CN110490874A - Weaving cloth surface flaw detecting method based on YOLO neural network - Google Patents
Weaving cloth surface flaw detecting method based on YOLO neural network Download PDFInfo
- Publication number
- CN110490874A CN110490874A CN201910832559.3A CN201910832559A CN110490874A CN 110490874 A CN110490874 A CN 110490874A CN 201910832559 A CN201910832559 A CN 201910832559A CN 110490874 A CN110490874 A CN 110490874A
- Authority
- CN
- China
- Prior art keywords
- flaw
- neural network
- training
- yolo
- picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- 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
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30124—Fabrics; Textile; Paper
-
- 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
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Pathology (AREA)
- Signal Processing (AREA)
- Immunology (AREA)
- Analytical Chemistry (AREA)
- Quality & Reliability (AREA)
- Biochemistry (AREA)
- Chemical & Material Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The present invention discloses a kind of weaving cloth surface flaw detecting method, comprising: the cloth exterior view piece of acquisition weaving in real time;Flaw identification is carried out to the weaving cloth exterior view piece that acquires in real time using the YOLO neural network detection model pre-established, obtain include flaw type and flaw location recognition result;And output flaw recognition result data.The building of YOLO neural network detection model includes flaw picture collection, processing, forming label and sample training process, using detecting and identify flaw based on the method for the supervised learning of big data, region where enabling established YOLO neural network detection model quickly to select flaw simultaneously detects corresponding type, the detection scheme of intelligent Fabric Defect can be provided for textile garment manufacturing enterprise, solve the problems such as artificial perching recall rate is low, speed is slow, personnel cost is high.
Description
Technical field
The present invention relates to textile surface Defect Detection technical fields, especially a kind of to be based on YOLO (You Only Look
Once only has a look at) the weaving cloth surface flaw detecting method of neural network.
Background technique
It stands at present in the cloth inspector that textile garment manufacturing enterprise mainly passes through profession and passes through naked eyes before perching equipment
It was found that fabric defect carries out the label of fault again.With continuing to optimize for Computer Vision Detection Technique, intelligent fabric inspecting system
The labor intensity of artificial detection will be mitigated significantly, improves the efficiency and precision of the monitoring of production process fabric quality.
With the continuous development of China's textile industryization level, the requirement to weaving cloth surface quality is also higher and higher,
How rapidly and accurately to detect that flaw becomes a link very crucial in cloth manufacturing process.The cloth of mainstream detects skill
Art is the machine vision technique based on image procossing, is divided into two steps substantially:
(1) first the image obtained by industrial camera is handled, then extracts feature, the method for this step mainly has statistics
Four class such as method, Structure Method, Spectrum Method, modelling, wherein using it is relatively broad be statistic law and Spectrum Method;
(2) by trained classifier, the feature extracted to (1) is classified, and has BP refreshing using more classifier at present
Through network, SVM etc..
But these methods are both limited by the low contrast and noise and fine defects between flaw and non-defect areas
Similitude exists and detects the problems such as speed is slow and accuracy of identification is low, is unable to satisfy the accuracy and real-time of caused by spinning industrial production
It is required that.
The continuous development of deep learning is that we provide better tools.Wherein, YOLO (You Only Look
Once only has a look at) be the target detection deep neural network based on convolutional network, because its on small target deteection table
The high frame rate and high-accuracy that reveal and be employed in many real-time detecting systems.
Summary of the invention
The technical problem to be solved in the present invention is to be examined automatically using the surface blemish of YOLO neural fusion weaving cloth
It surveys, promotes the accuracy and real-time of Defect Detection in textile industry cloth production process.
The technical scheme adopted by the invention is as follows: a kind of weaving cloth surface flaw detecting method, comprising:
The cloth exterior view piece of acquisition weaving in real time;
Flaw knowledge is carried out to the weaving cloth exterior view piece acquired in real time using the YOLO neural network detection model pre-established
, do not obtain include flaw type and flaw location recognition result;
Export flaw recognition result data.
Optionally, the method for building up of the YOLO neural network detection model includes:
Acquire multiple Fabric Defect pictures;
Fabric Defect picture is pre-processed;
The calibration that defect areas and type are carried out to pretreated Fabric Defect picture, obtains the flaw of each Fabric Defect picture
Label;
Each Fabric Defect picture and its flaw label are divided into training set sample and test set sample;
Using training set sample and test machine sample, using computer graphics processor GPU to the YOLO depth mind built in advance
It is trained through network, obtains YOLO neural network detection model.
When establishing YOLO neural network detection model, should shooting Fabric Defect picture as much as possible, cover as far as possible
Various known flaw types, the target detection model recall rate and accuracy rate for this training being obtained are all higher.
Optionally, the pretreatment includes that Fabric Defect picture is normalized: the cloth after equal proportion is scaled
In the blank picture of flaw picture filling presetted pixel size, the then pixel by blank sheet on piece in addition to Fabric Defect picture
It is stuffed entirely with as setpoint color.
Optionally, the pixel size of the blank picture is 416*416, and blank picture can cover the cloth flaw after scaling
Defect picture;The setpoint color is grey.
The above normalized processing of picture can keep the original textural characteristics of picture as far as possible, so that training obtained
YOLO model inspection accuracy is higher.
Optionally, the pretreatment further includes carrying out data enhancing to the Fabric Defect picture obtained after normalized.
The strategies such as existing turnover translation moving can be used in data enhancing.
Optionally, the present invention selects mode to carry out defect areas and kind to pretreated Fabric Defect picture by artificial frame
The flaw type of the upper left angle point in each calibration region and bottom right angular coordinate and corresponding region is stored as label by the calibration of class.
Optionally, the flaw less for elongated, pixel (such as wrong flower and latitude), using multiple continuous small frames into
Rower is fixed.Flaw proportion in rectangle frame can be improved.
Optionally, the training set sample and test machine sample random division, division proportion 9:1.
Optionally, described be trained to the YOLO neural network built in advance includes:
S1 is arranged training parameter: setting 10000 for iterative steps epochs, and learning rate optimizer optimizer is arranged
For ' adam', 64, and the size of 9 priori frame anchor box of setting are set by crowd number of training batch_size;
Training YOLO neural network: training set sample data and each sample corresponding flaw type data are input to YOLO by S2
Model training is carried out in neural network;
Model measurement: S3 every time after training, is tested, if model using the model that test set sample obtains training
To the flaw recall rate in test set sample be more than 95% and Detection accuracy is not less than 95%, then obtains last time training
Model as final YOLO neural network detection model;Otherwise model last time training obtained is as currently wait train
YOLO neural network, step S2-S3 is repeated, until obtain final YOLO neural network detection model.
Optionally, the size of the priori frame is arranged by k-means clustering method, comprising: randomly selects 9 flaw squares
Cluster centre of the shape frame object as training set distributes each object according to the distance between each object and each cluster centre
To apart from nearest cluster centre, one sample of every distribution, cluster centre is recalculated once, duplicate allocation process, until not having
There is cluster centre to change again, the size of the value i.e. 9 priori frame of current 9 cluster centres.Pass through k-means clustering method
The size for determining and being arranged 9 priori frame anchor box, can make model obtain more accurately target sizes in the training process
And location expression.
Beneficial effect
(1) present invention utilizes the target detection model of YOLO neural network Fabric Defect, using it on small target deteection
The high frame rate and high-accuracy that show ensure inspection, it can be achieved that for the real-time detection of surface blemish in cloth production process
Survey the accuracy of result;
(2) when establishing YOLO neural network detection model, the present invention is adopted by manually demarcating to defect areas and type
With the supervised learning detection algorithm based on big data, therefore it can be good at the study from data and improved to effectively expressing
The accuracy rate of detection;
(2) the target detection recognition methods based on deep learning that the present invention provides is by feature construction and fusion for classification at one
A entirety, i.e. input are initial data, direct output category result, do not need artificial constructed feature, are more suitable for solving complicated
The automatic detection identification problem of Fabric Defect under grain background;
(3) classification method of the present invention uses YOLO deep neural network frame, which will test task as regression problem
It handles, directly obtains bounding box coordinates and confidence level comprising object by all pixels of entire image and classification is general
Rate is significantly better than R-CNN, Fast R-CNN even depth learning network frame in detection speed.And it is clustered using k-means
Algorithm determines the size of priori frame anchor box, and YOLO model is made to obtain more accurately target sizes and position in the training process
Description is set, Defect Detection accuracy rate is improved.
Detailed description of the invention
Fig. 1 show model foundation of the present invention to a kind of specific embodiment method flow schematic diagram of real-time detection process;
Fig. 2 is picture normalization processes result figure provided by the present invention;
Fig. 3 is two different picture flaw mark modes provided by the present invention;
Fig. 4 is YOLO deep neural network structural schematic diagram provided by the present invention;
Fig. 5 is the operation principle schematic diagram of YOLO network provided by the present invention;
Fig. 6 is detection effect illustrated example of the present invention.
Specific embodiment
It is further described below in conjunction with the drawings and specific embodiments.
With reference to Fig. 1, a kind of weaving cloth surface flaw detecting method of the present invention, comprising:
The cloth exterior view piece of acquisition weaving in real time;
Flaw knowledge is carried out to the weaving cloth exterior view piece acquired in real time using the YOLO neural network detection model pre-established
, do not obtain include flaw type and flaw location recognition result;
Export flaw recognition result data.
Wherein, the method for building up of the YOLO neural network detection model includes:
Acquire multiple Fabric Defect pictures;
Fabric Defect picture is pre-processed;
The calibration that defect areas and type are carried out to pretreated Fabric Defect picture, obtains the flaw of each Fabric Defect picture
Label;
Each Fabric Defect picture and its flaw label are divided into training set sample and test set sample;
Using training set sample and test machine sample, using computer graphics processor GPU to the YOLO depth mind built in advance
It is trained through network, obtains YOLO neural network detection model.
Embodiment
Embodiment as shown in Figure 1, when establishing YOLO neural network detection model, data acquisition phase should be utilized and be taken the photograph
The shooting Fabric Defect picture as much as possible such as camera, to cover various known flaw types, so that the target inspection that training obtains
It surveys model recall rate and accuracy rate is all higher.
Flaw picture is normalized in data preprocessing phase, which is normalization process by picture
Original length-width ratio zooms to correspondingly sized, is filled among 416*416 blank picture, other vacant pixels are all filled out as grey,
To keep the original textural characteristics of picture as far as possible, as shown in Figure 2;And by overturning, translation strategy to the flaw figure after normalization
Piece carries out data enhancing.
In picture calibration phase, the production of flaw label comprising steps of
(1) marker software can be used, defect areas and flaw type are manually demarcated with rectangle frame, wherein for elongated, pixel
Such as wrong flower of less flaw and latitude, are demarcated with multiple continuous small frames, so as to improve flaw in rectangle frame institute's accounting
Example, as shown in Figure 3;
(2) transverse and longitudinal coordinate of the upper left angle point of rectangle frame and bottom right angle point is stored in txt file, while the rectangle frame is wrapped
The flaw type contained is also recorded in file.
Existing graphic plotting modification software, such as labelImg software can be used in the marker software, only for convenience of in people
The choosing of work frame can automatically or artificially obtain corresponding coordinate data.
Training set sample and test machine sample random division, division proportion are about 9:1, can be floated up and down.
The network structure for the YOLO neural network that the present embodiment is built in advance can refer to shown in Fig. 4, to YOLO neural network
It is trained and includes:
S1 is arranged training parameter: setting 10000 for iterative steps epochs, and learning rate optimizer optimizer is arranged
For ' adam', i.e., the learning rate of each network node parameter is dynamically adjusted, sets 64 for crowd number of training batch_size,
Determining by k-means clustering method and 9 priori frame anchor box of setting size;
Training YOLO neural network: training set sample data and each sample corresponding flaw type data are input to YOLO by S2
In neural network, according to the parameter of setting, with computer graphics processor GPU training objective detection model;;
Model measurement: S3 every time after training, is tested, if model using the model that test set sample obtains training
To the flaw recall rate in test set sample be more than 95% and Detection accuracy is not less than 95%, then obtains last time training
Model as final YOLO neural network detection model;Otherwise model last time training obtained is as currently wait train
YOLO neural network, step S2-S3 is repeated, until obtain final YOLO neural network detection model.
The subsequent real-time detection that Fabric Defect is carried out using final YOLO neural network detection model: camera is utilized
Weaving cloth surface image to be detected is acquired in real time, and according to trained YOLO network, whether real-time detection cloth surface has
Flaw, and the flaw detected is classified and positioned.
Realize k-means clustering algorithm the specific steps are randomly select 9 flaw rectangle frame objects as training set
Then cluster centre calculates the distance between each object and each cluster centre, each object is distributed to apart from it recently
Cluster centre, one sample of every distribution, cluster centre can be recalculated according to object existing in cluster, repeat above-mentioned point
With process, until not having cluster centre to change again, the value of 9 cluster centres obtains the size of 9 priori frames.
Weaving cloth surface flaw detecting method based on YOLO network in order to better understand, here to YOLO network
Working principle is briefly described, as shown in Figure 5:
The image averaging of input is divided into S × S cell (as shown in Fig. 5 (a)) by a.YOLO network;
B. each cell can predict B bounding box (Bounding Box), provide these bounding boxes in vector form
Information.The information of bounding box includes location information (rectangle frame center point coordinate, wide and high), confidence level (Confidence) with
And the classification information of prediction object (shown in such as Fig. 5 (b) (c)).
C. for training data, after picture and label input, by five parameters of cell output, (rectangle frame central point is sat
Mark, wide and high, confidence level) substituting into loss function, (loss function is used to calculate five parameters and mark that YOLO network query function goes out
Five parameters between gap) calculate, weight is adjusted by backpropagation, increases the confidence level of correct flaw type,
The confidence level of incorrect flaw type reduces;Data for acquiring in real time can also calculate five ginsengs after inputting YOLO network
Number (rectangle frame center point coordinate, wide and high, confidence level) can obtain making to damage after calculating this five data by loss function
The minimum bounding box of functional value, that is, the classfying frame finally needed, i.e. testing result are lost, (shown in such as Fig. 5 (d)).
In example shown in Fig. 6, it is shown that after detecting the cloth picture of four high-definition camera shootings as a result, figure
In flaw is outlined automatically, while the upper left corner gives the type of flaw.
To sum up, the present invention realizes the real-time detection of Fabric Defect using YOLO deep learning method, can quickly frame select
Region where flaw simultaneously detects corresponding type, and flaw recall rate is up to 95%, and Detection accuracy is up to 90%.Because having used depth
Study has robustness high in this way, and accuracy rate is high, and fireballing advantage, for textile garment, manufacturing enterprise provides intelligence
The detection scheme of Fabric Defect solves the problems such as artificial perching recall rate is low, speed is slow, personnel cost is high.
It should be understood by those skilled in the art that, embodiments herein can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the application
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the application, which can be used in one or more,
The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) produces
The form of product.
The application is referring to method, the process of equipment (system) and computer program product according to the embodiment of the present application
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
The embodiment of the present invention is described in conjunction with attached drawing above, but the invention is not limited to above-mentioned specific
Embodiment, the above mentioned embodiment is only schematical, rather than restrictive, those skilled in the art
Under the inspiration of the present invention, without breaking away from the scope protected by the purposes and claims of the present invention, it can also make very much
Form, all of these belong to the protection of the present invention.
Claims (10)
1. a kind of weaving cloth surface flaw detecting method, characterized in that include:
The cloth exterior view piece of acquisition weaving in real time;
Flaw knowledge is carried out to the weaving cloth exterior view piece acquired in real time using the YOLO neural network detection model pre-established
, do not obtain include flaw type and flaw location recognition result;
Export flaw recognition result data.
2. according to the method described in claim 1, it is characterized in that, the method for building up packet of the YOLO neural network detection model
It includes:
Acquire multiple Fabric Defect pictures;
Fabric Defect picture is pre-processed;
The calibration that defect areas and type are carried out to pretreated Fabric Defect picture, obtains the flaw of each Fabric Defect picture
Label;
Each Fabric Defect picture and its flaw label are divided into training set sample and test set sample;
Using training set sample and test machine sample, using computer graphics processor GPU to the YOLO depth mind built in advance
It is trained through network, obtains YOLO neural network detection model.
3. according to the method described in claim 2, it is characterized in that, the pretreatment includes that Fabric Defect picture is normalized
Processing: in the blank picture of the Fabric Defect picture filling presetted pixel size after equal proportion is scaled, then by blank picture
On pixel in addition to Fabric Defect picture be stuffed entirely with as setpoint color.
4. according to the method described in claim 3, it is characterized in that, the pixel size of the blank picture is 416*416, blank sheet
Piece can cover the Fabric Defect picture after scaling;The setpoint color is grey.
5. according to the method described in claim 2, it is characterized in that, the pretreatment further includes to the cloth obtained after normalized
Flaw picture carries out data enhancing.
6. according to the method described in claim 2, it is characterized in that, select mode to pretreated Fabric Defect figure by artificial frame
Piece carries out the calibration of defect areas and type, by the flaw of the upper left angle point in each calibration region and bottom right angular coordinate and corresponding region
Defect type is stored as flaw label.
7. according to the method described in claim 6, it is characterized in that, the flaw less for elongated, pixel is (such as wrong flower and latitude
Deng), it is demarcated using multiple continuous small frames.
8. according to the method described in claim 2, it is characterized in that, the training set sample and test machine sample random division, draw
Dividing ratio is 9:1.
9. according to the method described in claim 2, it is characterized in that, it is described that packet is trained to the YOLO neural network built in advance
It includes:
S1 is arranged training parameter: setting 10000 for iterative steps epochs, and learning rate optimizer optimizer is arranged
For ' adam', 64, and the size of 9 priori frame anchor box of setting are set by crowd number of training batch_size;
Training YOLO neural network: training set sample data and each sample corresponding flaw type data are input to YOLO by S2
Model training is carried out in neural network;
Model measurement: S3 every time after training, is tested, if model using the model that test set sample obtains training
To the flaw recall rate in test set sample be more than 95% and Detection accuracy is not less than 95%, then obtains last time training
Model as final YOLO neural network detection model;Otherwise model last time training obtained is as currently wait train
YOLO neural network, step S2-S3 is repeated, until obtain final YOLO neural network detection model.
10. according to the method described in claim 9, it is characterized in that, the size of the priori frame is set by k-means clustering method
It sets, comprising: cluster centre of 9 flaw rectangle frame objects as training set is randomly selected, according in each object and each cluster
The distance between heart distributes to each object apart from nearest cluster centre, one sample of every distribution, and cluster centre recalculates
Once, duplicate allocation process, until there is no cluster centre to change again, the value i.e. 9 priori frame of current 9 cluster centres
Size.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910832559.3A CN110490874A (en) | 2019-09-04 | 2019-09-04 | Weaving cloth surface flaw detecting method based on YOLO neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910832559.3A CN110490874A (en) | 2019-09-04 | 2019-09-04 | Weaving cloth surface flaw detecting method based on YOLO neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110490874A true CN110490874A (en) | 2019-11-22 |
Family
ID=68556468
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910832559.3A Pending CN110490874A (en) | 2019-09-04 | 2019-09-04 | Weaving cloth surface flaw detecting method based on YOLO neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110490874A (en) |
Cited By (36)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110827277A (en) * | 2019-11-26 | 2020-02-21 | 山东浪潮人工智能研究院有限公司 | Cloth flaw detection method based on yolo3 network |
CN111062925A (en) * | 2019-12-18 | 2020-04-24 | 华南理工大学 | Intelligent cloth defect identification method based on deep learning |
CN111126404A (en) * | 2019-12-11 | 2020-05-08 | 杭州电子科技大学 | Ancient character and font identification method based on improved YOLO v3 |
CN111260614A (en) * | 2020-01-13 | 2020-06-09 | 华南理工大学 | Convolutional neural network cloth flaw detection method based on extreme learning machine |
CN111340802A (en) * | 2020-03-25 | 2020-06-26 | 嘉兴学院 | Intelligent cloth inspecting machine capable of automatically inspecting cloth by adopting artificial intelligence technology |
CN111462051A (en) * | 2020-03-14 | 2020-07-28 | 华中科技大学 | Cloth defect detection method and system based on deep neural network |
CN111553898A (en) * | 2020-04-27 | 2020-08-18 | 东华大学 | Fabric defect detection method based on convolutional neural network |
CN111598833A (en) * | 2020-04-01 | 2020-08-28 | 江汉大学 | Method and device for detecting defects of target sample and electronic equipment |
CN111681229A (en) * | 2020-06-10 | 2020-09-18 | 创新奇智(上海)科技有限公司 | Deep learning model training method, wearable clothes flaw identification method and wearable clothes flaw identification device |
CN111767858A (en) * | 2020-06-30 | 2020-10-13 | 北京百度网讯科技有限公司 | Image recognition method, device, equipment and computer storage medium |
CN111881774A (en) * | 2020-07-07 | 2020-11-03 | 上海艾豚科技有限公司 | Method and system for identifying foreign matters in textile raw materials |
CN111929327A (en) * | 2020-09-09 | 2020-11-13 | 深兰人工智能芯片研究院(江苏)有限公司 | Cloth defect detection method and device |
CN112085024A (en) * | 2020-09-21 | 2020-12-15 | 江苏理工学院 | Tank surface character recognition method |
CN112115287A (en) * | 2020-08-17 | 2020-12-22 | 联通(广东)产业互联网有限公司 | Method, system and medium based on AI quality inspection and industrial big data analysis |
CN112215824A (en) * | 2020-10-16 | 2021-01-12 | 南通大学 | YOLO-v 3-based cloth cover defect detection and auxiliary device and method |
CN112348776A (en) * | 2020-10-16 | 2021-02-09 | 上海布眼人工智能科技有限公司 | Fabric flaw detection method based on EfficientDet |
CN112785574A (en) * | 2021-01-25 | 2021-05-11 | 金陵科技学院 | Scarf pattern defect detection method based on improved YOLOv3 |
CN112907529A (en) * | 2021-02-09 | 2021-06-04 | 南京航空航天大学 | Image-based woven preform defect detection method and device |
CN112967223A (en) * | 2021-01-29 | 2021-06-15 | 绍兴隆芙力智能科技发展有限公司 | Artificial intelligence-based textile detection system, method and medium |
CN113139946A (en) * | 2021-04-26 | 2021-07-20 | 北方工业大学 | Shirt stain positioning device based on vision |
CN113192040A (en) * | 2021-05-10 | 2021-07-30 | 浙江理工大学 | Fabric flaw detection method based on YOLO v4 improved algorithm |
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 |
CN113267501A (en) * | 2020-02-17 | 2021-08-17 | 远传电信股份有限公司 | Sheet material rapid defect detection integration system and use method thereof |
CN113506242A (en) * | 2021-05-26 | 2021-10-15 | 杭州电子科技大学 | Corn aflatoxin detection method based on YOLO |
CN113567443A (en) * | 2021-06-22 | 2021-10-29 | 浙江大豪科技有限公司 | Control method and device of fabric flaw detection system |
CN113627435A (en) * | 2021-07-13 | 2021-11-09 | 南京大学 | Method and system for detecting and identifying flaws of ceramic tiles |
CN114240822A (en) * | 2021-12-28 | 2022-03-25 | 中山小池科技有限公司 | Cotton cloth flaw detection method based on YOLOv3 and multi-scale feature fusion |
CN114240885A (en) * | 2021-12-17 | 2022-03-25 | 成都信息工程大学 | Cloth flaw detection method based on improved Yolov4 network |
CN114638779A (en) * | 2021-07-29 | 2022-06-17 | 广州机智云物联网科技有限公司 | Textile quality inspection system, method and device, computer equipment and storage medium |
CN115294136A (en) * | 2022-10-09 | 2022-11-04 | 播种者工业智能科技(苏州)有限公司 | Artificial intelligence-based construction and detection method for textile fabric flaw detection model |
CN115980322A (en) * | 2023-02-09 | 2023-04-18 | 百福工业缝纫机(张家港)有限公司 | Intelligent detection method and system for fabric defects |
CN116106319A (en) * | 2023-02-14 | 2023-05-12 | 浙江迈沐智能科技有限公司 | Automatic detection method and system for defects of synthetic leather |
CN116309493A (en) * | 2023-03-24 | 2023-06-23 | 南通飞隼信息科技有限公司 | Method and system for detecting defects of textile products |
CN116593486A (en) * | 2023-07-18 | 2023-08-15 | 佛山市南海德耀纺织实业有限公司 | Intelligent detection method for cloth flaws and related equipment thereof |
CN118052825A (en) * | 2024-04-16 | 2024-05-17 | 浙江泰伦绝缘子有限公司 | Glass insulator surface flaw detection method |
CN113506242B (en) * | 2021-05-26 | 2024-07-12 | 杭州电子科技大学 | Corn aflatoxin detection method based on YOLO |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108647742A (en) * | 2018-05-19 | 2018-10-12 | 南京理工大学 | Fast target detection method based on lightweight neural network |
CN109615610A (en) * | 2018-11-13 | 2019-04-12 | 浙江师范大学 | A kind of medical band-aid flaw detection method based on YOLO v2-tiny |
CN110136126A (en) * | 2019-05-17 | 2019-08-16 | 东南大学 | Cloth textured flaw detection method based on full convolutional neural networks |
-
2019
- 2019-09-04 CN CN201910832559.3A patent/CN110490874A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108647742A (en) * | 2018-05-19 | 2018-10-12 | 南京理工大学 | Fast target detection method based on lightweight neural network |
CN109615610A (en) * | 2018-11-13 | 2019-04-12 | 浙江师范大学 | A kind of medical band-aid flaw detection method based on YOLO v2-tiny |
CN110136126A (en) * | 2019-05-17 | 2019-08-16 | 东南大学 | Cloth textured flaw detection method based on full convolutional neural networks |
Non-Patent Citations (3)
Title |
---|
HONG-WEI ZHANG 等: "Yarn-dyed Fabric Defect Detection with YOLOV2 Based on Deep Convolution Neural Networks", 《IEEE》 * |
JOSEPH REDMON 等: "YOLOv3: An Incremental Improvement", 《ARXIV》 * |
禹万泓: "基于YOLOv2的纺织布疵点检测算法及系统研究", 《万方数据》 * |
Cited By (45)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110827277A (en) * | 2019-11-26 | 2020-02-21 | 山东浪潮人工智能研究院有限公司 | Cloth flaw detection method based on yolo3 network |
CN111126404A (en) * | 2019-12-11 | 2020-05-08 | 杭州电子科技大学 | Ancient character and font identification method based on improved YOLO v3 |
CN111126404B (en) * | 2019-12-11 | 2023-08-22 | 杭州电子科技大学 | Ancient character and font recognition method based on improved YOLO v3 |
CN111062925A (en) * | 2019-12-18 | 2020-04-24 | 华南理工大学 | Intelligent cloth defect identification method based on deep learning |
CN111260614A (en) * | 2020-01-13 | 2020-06-09 | 华南理工大学 | Convolutional neural network cloth flaw detection method based on extreme learning machine |
CN113267501A (en) * | 2020-02-17 | 2021-08-17 | 远传电信股份有限公司 | Sheet material rapid defect detection integration system and use method thereof |
CN111462051A (en) * | 2020-03-14 | 2020-07-28 | 华中科技大学 | Cloth defect detection method and system based on deep neural network |
CN111462051B (en) * | 2020-03-14 | 2022-09-27 | 华中科技大学 | Cloth defect detection method and system based on deep neural network |
CN111340802A (en) * | 2020-03-25 | 2020-06-26 | 嘉兴学院 | Intelligent cloth inspecting machine capable of automatically inspecting cloth by adopting artificial intelligence technology |
CN111598833A (en) * | 2020-04-01 | 2020-08-28 | 江汉大学 | Method and device for detecting defects of target sample and electronic equipment |
CN111553898A (en) * | 2020-04-27 | 2020-08-18 | 东华大学 | Fabric defect detection method based on convolutional neural network |
CN111681229A (en) * | 2020-06-10 | 2020-09-18 | 创新奇智(上海)科技有限公司 | Deep learning model training method, wearable clothes flaw identification method and wearable clothes flaw identification device |
CN111767858B (en) * | 2020-06-30 | 2024-03-22 | 北京百度网讯科技有限公司 | Image recognition method, device, equipment and computer storage medium |
CN111767858A (en) * | 2020-06-30 | 2020-10-13 | 北京百度网讯科技有限公司 | Image recognition method, device, equipment and computer storage medium |
CN111881774A (en) * | 2020-07-07 | 2020-11-03 | 上海艾豚科技有限公司 | Method and system for identifying foreign matters in textile raw materials |
CN112115287A (en) * | 2020-08-17 | 2020-12-22 | 联通(广东)产业互联网有限公司 | Method, system and medium based on AI quality inspection and industrial big data analysis |
CN111929327A (en) * | 2020-09-09 | 2020-11-13 | 深兰人工智能芯片研究院(江苏)有限公司 | Cloth defect detection method and device |
CN112085024A (en) * | 2020-09-21 | 2020-12-15 | 江苏理工学院 | Tank surface character recognition method |
CN112348776A (en) * | 2020-10-16 | 2021-02-09 | 上海布眼人工智能科技有限公司 | Fabric flaw detection method based on EfficientDet |
CN112215824A (en) * | 2020-10-16 | 2021-01-12 | 南通大学 | YOLO-v 3-based cloth cover defect detection and auxiliary device and method |
CN112785574B (en) * | 2021-01-25 | 2023-06-06 | 金陵科技学院 | Scarf pattern defect detection method based on improved YOLOv3 |
CN112785574A (en) * | 2021-01-25 | 2021-05-11 | 金陵科技学院 | Scarf pattern defect detection method based on improved YOLOv3 |
CN112967223A (en) * | 2021-01-29 | 2021-06-15 | 绍兴隆芙力智能科技发展有限公司 | Artificial intelligence-based textile detection system, method and medium |
CN112907529A (en) * | 2021-02-09 | 2021-06-04 | 南京航空航天大学 | Image-based woven preform defect detection method and device |
CN113139946A (en) * | 2021-04-26 | 2021-07-20 | 北方工业大学 | Shirt stain positioning device based on vision |
CN113192040A (en) * | 2021-05-10 | 2021-07-30 | 浙江理工大学 | Fabric flaw detection method based on YOLO v4 improved algorithm |
CN113192040B (en) * | 2021-05-10 | 2023-09-22 | 浙江理工大学 | Fabric flaw detection method based on YOLO v4 improved algorithm |
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 |
CN113506242B (en) * | 2021-05-26 | 2024-07-12 | 杭州电子科技大学 | Corn aflatoxin detection method based on YOLO |
CN113506242A (en) * | 2021-05-26 | 2021-10-15 | 杭州电子科技大学 | Corn aflatoxin detection method based on YOLO |
CN113567443A (en) * | 2021-06-22 | 2021-10-29 | 浙江大豪科技有限公司 | Control method and device of fabric flaw detection system |
CN113627435A (en) * | 2021-07-13 | 2021-11-09 | 南京大学 | Method and system for detecting and identifying flaws of ceramic tiles |
CN114638779A (en) * | 2021-07-29 | 2022-06-17 | 广州机智云物联网科技有限公司 | Textile quality inspection system, method and device, computer equipment and storage medium |
CN114638779B (en) * | 2021-07-29 | 2023-09-29 | 广州机智云物联网科技有限公司 | Textile quality inspection system, method, device, computer equipment and storage medium |
CN114240885B (en) * | 2021-12-17 | 2022-08-16 | 成都信息工程大学 | Cloth flaw detection method based on improved Yolov4 network |
CN114240885A (en) * | 2021-12-17 | 2022-03-25 | 成都信息工程大学 | Cloth flaw detection method based on improved Yolov4 network |
CN114240822A (en) * | 2021-12-28 | 2022-03-25 | 中山小池科技有限公司 | Cotton cloth flaw detection method based on YOLOv3 and multi-scale feature fusion |
CN115294136A (en) * | 2022-10-09 | 2022-11-04 | 播种者工业智能科技(苏州)有限公司 | Artificial intelligence-based construction and detection method for textile fabric flaw detection model |
CN115980322A (en) * | 2023-02-09 | 2023-04-18 | 百福工业缝纫机(张家港)有限公司 | Intelligent detection method and system for fabric defects |
CN116106319A (en) * | 2023-02-14 | 2023-05-12 | 浙江迈沐智能科技有限公司 | Automatic detection method and system for defects of synthetic leather |
CN116309493A (en) * | 2023-03-24 | 2023-06-23 | 南通飞隼信息科技有限公司 | Method and system for detecting defects of textile products |
CN116593486A (en) * | 2023-07-18 | 2023-08-15 | 佛山市南海德耀纺织实业有限公司 | Intelligent detection method for cloth flaws and related equipment thereof |
CN116593486B (en) * | 2023-07-18 | 2023-12-08 | 佛山市南海德耀纺织实业有限公司 | Intelligent detection method for cloth flaws and related equipment thereof |
CN118052825A (en) * | 2024-04-16 | 2024-05-17 | 浙江泰伦绝缘子有限公司 | Glass insulator surface flaw detection method |
CN118052825B (en) * | 2024-04-16 | 2024-06-21 | 浙江泰伦绝缘子有限公司 | Glass insulator surface flaw detection method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110490874A (en) | Weaving cloth surface flaw detecting method based on YOLO neural network | |
CN113192040B (en) | Fabric flaw detection method based on YOLO v4 improved algorithm | |
CN102221559B (en) | Online automatic detection method of fabric defects based on machine vision and device thereof | |
CN106952258B (en) | A kind of bottle mouth defect detection method based on gradient orientation histogram | |
CN110378909B (en) | Single wood segmentation method for laser point cloud based on Faster R-CNN | |
CN104268505B (en) | Fabric Defects Inspection automatic detecting identifier and method based on machine vision | |
CN109636772A (en) | The defect inspection method on the irregular shape intermetallic composite coating surface based on deep learning | |
CN109711288A (en) | Remote sensing ship detecting method based on feature pyramid and distance restraint FCN | |
CN111982910B (en) | Weak supervision machine vision detection method and system based on artificial defect simulation | |
CN111507976B (en) | Defect detection method and system based on multi-angle imaging | |
CN110400322A (en) | Fruit point cloud segmentation method based on color and three-dimensional geometric information | |
CN110175988A (en) | Cloth defect inspection method based on deep learning | |
CN110232379A (en) | A kind of vehicle attitude detection method and system | |
CN102073846A (en) | Method for acquiring traffic information based on aerial images | |
CN102621154B (en) | Method and device for automatically detecting cloth defects on line based on improved differential box multi-fractal algorithm | |
CN111047655A (en) | High-definition camera cloth defect detection method based on convolutional neural network | |
CN109509171A (en) | A kind of Fabric Defects Inspection detection method based on GMM and image pyramid | |
CN106780483A (en) | Many continuous casting billet end face visual identifying systems and centre coordinate acquiring method | |
CN104331693B (en) | A kind of printed matter symmetry detection methods and system | |
CN107154042A (en) | A kind of seed-coating machine visible detection method and device | |
CN109584206B (en) | Method for synthesizing training sample of neural network in part surface flaw detection | |
CN109540925A (en) | Complicated ceramic tile surface defect inspection method based on difference shadow method and local variance measurement operator | |
CN109087305A (en) | A kind of crack image partition method based on depth convolutional neural networks | |
CN110263794A (en) | Safety belt images steganalysis method based on data enhancing | |
CN105956592B (en) | A kind of Aircraft Targets detection method based on saliency and SVM |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191122 |
|
RJ01 | Rejection of invention patent application after publication |