image to be detected collected in real time, whether detection metal plate and belt surface is defective in real time, and
The defect detected is classified and positioned.
Further, the enhancing module, for passing through upset, the tactful metal plate and belt surface to collecting of translation
Image carry out data enhancing.
Further, the acquisition module includes：
Unit is marked, for will be after the image on the metal plate and belt surface collected and data enhancing in obtained image
Defect part be labeled with rectangle frame；
Acquiring unit, the center point coordinate for obtaining the rectangle frame, the width of the rectangle frame, height, and it is described
Rectangle frame includes the species of defect；
Generation unit, for will the image on the metal plate and belt surface collected and data enhancing after obtain image,
The center point coordinate of the rectangle frame obtained, the width of the rectangle frame include defect with height and the rectangle frame
Species generates data set.
Further, described device also includes：
Numbering module, for according to the defect classification to be detected, it is total that modification Faster R-CNN networks can be detected
Defect classification, and be the fixed numbering of each the defect classification to be detected distribution.
Further, the training module, is trained for the data set according to generation to Faster R-CNN networks.
The above-mentioned technical proposal of the present invention has the beneficial effect that：
In such scheme, the image on the metal plate and belt surface for training Faster R-CNN networks is gathered；To collecting
The metal plate and belt surface image carry out data enhancing；The image on the metal plate and belt surface collected and data are increased
Defect part in the image obtained after strong is labeled with rectangle frame, obtains the coordinate of the rectangle frame and the rectangle frame institute
Species comprising defect；According to the image obtained after the image on the metal plate and belt surface collected and data enhancing, and obtain
The species that the coordinate of the rectangle frame taken includes defect with the rectangle frame is trained to Faster R-CNN networks；It is real
When gather metal strip surface image to be detected, according to the Faster R-CNN networks trained and collect in real time it is to be checked
The metal strip surface image of survey, whether detection metal plate and belt surface is defective in real time, and the defect detected is classified
And positioning, not only meet the demand that industrial production detects metal plate and belt surface defect in real time, moreover it is possible to improve metal plate and belt surface
The accuracy rate of defects detection.
Brief description of the drawings
Fig. 1 is the metal plate and belt detection method of surface flaw provided in an embodiment of the present invention based on Faster R-CNN networks
Schematic flow sheet；
Fig. 2 is the operation principle schematic diagram of Faster R-CNN networks provided in an embodiment of the present invention；
Fig. 3 is the metal strip surface image to be detected provided in an embodiment of the present invention collected in real time；
Fig. 4 is the metal plate and belt detection method of surface flaw provided in an embodiment of the present invention based on Faster R-CNN networks
Design sketch after detection；
Fig. 5 is the metal plate and belt surface defect detection apparatus provided in an embodiment of the present invention based on Faster R-CNN networks
Structural representation.
Embodiment
To make the technical problem to be solved in the present invention, technical scheme and advantage clearer, below in conjunction with accompanying drawing and tool
Body embodiment is described in detail.
The present invention for it is existing can not meet industrial accuracy and requirement of real-time the problem of there is provided a kind of base
In the metal plate and belt detection method of surface flaw and device of Faster R-CNN networks.
Embodiment one
Referring to shown in Fig. 1, the metal plate and belt surface defect provided in an embodiment of the present invention based on Faster R-CNN networks
Detection method, including：
S101, gathers the image on the metal plate and belt surface for training Faster R-CNN networks；
S102, the image on the metal plate and belt surface to collecting carries out data enhancing；
Defective part in S103, the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
Divide and be labeled with rectangle frame, the species that the coordinate and the rectangle frame for obtaining the rectangle frame include defect；
S104, according to the image obtained after the image on the metal plate and belt surface collected and data enhancing, and is obtained
The coordinate species that includes defect with the rectangle frame of the rectangle frame Faster R-CNN networks are trained；
S105, gathers metal strip surface image to be detected in real time, according to the Faster R-CNN networks that train and
The metal strip surface image to be detected collected in real time, whether detection metal plate and belt surface is defective in real time, and to detection
To defect classified and positioned.
The metal plate and belt detection method of surface flaw based on Faster R-CNN networks described in the embodiment of the present invention, collection
For the image on the metal plate and belt surface for training Faster R-CNN networks；To the image on the metal plate and belt surface collected
Carry out data enhancing；Defective part in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
Divide and be labeled with rectangle frame, the species that the coordinate and the rectangle frame for obtaining the rectangle frame include defect；According to collection
To the metal plate and belt surface image and data enhancing after obtained image, and the rectangle frame obtained coordinate and institute
The species that rectangle frame includes defect is stated to be trained Faster R-CNN networks；Metal plate and belt table to be detected is gathered in real time
Face image, it is real according to the Faster R-CNN networks trained and the metal strip surface image to be detected collected in real time
When detection metal plate and belt surface it is whether defective, and the defect detected is classified and positioned, not only meet industrial life
Production detects the demand of metal plate and belt surface defect in real time, moreover it is possible to improve the accuracy rate of metal plate and belt surface defects detection.
In the present embodiment, in order to more fully understand the metal plate and belt surface defects detection side based on Faster R-CNN networks
Method, first the operation principle to Faster R-CNN networks is illustrated, as shown in Fig. 2 FasterR-CNN is main by two parts
Composition：(1) it is used for candidate region network (the Region Proposal for generating region proposal (candidate region)
Network, RPN), RPN is a full convolutional network, and its main function is to carry out calculating analysis to the convolutional layer feature of picture,
Then under different image scaleds, generating some rectangle frames for different defect types, (coordinate of rectangle frame is by four
Count to represent, frame center point coordinate x and y, height h, the same pictures of width w) can produce thousands of rectangle frames, these rectangles
Frame is exactly region proposal (region for being likely to be defect)；(2) defects detection is carried out to region proposal
Depth convolutional neural networks Fast R-CNN, Fast R-CNN obtained region proposal exported to RPN calculated
Analysis, weeds out the region proposal of redundancy or mistake, obtains optimal rectangle frame and category score, as last inspection
Survey result；In Fig. 2, image represents image；Conv layers represent convolutional layer；Feature maps represent Feature Mapping；
Proposals represents candidate region；Classifier presentation class devices；RoI pooling full name are region of
Interest pooling, Chinese is area-of-interest pond layer/RoI ponds layer.
In the embodiment of the foregoing metal plate and belt detection method of surface flaw based on Faster R-CNN networks,
Further, the described pair of image on the metal plate and belt surface collected, which carries out data enhancing, includes：
The image on the metal plate and belt surface by upset, translation strategy to collecting carries out data enhancing.
In the present embodiment, the metal plate and belt surface for training Faster R-CNN networks can be gathered by industrial camera
Image, and the strategies such as upset, translation passed through to the image on the metal plate and belt surface of collection carry out data enhancing.
In the present embodiment, data enhancing is very few or in the case that data are difficult to obtain in data intensive data, to existing
Data set be enlarged by strategies such as turnover translation movings, obtained data set include original data and enhanced data.
Same object, is observed under different angles and different backgrounds, and obtained image may be entirely different, and computer is possibly can not
These images are correctly recognized, therefore mirror image switch is carried out to image and moved horizontally, data set can be made to include same figure
Various data of the piece in different angle diverse locations.
In the embodiment of the foregoing metal plate and belt detection method of surface flaw based on Faster R-CNN networks,
Further, the defective part in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
Divide and be labeled with rectangle frame, the species that the coordinate for obtaining the rectangle frame includes defect with the rectangle frame includes：
Defect part in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing is used
Rectangle frame is labeled, and obtains the center point coordinate of the rectangle frame, the width of the rectangle frame, height, and the rectangle frame
The species of included defect；
By the image obtained after the image on the metal plate and belt surface collected and data enhancing, the rectangle obtained
The center point coordinate of frame, the width of the rectangle frame include the species generation data set of defect with height and the rectangle frame.
In the present embodiment, in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
Defect part is labeled with rectangle frame, and the species that the coordinate and the rectangle frame for taking the rectangle frame include defect is (specific
's：Five parameters of each rectangle frame are obtained, five parameters of the rectangle frame include：The center point coordinate of rectangle frame is (described
Center point coordinate is two-dimensional coordinate), width and the height, and rectangle frame of rectangle frame include the species of defect), it will collect
The metal plate and belt surface image and data enhancing after obtain image, this five parameters generation texts, text text
Part includes five parameters of all rectangle frames in every pictures, so that data set is obtained, according to the data set pair of generation
FasterR-CNN networks are trained, wherein, data set includes whole images and text.
In the present embodiment, the data set includes three parts, and Part I is training set, for training FasterR-CNN
Network；Part II part is checking collection, for adjusting Faster R-CNN network hyper parameters；Part III is test set, is used
To weigh network performance.
In the present embodiment, for each bounding box when calculating confidence level, it can be calculated for each defect kind
The confidence level of class.In the embodiment of the foregoing metal plate and belt detection method of surface flaw based on Faster R-CNN networks
In, further, the image obtained after the image according to the metal plate and belt surface collected and data enhancing, and obtain
The rectangle frame coordinate and the rectangle frame species that includes defect Faster R-CNN networks are trained before,
Methods described also includes：
According to the defect classification to be detected, total defect classification that modification Faster R-CNN networks can be detected, and be
The fixed numbering of each defect classification to be detected distribution.
In the present embodiment, according to the defect classification to be detected (such as roll marks, spot, cut, hole, holiday, depression, gas
Bubble, foreign matter, peeling etc.), total defect classification that modification Faster R-CNN networks can be detected, and to be detected for each
The fixed numbering of defect classification distribution.
In the present embodiment, when the good Faster R-CNN networks of application training detect metal plate and belt surface defect, reality is needed
When gather metal strip surface image metal plate and belt surface defect to be detected, further according to the Faster R-CNN networks trained
The metal strip surface image to be detected collected in real time, whether detection metal plate and belt surface is defective in real time, and to inspection
The defect measured is classified and positioned, the metal plate and belt surface defect based on Faster R-CNN networks that the present embodiment is provided
Detection method, can fast and accurately detect the defect on metal plate and belt surface, have higher for various types of defects
Accuracy rate, testing result can reach more than 95 percent accuracy rate, and the defect for partial category can even reach
To absolutely accuracy rate, Fig. 3 and Fig. 4 is the metal strip surface image to be detected gathered in real time respectively with using this reality
Design sketch after the detection of the metal plate and belt detection method of surface flaw based on Faster R-CNN networks of example offer is provided.
In the present embodiment, industrial camera can be set up on metal plate and belt production line to gather metal to be detected in real time
Strip surface image, industrial camera is connected with computer, the metal plate and belt surface to be detected collected whenever industrial camera
Image and when being transferred to computer, the Faster R-CNN networks trained can realize the real-time processing to image, by image
In defect marked out come and exported with rectangle frame.
Embodiment two
The present invention also provides a kind of the specific of metal plate and belt surface defect detection apparatus based on Faster R-CNN networks
Embodiment, the metal plate and belt surface defect detection apparatus based on Faster R-CNN networks provided due to the present invention with it is foregoing
The embodiment of metal plate and belt detection method of surface flaw based on Faster R-CNN networks is corresponding, and this is based on
The metal plate and belt surface defect detection apparatus of Faster R-CNN networks can be by performing in above method embodiment
Process step realize the purpose of the present invention, therefore the above-mentioned metal plate and belt surface defect inspection based on Faster R-CNN networks
Explanation in survey method embodiment, is also applied for the metal based on Faster R-CNN networks that the present invention is provided
The embodiment of strip surface defect detection means, will not be described in great detail in the embodiment below the present invention.
Referring to shown in Fig. 5, the embodiment of the present invention also provides a kind of metal plate and belt surface based on Faster R-CNN networks
Defect detecting device, including：
Acquisition module 11, the image for gathering the metal plate and belt surface for being used to train Faster R-CNN networks；
Strengthen module 12, the image for the metal plate and belt surface to collecting carries out data enhancing；
Acquisition module 13, for the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
In defect part be labeled with rectangle frame, the kind that the coordinate and the rectangle frame for obtaining the rectangle frame include defect
Class；
Training module 14, for the figure obtained after the image according to the metal plate and belt surface collected and data enhancing
Picture, and the coordinate of the rectangle frame obtained include the species of defect to the progress of Faster R-CNN networks with the rectangle frame
Training；
Detection module 15, for gathering metal strip surface image to be detected in real time, according to the Faster R- trained
CNN networks and the metal strip surface image to be detected collected in real time, whether detection metal plate and belt surface is defective in real time,
And the defect detected is classified and positioned.
The metal plate and belt surface defect detection apparatus based on Faster R-CNN networks described in the embodiment of the present invention, collection
For the image on the metal plate and belt surface for training Faster R-CNN networks；To the image on the metal plate and belt surface collected
Carry out data enhancing；Defective part in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
Divide and be labeled with rectangle frame, the species that the coordinate and the rectangle frame for obtaining the rectangle frame include defect；According to collection
To the metal plate and belt surface image and data enhancing after obtained image, and the rectangle frame obtained coordinate and institute
The species that rectangle frame includes defect is stated to be trained Faster R-CNN networks；Metal plate and belt table to be detected is gathered in real time
Face image, it is real according to the Faster R-CNN networks trained and the metal strip surface image to be detected collected in real time
When detection metal plate and belt surface it is whether defective, and the defect detected is classified and positioned, not only meet industrial life
The property demand of production, moreover it is possible to improve the accuracy rate of surface defects detection.
In the embodiment of the foregoing metal plate and belt surface defect detection apparatus based on Faster R-CNN networks,
Further, the enhancing module, the image for the metal plate and belt surface by upset, translation strategy to collecting enters
Row data strengthen.
In the embodiment of the foregoing metal plate and belt surface defect detection apparatus based on Faster R-CNN networks,
Further, the acquisition module includes：
Unit is marked, for will be after the image on the metal plate and belt surface collected and data enhancing in obtained image
Defect part be labeled with rectangle frame；
Acquiring unit, the center point coordinate for obtaining the rectangle frame, the width of the rectangle frame, height, and it is described
Rectangle frame includes the species of defect；
Generation unit, for will the image on the metal plate and belt surface collected and data enhancing after obtain image,
The center point coordinate of the rectangle frame obtained, the width of the rectangle frame include defect with height and the rectangle frame
Species generates data set.
In the embodiment of the foregoing metal plate and belt surface defect detection apparatus based on Faster R-CNN networks,
Further, described device also includes：
Numbering module, for according to the defect classification to be detected, it is total that modification Faster R-CNN networks can be detected
Defect classification, and be the fixed numbering of each the defect classification to be detected distribution.
In the embodiment of the foregoing metal plate and belt surface defect detection apparatus based on Faster R-CNN networks,
Further, the training module, is trained for the data set according to generation to Faster R-CNN networks.
The above is the preferred embodiment of the present invention, it is noted that for those skilled in the art
For, on the premise of principle of the present invention is not departed from, some improvements and modifications can also be made, these improvements and modifications
It should be regarded as protection scope of the present invention.
Claims (10)
1. a kind of metal plate and belt detection method of surface flaw based on Faster R-CNN networks, it is characterised in that including：
Gather the image on the metal plate and belt surface for training Faster R-CNN networks；
The image on the metal plate and belt surface to collecting carries out data enhancing；
Defect part rectangle in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
Frame is labeled, the species that the coordinate and the rectangle frame for obtaining the rectangle frame include defect；
According to the image obtained after the image on the metal plate and belt surface collected and data enhancing, and the rectangle obtained
The species that the coordinate of frame includes defect with the rectangle frame is trained to Faster R-CNN networks；
Metal strip surface image to be detected is gathered in real time, is collected according to the Faster R-CNN networks trained and in real time
Metal strip surface image to be detected, whether detection metal plate and belt surface defective in real time, and the defect detected is entered
Row classification and positioning.
2. the metal plate and belt detection method of surface flaw according to claim 1 based on Faster R-CNN networks, it is special
Levy and be, the described pair of image on the metal plate and belt surface collected, which carries out data enhancing, to be included：
The image on the metal plate and belt surface by upset, translation strategy to collecting carries out data enhancing.
3. the metal plate and belt detection method of surface flaw according to claim 1 based on Faster R-CNN networks, it is special
Levy and be, the defect part in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
It is labeled with rectangle frame, the species that the coordinate for obtaining the rectangle frame includes defect with the rectangle frame includes：
Defect part rectangle in the image that will be obtained after the image on the metal plate and belt surface collected and data enhancing
Frame is labeled, and obtains the center point coordinate of the rectangle frame, the width of the rectangle frame, height, and the rectangle frame are wrapped
Species containing defect；
By the image obtained after the image on the metal plate and belt surface collected and data enhancing, the rectangle frame obtained
Center point coordinate, the width of the rectangle frame include the species generation data set of defect with height and the rectangle frame.
4. the metal plate and belt detection method of surface flaw according to claim 1 based on Faster R-CNN networks, it is special
Levy and be, the image obtained after the image according to the metal plate and belt surface collected and data enhancing, and the institute obtained
It is described before the species that the coordinate and the rectangle frame for stating rectangle frame include defect is trained to Faster R-CNN networks
Method also includes：
According to the defect classification to be detected, total defect classification that modification Faster R-CNN networks can be detected, and be each
The fixed numbering of the individual defect classification to be detected distribution.
5. the sheet metal surface defect inspection method according to claim 3 based on Faster R-CNN networks, it is special
Levy and be, the image obtained after image and the data enhancing on the metal plate and belt surface that the basis is collected in real time, and obtain
The coordinate of the rectangle frame taken is trained bag with the species that the rectangle frame includes defect to Faster R-CNN networks
Include：
Faster R-CNN networks are trained according to the data set of generation.
6. a kind of metal plate and belt surface defect detection apparatus based on Faster R-CNN networks, it is characterised in that including：
Acquisition module, the image for gathering the metal plate and belt surface for being used to train Faster R-CNN networks；
Strengthen module, the image for the metal plate and belt surface to collecting carries out data enhancing；
Acquisition module, for will lacking in obtained image after the image on the metal plate and belt surface collected and data enhancing
Fall into part to be labeled with rectangle frame, the species that the coordinate and the rectangle frame for obtaining the rectangle frame include defect；
Training module, for the image obtained after the image according to the metal plate and belt surface collected and data enhancing, and
The species that the coordinate of the rectangle frame obtained includes defect with the rectangle frame is trained to Faster R-CNN networks；
Detection module, for gathering metal strip surface image to be detected in real time, according to the Faster R-CNN nets trained
Network and the metal strip surface image to be detected collected in real time, whether detection metal plate and belt surface is defective and right in real time
The defect detected is classified and positioned.
7. the metal plate and belt surface defect detection apparatus according to claim 6 based on Faster R-CNN networks, it is special
Levy and be, the enhancing module, the image for the metal plate and belt surface by upset, translation strategy to collecting is carried out
Data strengthen.
8. the metal plate and belt surface defect detection apparatus according to claim 6 based on Faster R-CNN networks, it is special
Levy and be, the acquisition module includes：
Unit is marked, for will lacking in obtained image after the image on the metal plate and belt surface collected and data enhancing
Part is fallen into be labeled with rectangle frame；
Acquiring unit, the center point coordinate for obtaining the rectangle frame, the width of the rectangle frame, height, and the rectangle
Frame includes the species of defect；
Generation unit, for will the image on the metal plate and belt surface collected and data enhancing after obtain image, obtain
The center point coordinate of the rectangle frame, the width of the rectangle frame species of defect is included with height and the rectangle frame
Generate data set.
9. the metal plate and belt surface defect detection apparatus according to claim 6 based on Faster R-CNN networks, it is special
Levy and be, described device also includes：
Numbering module, for the total defect that can be detected according to the defect classification to be detected, modification Faster R-CNN networks
Classification, and be the fixed numbering of each the defect classification to be detected distribution.
10. the sheet metal strip surface defect detection means according to claim 8 based on Faster R-CNN networks,
Characterized in that, the training module, is trained for the data set according to generation to Faster R-CNN networks.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710110525.4A CN106952250B (en) | 2017-02-28 | 2017-02-28 | Metal plate strip surface defect detection method and device based on fast R-CNN network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710110525.4A CN106952250B (en) | 2017-02-28 | 2017-02-28 | Metal plate strip surface defect detection method and device based on fast R-CNN network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106952250A true CN106952250A (en) | 2017-07-14 |
| CN106952250B CN106952250B (en) | 2021-05-07 |
Family
ID=59468055
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710110525.4A Active CN106952250B (en) | 2017-02-28 | 2017-02-28 | Metal plate strip surface defect detection method and device based on fast R-CNN network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106952250B (en) |
Cited By (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107424150A (en) * | 2017-07-27 | 2017-12-01 | 济南浪潮高新科技投资发展有限公司 | A kind of road damage testing method and device based on convolutional neural networks |
| CN107561738A (en) * | 2017-08-30 | 2018-01-09 | 湖南理工学院 | TFT LCD surface defect quick determination methods based on FCN |
| CN107607554A (en) * | 2017-09-26 | 2018-01-19 | 天津工业大学 | A kind of Defect Detection and sorting technique of the zinc-plated stamping parts based on full convolutional neural networks |
| CN107657603A (en) * | 2017-08-21 | 2018-02-02 | 北京精密机电控制设备研究所 | A kind of industrial appearance detecting method based on intelligent vision |
| CN107944505A (en) * | 2017-12-19 | 2018-04-20 | 青岛科技大学 | A kind of metal failure type automatization judgement method |
| CN108021926A (en) * | 2017-09-28 | 2018-05-11 | 东南大学 | A kind of vehicle scratch detection method and system based on panoramic looking-around system |
| CN108061735A (en) * | 2017-12-01 | 2018-05-22 | 工业互联网创新中心（上海）有限公司 | The recognition methods of component surface defect and device |
| CN108416294A (en) * | 2018-03-08 | 2018-08-17 | 南京天数信息科技有限公司 | A kind of fan blade fault intelligent identification method based on deep learning |
| CN108492291A (en) * | 2018-03-12 | 2018-09-04 | 苏州天准科技股份有限公司 | A kind of photovoltaic silicon chip Defect Detection system and method based on CNN segmentations |
| CN108520274A (en) * | 2018-03-27 | 2018-09-11 | 天津大学 | High reflecting surface defect inspection method based on image procossing and neural network classification |
| CN108562589A (en) * | 2018-03-30 | 2018-09-21 | 慧泉智能科技（苏州）有限公司 | A method of magnetic circuit material surface defect is detected |
| CN108564563A (en) * | 2018-03-07 | 2018-09-21 | 浙江大学 | A kind of tire X-ray defect detection method based on Faster R-CNN |
| CN108596883A (en) * | 2018-04-12 | 2018-09-28 | 福州大学 | It is a kind of that method for diagnosing faults is slid based on the Aerial Images stockbridge damper of deep learning and distance restraint |
| CN108830144A (en) * | 2018-05-03 | 2018-11-16 | 华南农业大学 | A kind of milking sow gesture recognition method based on improvement Faster-R-CNN |
| 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 |
| CN108986090A (en) * | 2018-07-11 | 2018-12-11 | 天津工业大学 | A kind of depth convolutional neural networks method applied to the detection of cabinet surface scratch |
| CN109115501A (en) * | 2018-07-12 | 2019-01-01 | 哈尔滨工业大学（威海） | A kind of Civil Aviation Engine Gas path fault diagnosis method based on CNN and SVM |
| CN109242830A (en) * | 2018-08-18 | 2019-01-18 | 苏州翔升人工智能科技有限公司 | A kind of machine vision technique detection method based on deep learning |
| CN109300114A (en) * | 2018-08-30 | 2019-02-01 | 西南交通大学 | The minimum target components of high iron catenary support device hold out against missing detection method |
| CN109291657A (en) * | 2018-09-11 | 2019-02-01 | 东华大学 | Laser Jet system is identified based on convolutional neural networks space structure part industry Internet of Things |
| CN109360204A (en) * | 2018-11-28 | 2019-02-19 | 燕山大学 | A kind of multiple layer metal lattice structure material internal defect detection method based on Faster R-CNN |
| CN109377483A (en) * | 2018-09-30 | 2019-02-22 | 云南电网有限责任公司普洱供电局 | Porcelain insulator crack detecting method and device |
| CN109472769A (en) * | 2018-09-26 | 2019-03-15 | 成都数之联科技有限公司 | A kind of bad image defect detection method and system |
| CN109544533A (en) * | 2018-11-23 | 2019-03-29 | 聚时科技(上海)有限公司 | A kind of metal plate defect detection and measure based on deep learning |
| CN109711474A (en) * | 2018-12-24 | 2019-05-03 | 中山大学 | A kind of aluminium material surface defects detection algorithm based on deep learning |
| CN109724984A (en) * | 2018-12-07 | 2019-05-07 | 上海交通大学 | A kind of defects detection identification device and method based on deep learning algorithm |
| CN109859207A (en) * | 2019-03-06 | 2019-06-07 | 华南理工大学 | A kind of defect inspection method of high density flexible substrate |
| CN109934088A (en) * | 2019-01-10 | 2019-06-25 | 海南大学 | Sea ship discrimination method based on deep learning |
| CN110021005A (en) * | 2018-01-05 | 2019-07-16 | 财团法人工业技术研究院 | Circuit board flaw screening technique and its device and computer-readable recording medium |
| CN110111293A (en) * | 2018-01-29 | 2019-08-09 | 国科赛思(北京)科技有限公司 | The failure recognition methods of plastic device and device |
| CN110120035A (en) * | 2019-04-17 | 2019-08-13 | 杭州数据点金科技有限公司 | A kind of tire X-ray defect detection method differentiating defect grade |
| CN110175982A (en) * | 2019-04-16 | 2019-08-27 | 浙江大学城市学院 | A kind of defect inspection method based on target detection |
| CN110276754A (en) * | 2019-06-21 | 2019-09-24 | 厦门大学 | A kind of detection method of surface flaw, terminal device and storage medium |
| CN110322430A (en) * | 2019-05-17 | 2019-10-11 | 杭州数据点金科技有限公司 | A kind of two stage tire X-ray defect detection method |
| CN110335238A (en) * | 2019-05-08 | 2019-10-15 | 吉林大学 | A kind of Automotive Paint Film Properties defect recognition system constituting method of deep learning |
| CN110346953A (en) * | 2019-07-02 | 2019-10-18 | 盐城华昱光电技术有限公司 | A kind of the sheet detection system and method for the removing of liquid crystal display die set polaroid |
| CN110570316A (en) * | 2018-08-31 | 2019-12-13 | 阿里巴巴集团控股有限公司 | method and device for training damage recognition model |
| 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 |
| 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 |
| CN110782183A (en) * | 2019-11-07 | 2020-02-11 | 山东浪潮人工智能研究院有限公司 | Aluminum product defect detection method based on Faster R-CNN |
| CN111079820A (en) * | 2019-12-12 | 2020-04-28 | 哈尔滨市科佳通用机电股份有限公司 | Image recognition-based rail wagon fire-proof plate fault recognition method |
| CN113245238A (en) * | 2021-05-13 | 2021-08-13 | 苏州迪宏人工智能科技有限公司 | Elbow welding flaw detection method, device and system |
| CN113379745A (en) * | 2021-08-13 | 2021-09-10 | 深圳市信润富联数字科技有限公司 | Product defect identification method and device, electronic equipment and storage medium |
| CN116580030A (en) * | 2023-07-13 | 2023-08-11 | 厦门微图软件科技有限公司 | Welding quality anomaly detection method based on anomaly simulation |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103886610A (en) * | 2014-04-05 | 2014-06-25 | 东北电力大学 | Image type defect detecting method for insulator |
| CN104424629A (en) * | 2013-08-19 | 2015-03-18 | 深圳先进技术研究院 | X-ray chest radiography lung segmentation method and device |
| CN104463104A (en) * | 2014-11-14 | 2015-03-25 | 武汉工程大学 | Fast detecting method and device for static vehicle target |
| CN104515786A (en) * | 2015-01-08 | 2015-04-15 | 北京科技大学 | Method for detecting and analyzing internal defect evolution of metal casting in fatigue process |
| WO2016168378A1 (en) * | 2015-04-13 | 2016-10-20 | Gerard Dirk Smits | Machine vision for ego-motion, segmenting, and classifying objects |
| CN106127173A (en) * | 2016-06-30 | 2016-11-16 | 北京小白世纪网络科技有限公司 | A kind of human body attribute recognition approach based on degree of depth study |
| CN106156744A (en) * | 2016-07-11 | 2016-11-23 | 西安电子科技大学 | SAR target detection method based on CFAR detection with degree of depth study |
-
2017
- 2017-02-28 CN CN201710110525.4A patent/CN106952250B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104424629A (en) * | 2013-08-19 | 2015-03-18 | 深圳先进技术研究院 | X-ray chest radiography lung segmentation method and device |
| CN103886610A (en) * | 2014-04-05 | 2014-06-25 | 东北电力大学 | Image type defect detecting method for insulator |
| CN104463104A (en) * | 2014-11-14 | 2015-03-25 | 武汉工程大学 | Fast detecting method and device for static vehicle target |
| CN104515786A (en) * | 2015-01-08 | 2015-04-15 | 北京科技大学 | Method for detecting and analyzing internal defect evolution of metal casting in fatigue process |
| WO2016168378A1 (en) * | 2015-04-13 | 2016-10-20 | Gerard Dirk Smits | Machine vision for ego-motion, segmenting, and classifying objects |
| CN106127173A (en) * | 2016-06-30 | 2016-11-16 | 北京小白世纪网络科技有限公司 | A kind of human body attribute recognition approach based on degree of depth study |
| CN106156744A (en) * | 2016-07-11 | 2016-11-23 | 西安电子科技大学 | SAR target detection method based on CFAR detection with degree of depth study |
Non-Patent Citations (2)
| Title |
|---|
| JOSEPH REDMON ET AL: "YOLO9000: Better, Faster, Stronger", 《EPRINT ARXIV:1612.08242》 * |
| LI YI ET AL: "An End‐to‐End Steel Strip Surface Defects Recognition System Based on Convolutional Neural Networks", 《STEEL RESEARCH INTERNATIONAL》 * |
Cited By (64)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107424150A (en) * | 2017-07-27 | 2017-12-01 | 济南浪潮高新科技投资发展有限公司 | A kind of road damage testing method and device based on convolutional neural networks |
| CN107657603B (en) * | 2017-08-21 | 2020-07-14 | 北京精密机电控制设备研究所 | Industrial appearance detection method based on intelligent vision |
| CN107657603A (en) * | 2017-08-21 | 2018-02-02 | 北京精密机电控制设备研究所 | A kind of industrial appearance detecting method based on intelligent vision |
| CN107561738A (en) * | 2017-08-30 | 2018-01-09 | 湖南理工学院 | TFT LCD surface defect quick determination methods based on FCN |
| CN107561738B (en) * | 2017-08-30 | 2020-06-12 | 湖南理工学院 | Fast TFT-LCD surface defect detection method based on FCN |
| CN107607554A (en) * | 2017-09-26 | 2018-01-19 | 天津工业大学 | A kind of Defect Detection and sorting technique of the zinc-plated stamping parts based on full convolutional neural networks |
| CN108021926A (en) * | 2017-09-28 | 2018-05-11 | 东南大学 | A kind of vehicle scratch detection method and system based on panoramic looking-around system |
| CN108061735A (en) * | 2017-12-01 | 2018-05-22 | 工业互联网创新中心（上海）有限公司 | The recognition methods of component surface defect and device |
| CN107944505A (en) * | 2017-12-19 | 2018-04-20 | 青岛科技大学 | A kind of metal failure type automatization judgement method |
| CN110021005B (en) * | 2018-01-05 | 2022-03-15 | 财团法人工业技术研究院 | Method and device for screening defects of circuit board and computer readable recording medium |
| CN110021005A (en) * | 2018-01-05 | 2019-07-16 | 财团法人工业技术研究院 | Circuit board flaw screening technique and its device and computer-readable recording medium |
| CN110111293B (en) * | 2018-01-29 | 2021-05-11 | 国科赛思(北京)科技有限公司 | Failure identification method and device for plastic package device |
| CN110111293A (en) * | 2018-01-29 | 2019-08-09 | 国科赛思(北京)科技有限公司 | The failure recognition methods of plastic device and device |
| CN108564563A (en) * | 2018-03-07 | 2018-09-21 | 浙江大学 | A kind of tire X-ray defect detection method based on Faster R-CNN |
| CN108416294A (en) * | 2018-03-08 | 2018-08-17 | 南京天数信息科技有限公司 | A kind of fan blade fault intelligent identification method based on deep learning |
| CN108492291B (en) * | 2018-03-12 | 2022-07-22 | 苏州天准科技股份有限公司 | CNN segmentation-based solar photovoltaic silicon wafer defect detection system and method |
| CN108492291A (en) * | 2018-03-12 | 2018-09-04 | 苏州天准科技股份有限公司 | A kind of photovoltaic silicon chip Defect Detection system and method based on CNN segmentations |
| CN108520274A (en) * | 2018-03-27 | 2018-09-11 | 天津大学 | High reflecting surface defect inspection method based on image procossing and neural network classification |
| CN108520274B (en) * | 2018-03-27 | 2022-03-11 | 天津大学 | High-reflectivity surface defect detection method based on image processing and neural network classification |
| CN108562589B (en) * | 2018-03-30 | 2020-12-01 | 慧泉智能科技（苏州）有限公司 | Method for detecting surface defects of magnetic circuit material |
| CN108562589A (en) * | 2018-03-30 | 2018-09-21 | 慧泉智能科技（苏州）有限公司 | A method of magnetic circuit material surface defect is detected |
| CN108596883A (en) * | 2018-04-12 | 2018-09-28 | 福州大学 | It is a kind of that method for diagnosing faults is slid based on the Aerial Images stockbridge damper of deep learning and distance restraint |
| CN108596883B (en) * | 2018-04-12 | 2021-07-13 | 福州大学 | Aerial image vibration damper slip fault diagnosis method based on deep learning and distance constraint |
| CN108830144B (en) * | 2018-05-03 | 2022-02-22 | 华南农业大学 | Lactating sow posture identification method based on improved Faster-R-CNN |
| CN108830144A (en) * | 2018-05-03 | 2018-11-16 | 华南农业大学 | A kind of milking sow gesture recognition method based on improvement Faster-R-CNN |
| 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 |
| 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 |
| 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 |
| 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 |
| 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 |
| CN108986090A (en) * | 2018-07-11 | 2018-12-11 | 天津工业大学 | A kind of depth convolutional neural networks method applied to the detection of cabinet surface scratch |
| CN109115501A (en) * | 2018-07-12 | 2019-01-01 | 哈尔滨工业大学（威海） | A kind of Civil Aviation Engine Gas path fault diagnosis method based on CNN and SVM |
| CN109242830A (en) * | 2018-08-18 | 2019-01-18 | 苏州翔升人工智能科技有限公司 | A kind of machine vision technique detection method based on deep learning |
| CN109300114A (en) * | 2018-08-30 | 2019-02-01 | 西南交通大学 | The minimum target components of high iron catenary support device hold out against missing detection method |
| CN110570316A (en) * | 2018-08-31 | 2019-12-13 | 阿里巴巴集团控股有限公司 | method and device for training damage recognition model |
| CN109291657A (en) * | 2018-09-11 | 2019-02-01 | 东华大学 | Laser Jet system is identified based on convolutional neural networks space structure part industry Internet of Things |
| CN109291657B (en) * | 2018-09-11 | 2020-10-30 | 东华大学 | Convolutional neural network-based aerospace structure industrial Internet of things identification laser coding system |
| CN109472769A (en) * | 2018-09-26 | 2019-03-15 | 成都数之联科技有限公司 | A kind of bad image defect detection method and system |
| CN109377483A (en) * | 2018-09-30 | 2019-02-22 | 云南电网有限责任公司普洱供电局 | Porcelain insulator crack detecting method and device |
| CN109544533A (en) * | 2018-11-23 | 2019-03-29 | 聚时科技(上海)有限公司 | A kind of metal plate defect detection and measure based on deep learning |
| CN109544533B (en) * | 2018-11-23 | 2021-04-02 | 聚时科技(上海)有限公司 | Metal plate defect detection and measurement method based on deep learning |
| CN109360204B (en) * | 2018-11-28 | 2021-07-16 | 燕山大学 | Inner defect detection method of multilayer metal lattice structure material based on Faster R-CNN |
| CN109360204A (en) * | 2018-11-28 | 2019-02-19 | 燕山大学 | A kind of multiple layer metal lattice structure material internal defect detection method based on Faster R-CNN |
| CN109724984B (en) * | 2018-12-07 | 2021-11-02 | 上海交通大学 | Defect detection and identification device and method based on deep learning algorithm |
| CN109724984A (en) * | 2018-12-07 | 2019-05-07 | 上海交通大学 | A kind of defects detection identification device and method based on deep learning algorithm |
| CN109711474B (en) * | 2018-12-24 | 2023-01-17 | 中山大学 | Aluminum product surface defect detection algorithm based on deep learning |
| CN109711474A (en) * | 2018-12-24 | 2019-05-03 | 中山大学 | A kind of aluminium material surface defects detection algorithm based on deep learning |
| CN109934088A (en) * | 2019-01-10 | 2019-06-25 | 海南大学 | Sea ship discrimination method based on deep learning |
| CN109859207A (en) * | 2019-03-06 | 2019-06-07 | 华南理工大学 | A kind of defect inspection method of high density flexible substrate |
| CN110175982B (en) * | 2019-04-16 | 2021-11-02 | 浙江大学城市学院 | Defect detection method based on target detection |
| CN110175982A (en) * | 2019-04-16 | 2019-08-27 | 浙江大学城市学院 | A kind of defect inspection method based on target detection |
| CN110120035A (en) * | 2019-04-17 | 2019-08-13 | 杭州数据点金科技有限公司 | A kind of tire X-ray defect detection method differentiating defect grade |
| CN110335238A (en) * | 2019-05-08 | 2019-10-15 | 吉林大学 | A kind of Automotive Paint Film Properties defect recognition system constituting method of deep learning |
| CN110322430A (en) * | 2019-05-17 | 2019-10-11 | 杭州数据点金科技有限公司 | A kind of two stage tire X-ray defect detection method |
| CN110276754A (en) * | 2019-06-21 | 2019-09-24 | 厦门大学 | A kind of detection method of surface flaw, terminal device and storage medium |
| CN110276754B (en) * | 2019-06-21 | 2021-08-20 | 厦门大学 | Surface defect detection method, terminal device and storage medium |
| CN110346953B (en) * | 2019-07-02 | 2022-08-09 | 晋城市龙鑫达光电科技有限公司 | Tearing detection system and method for stripping polaroid of liquid crystal display module |
| CN110346953A (en) * | 2019-07-02 | 2019-10-18 | 盐城华昱光电技术有限公司 | A kind of the sheet detection system and method for the removing of liquid crystal display die set polaroid |
| CN110782183A (en) * | 2019-11-07 | 2020-02-11 | 山东浪潮人工智能研究院有限公司 | Aluminum product defect detection method based on Faster R-CNN |
| CN111079820A (en) * | 2019-12-12 | 2020-04-28 | 哈尔滨市科佳通用机电股份有限公司 | Image recognition-based rail wagon fire-proof plate fault recognition method |
| CN113245238A (en) * | 2021-05-13 | 2021-08-13 | 苏州迪宏人工智能科技有限公司 | Elbow welding flaw detection method, device and system |
| CN113379745A (en) * | 2021-08-13 | 2021-09-10 | 深圳市信润富联数字科技有限公司 | Product defect identification method and device, electronic equipment and storage medium |
| CN116580030A (en) * | 2023-07-13 | 2023-08-11 | 厦门微图软件科技有限公司 | Welding quality anomaly detection method based on anomaly simulation |
| CN116580030B (en) * | 2023-07-13 | 2023-10-20 | 厦门微图软件科技有限公司 | Welding quality anomaly detection method based on anomaly simulation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106952250B (en) | 2021-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106952250A (en) | A kind of metal plate and belt detection method of surface flaw and device based on Faster R CNN networks | |
| CN111080622B (en) | Neural network training method, workpiece surface defect classification and detection method and device | |
| CN106934800A (en) | A kind of metal plate and belt detection method of surface flaw and device based on YOLO9000 networks | |
| CN109886298B (en) | Weld quality detection method based on convolutional neural network | |
| CN113362326B (en) | Method and device for detecting defects of welding spots of battery | |
| CN109239102B (en) | CNN-based flexible circuit board appearance defect detection method | |
| CN110473173A (en) | A kind of defect inspection method based on deep learning semantic segmentation | |
| CN107966454A (en) | A kind of end plug defect detecting device and detection method based on FPGA | |
| CN110243937A (en) | A kind of Analyse of Flip Chip Solder Joint missing defect intelligent detecting method based on high frequency ultrasound | |
| CN117250208A (en) | Machine vision-based nano-imprint wafer defect accurate detection system and method | |
| US11315229B2 (en) | Method for training defect detector | |
| CN113222913B (en) | Circuit board defect detection positioning method, device and storage medium | |
| CN115035092A (en) | Image-based bottle detection method, device, equipment and storage medium | |
| CN112200776A (en) | Chip packaging defect detection method and detection device | |
| TW201818090A (en) | Detection method for blind holes on printed circuit board | |
| CN108414531A (en) | A kind of fexible film defect detecting device and its detection method based on machine vision | |
| CN114240944A (en) | Welding defect detection method based on point cloud information | |
| CN108802051A (en) | A kind of flexibility IC substrate straight path bubbles and folding line defect detecting system and method | |
| CN113947583B (en) | Weld joint nondestructive testing method based on deep learning two-dimensional time sequence image | |
| JP3913517B2 (en) | Defect detection method | |
| CN201081763Y (en) | Textile defect detector | |
| CN115937555A (en) | Industrial defect detection algorithm based on standardized flow model | |
| CN109685008A (en) | A kind of real-time video object detection method | |
| US20240169510A1 (en) | Surface defect detection model training method, and surface defect detection method and system | |
| CN112862826B (en) | Normal sample nondestructive generation method for surface defect detection task |
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 |