CN106875373A - Mobile phone screen MURA defect inspection methods based on convolutional neural networks pruning algorithms - Google Patents
Mobile phone screen MURA defect inspection methods based on convolutional neural networks pruning algorithms Download PDFInfo
- Publication number
- CN106875373A CN106875373A CN201611154333.5A CN201611154333A CN106875373A CN 106875373 A CN106875373 A CN 106875373A CN 201611154333 A CN201611154333 A CN 201611154333A CN 106875373 A CN106875373 A CN 106875373A
- Authority
- CN
- China
- Prior art keywords
- picture
- neural networks
- convolutional neural
- mobile phone
- phone screen
- 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
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000007547 defect Effects 0.000 title claims abstract description 28
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 27
- 238000013138 pruning Methods 0.000 title claims abstract description 24
- 238000007689 inspection Methods 0.000 title claims abstract description 8
- 230000004044 response Effects 0.000 claims abstract description 13
- 238000010586 diagram Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 9
- 230000003044 adaptive effect Effects 0.000 claims abstract description 8
- 230000011218 segmentation Effects 0.000 claims abstract description 5
- 238000001514 detection method Methods 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 4
- 230000000452 restraining effect Effects 0.000 claims description 2
- 230000000717 retained effect Effects 0.000 claims description 2
- 241000208340 Araliaceae Species 0.000 claims 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 1
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 230000007935 neutral effect Effects 0.000 abstract 1
- 238000013135 deep learning Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000009776 industrial production Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 210000000653 nervous system Anatomy 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000007634 remodeling Methods 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- 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
-
- 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
- G06T2207/30148—Semiconductor; IC; Wafer
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
Mobile phone screen MURA defect inspection methods based on convolutional neural networks pruning algorithms, including：1) self-defined depth convolutional neural networks, a neutral net for being used for detecting mobile phone screen MURA defects is trained using existing training data；2) cut operator of convolutional neural networks is carried out using the method for adaptive template matching, reduces network size, shorten Riming time of algorithm；3) the mobile phone screen picture that high resolution camera shoots is carried out the scaling of different proportion, form picture pyramid, for the picture of each yardstick, using the method for sliding window by picture segmentation into fritter, sent into all fritter pictures as a group together in convolutional neural networks；4) all characteristic patterns in intermediate layer are chosen as the response diagram of defect, mobile phone screen MURA defect areas position is finally obtained using the method for Threshold segmentation.
Description
Technical field
The invention belongs to object detection and recognition field, it is related to detect specific objective from image, specifically detects mobile phone
The method of screen defect.
Background technology
There is many weak points in traditional manually detection screen flaw method, in the present of industrial production high speed development
My god, it cannot adapt to current industrial production and efficiently, accurately require completely.For mobile phone screen business men, find a kind of
Efficiently, accurate automatic detection system is used to substitute manual detection link, becomes urgent demand.As computer is regarded
The development in the fields such as feel, image procossing, the automated detection system based on machine vision becomes a kind of good solution.
The program gathers mobile phone screen image by high resolution industrial camera, and then image information is carried out by image analysis module
Treatment in real time, so as to judge whether mobile phone screen is qualified.
Traditional screen defects detection algorithm based on machine vision, the screen that one or more classifications are directed to mostly lacks
Fall into what is be designed, without versatility, so for special flaw, it is necessary to write special algorithm.It is special due to screen
Property, moire fringes when taking pictures in imaging are inevitable, and traditional algorithm can not well solve the problems, such as moire fringes.Separately
Outward, although traditional algorithm can detect obvious wire, spot defect, lack for bulk MURA very light in imaging
Fall into, accuracy rate is very low.Finally, traditional screen defects detection algorithm needs to adjust quantity of parameters, is especially changed in screen product
During type, adjustment quantity of parameters can cause waste of time.Therefore, designing one has the algorithm of good generalization with very real
Value.
In recent years, deep learning method generates tremendous influence in computer vision field.Deep learning uses multilayer
Network structure, the hierarchical relationship and transfer mode of nervous system in simulation human brain, it is obtained in the multiple fields of pattern-recognition
To being widely applied and achieve good achievement.This method is made using the sorting algorithm based on depth convolutional neural networks
With pretreated mobile phone screen topography block as the input of grader, the characteristic pattern work of convolutional neural networks is then extracted
It is testing result, defects detection problem is transformed into an image block classification problem, the depth model obtained by the method is not
Only can effectively learn the background texture pattern to image, position defect exactly from the image block containing background texture
Position, and have accuracy rate very high for bulk MURA defects.Additionally, being calculated compared to tradition using convolutional neural networks algorithm
Method, parameter setting is less, because algorithm has good versatility, is particularly suited for the quick of screen product and remodels, and shortening is changed
The type time, improve producing line efficiency.
The content of the invention
The present invention will overcome the drawbacks described above of the screen defects detection algorithm based on machine vision, there is provided one kind is based on convolution
The mobile phone screen MURA defect inspection methods of neural networks pruning algorithm.
To achieve the above object, the mobile phone screen MURA defects based on convolutional neural networks pruning algorithms of the present invention
Detection method comprises the following steps：
1) self-defined convolutional neural networks, train the network until restraining and have compared with high-accuracy by training data；
2) beta pruning of convolutional neural networks is carried out by the method for adaptive template matching, network size and network is reduced
Parameter；
3) mobile phone screen image data is gathered, picture pyramid is generated, picture block is divided into, for the life of test phase data
Into being sent to step 2) computing is carried out in convolutional neural networks after the beta pruning that obtains；
4) the characteristic pattern sum of middle hidden layer is taken as response diagram, and defect final position is obtained using the method for Threshold segmentation
And iris out, the method is particularly for detection MURA defects.
Step 2) described in adaptive template matching beta pruning be specifically：The characteristic pattern of hidden layer in the middle of network there is response
Part is corresponded in artwork, and part is left as background in the part as prospect, calculates the mean luminance differences of foreground and background
It is different；Several maximum characteristic patterns of difference are taken to be retained；Remaining characteristic pattern and associated convolution kernel from network
Beta pruning.
Step 3) described in picture pyramid be specifically：Original high resolution picture is dwindled into the picture of different scale,
The picture of these different scales collectively one group of picture pyramid.Different chis are detected using the pyramidal purpose of picture
On degree, different size of flaw.
Step 4) described in response diagram be specifically：The 4th output of convolutional layer is used, this layer of network after beta pruning
Characteristic pattern sum as defect response diagram.
Beneficial effects of the present invention are as follows：
The present invention is a kind of mobile phone screen MURA defect inspection methods based on convolutional neural networks pruning algorithms.It is based on
The convolutional neural networks algorithm of deep learning, the method matched using adaptive template carries out network beta pruning, compression network model
And parameter, to reach the effect of near real-time.Further, since convolutional neural networks algorithm can automatically learn to background texture
Information, can preferably be processed into the moire fringes interference as in.When screen product is remodeled, it is only necessary to which a small amount of training picture enters
Row network is finely tuned, it is possible to reassure that the accuracy rate of network is up to standard.
Compared with conventional method, the present invention can more effectively detect the thin defect of imaging, such as bulk MURA defects.This
Outward, conventional method needs to set quantity of parameters, the accuracy for being required for adjusting parameter just to can guarantee that algorithm of remodeling every time.The present invention
The neural network algorithm of use has good self adaptation and generalisation properties, can quickly carry out screen product and remodel, and saves and produces
Line deployment time.
Brief description of the drawings
Fig. 1 is adaptive template matching pruning algorithms frame diagram of the invention.
Fig. 2 is picture pyramid of the invention to the diagram for being divided into picture fritter.
Specific embodiment
Clear, complete explanation and description is carried out to technical scheme below.
The present invention proposes the mobile phone screen MURA defect inspection methods based on convolutional neural networks pruning algorithms, the method
On the mobile phone screen picture having been taken, determine position of the flaw on picture using convolutional neural networks algorithm and iris out.
Mobile phone screen MURA defect inspection method of the present invention based on convolutional neural networks pruning algorithms comprises the following steps：
Step 1, training stage data acquisition：Collection respectively includes flaw and normal picture fritter, is marked (1 expression
Normal picture is represented comprising flaw picture, 0).According to 9：1 ratio is divided into training set and checking collects；Self-defined convolutional Neural net
Network, using above-mentioned training data, training convolutional neural networks can reach higher dividing until convergence on checking collection
Class accuracy rate.By observing the characteristic pattern of hidden layer in the middle of visualization convolutional neural networks, can analyze whether network is learned well
Unwanted visual characteristic is arrived.
Step 2, the method matched using adaptive template carries out beta pruning to convolutional neural networks, reduces network size and net
Network parameter.Specifically, all N number of characteristic pattern on l-th convolutional layer, the response bit that i-th (i ＜ N) characteristic pattern is acquired
Put and correspond to back artwork, the portion as foreground template, the rest position of artwork as background template, calculate foreground template and
The mean flow rate difference D of background templatei.To all of DiDescending sort is carried out, preceding K (K ＜ N) mean flow rate difference D is takeniMost
Big characteristic pattern sum is used as response diagram.Remaining characteristic pattern convolution kernel corresponding with its directly by beta pruning, as shown in Figure 1.Cut
Whole network re -training again after branch, it is ensured that the network after beta pruning still has accuracy rate higher.
Step 3, test phase data genaration：Collection mobile phone screen image data, picture is included and only includes complete hand
Machine screen position.Picture is transformed into image pyramid using the zoom scale of different proportion, for multiple scale detecting.For each
The picture of yardstick, is divided into the picture block of fixed size, using all of picture block as a group, as the one of convolutional neural networks
Secondary input data, as shown in Figure 2.
Step 4, each picture block chooses the 4th all characteristic pattern sums of convolutional layer as response diagram, by response diagram
Again artwork is corresponded to, by the method for Threshold segmentation, the final position of flaw is obtained, and iris out.
Claims (4)
1. the mobile phone screen MURA defect inspection methods based on convolutional neural networks pruning algorithms, comprise the following steps：
1) self-defined convolutional neural networks, train the network until restraining and have compared with high-accuracy by training data；
2) beta pruning of convolutional neural networks is carried out by the method for adaptive template matching, network size and network ginseng is reduced
Number；
3) mobile phone screen image data is gathered, picture pyramid is generated, picture block is divided into, for test phase data genaration,
Be sent to step 2) obtain beta pruning after convolutional neural networks in carry out computing；
4) the characteristic pattern sum of middle hidden layer is taken as response diagram, and defect final position doubling-up is obtained using the method for Threshold segmentation
Go out, the method is particularly for detection MURA defects.
2. method according to claim 1, it is characterised in that：Step 2) described in adaptive template matching beta pruning it is specific
It is：During the part that the characteristic pattern of hidden layer in the middle of network has response is corresponded to artwork, part conduct is left in the part as prospect
Background, calculates the mean flow rate difference of foreground and background；Several maximum characteristic patterns of difference are taken to be retained；Remaining spy
Levy the beta pruning from network of figure and associated convolution kernel.
3. method according to claim 1, it is characterised in that：Step 3) described in picture pyramid be specifically：Will be original
High-resolution pictures dwindle into the picture of different scale, the picture of these different scales collectively one group of picture pyramid.
Using the pyramidal purpose of picture to detect on different scale, different size of flaw.
4. method according to claim 1, it is characterised in that step 4) described in response diagram be specifically：Use the 4th
The output of individual convolutional layer, using the characteristic pattern sum of this layer of network after beta pruning as defect response diagram.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611154333.5A CN106875373B (en) | 2016-12-14 | 2016-12-14 | Mobile phone screen MURA defect detection method based on convolutional neural network pruning algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611154333.5A CN106875373B (en) | 2016-12-14 | 2016-12-14 | Mobile phone screen MURA defect detection method based on convolutional neural network pruning algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106875373A true CN106875373A (en) | 2017-06-20 |
| CN106875373B CN106875373B (en) | 2019-12-20 |
Family
ID=59164615
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611154333.5A Active CN106875373B (en) | 2016-12-14 | 2016-12-14 | Mobile phone screen MURA defect detection method based on convolutional neural network pruning algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106875373B (en) |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107742294A (en) * | 2017-11-27 | 2018-02-27 | 歌尔股份有限公司 | Scratch detection method, apparatus and electronic equipment |
| CN108171707A (en) * | 2018-01-23 | 2018-06-15 | 武汉精测电子集团股份有限公司 | A kind of Mura defects level evaluation method and device 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 |
| CN108802041A (en) * | 2018-03-16 | 2018-11-13 | 浙江大学 | A kind of method that the small sample set of screen detection is quickly remodeled |
| CN108844966A (en) * | 2018-07-09 | 2018-11-20 | 广东速美达自动化股份有限公司 | A kind of screen detection method and detection system |
| CN109063834A (en) * | 2018-07-12 | 2018-12-21 | 浙江工业大学 | A kind of neural networks pruning method based on convolution characteristic response figure |
| CN109387521A (en) * | 2017-08-04 | 2019-02-26 | 欧姆龙株式会社 | Image processing system |
| CN109509172A (en) * | 2018-09-25 | 2019-03-22 | 无锡动视宫原科技有限公司 | A kind of liquid crystal display flaw detection method and system based on deep learning |
| CN109784129A (en) * | 2017-11-14 | 2019-05-21 | 北京京东尚科信息技术有限公司 | Information output method and device |
| CN109829914A (en) * | 2019-02-26 | 2019-05-31 | 视睿（杭州）信息科技有限公司 | The method and apparatus of testing product defect |
| CN109829895A (en) * | 2019-01-09 | 2019-05-31 | 武汉精立电子技术有限公司 | A kind of AOI defect inspection method based on GAN |
| CN109840901A (en) * | 2019-01-09 | 2019-06-04 | 武汉精立电子技术有限公司 | A kind of quick judgment method of split screen Mura |
| CN109858613A (en) * | 2019-01-22 | 2019-06-07 | 鹏城实验室 | A kind of compression method of deep neural network, system and terminal device |
| CN110097084A (en) * | 2019-04-03 | 2019-08-06 | 浙江大学 | Pass through the knowledge fusion method of projection feature training multitask student network |
| CN110211121A (en) * | 2019-06-10 | 2019-09-06 | 北京百度网讯科技有限公司 | Method and apparatus for pushing model |
| CN110532859A (en) * | 2019-07-18 | 2019-12-03 | 西安电子科技大学 | Remote Sensing Target detection method based on depth evolution beta pruning convolution net |
| CN110619391A (en) * | 2019-09-19 | 2019-12-27 | 华南理工大学 | Detection model compression method and device and computer readable storage medium |
| CN110728681A (en) * | 2019-12-19 | 2020-01-24 | 武汉精立电子技术有限公司 | Mura defect detection method and device |
| CN111080633A (en) * | 2019-12-20 | 2020-04-28 | Oppo（重庆）智能科技有限公司 | Screen defect detection method and device, terminal equipment and storage medium |
| CN111476759A (en) * | 2020-03-13 | 2020-07-31 | 深圳市鑫信腾机器人科技有限公司 | Screen surface detection method and device, terminal and storage medium |
| CN111563883A (en) * | 2020-04-22 | 2020-08-21 | 惠州旭鑫智能技术有限公司 | Screen visual positioning method, positioning device and storage medium |
| CN111583129A (en) * | 2020-04-09 | 2020-08-25 | 天津大学 | Screen shot image moire removing method based on convolutional neural network AMNet |
| 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 |
| US11568324B2 (en) | 2018-12-20 | 2023-01-31 | Samsung Display Co., Ltd. | Adversarial training method for noisy labels |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104036323A (en) * | 2014-06-26 | 2014-09-10 | 叶茂 | Vehicle detection method based on convolutional neural network |
| CN104850858A (en) * | 2015-05-15 | 2015-08-19 | 华中科技大学 | Injection-molded product defect detection and recognition method |
| CN104977313A (en) * | 2014-04-09 | 2015-10-14 | 四川省特种设备检验研究院 | Method and device for detecting and identifying X-ray image defects of welding seam |
| CN105118044A (en) * | 2015-06-16 | 2015-12-02 | 华南理工大学 | Method for automatically detecting defects of wheel-shaped cast product |
-
2016
- 2016-12-14 CN CN201611154333.5A patent/CN106875373B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104977313A (en) * | 2014-04-09 | 2015-10-14 | 四川省特种设备检验研究院 | Method and device for detecting and identifying X-ray image defects of welding seam |
| CN104036323A (en) * | 2014-06-26 | 2014-09-10 | 叶茂 | Vehicle detection method based on convolutional neural network |
| CN104850858A (en) * | 2015-05-15 | 2015-08-19 | 华中科技大学 | Injection-molded product defect detection and recognition method |
| CN105118044A (en) * | 2015-06-16 | 2015-12-02 | 华南理工大学 | Method for automatically detecting defects of wheel-shaped cast product |
Cited By (38)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10885620B2 (en) | 2017-08-04 | 2021-01-05 | Omron Corporation | Neural network image processing system |
| CN109387521A (en) * | 2017-08-04 | 2019-02-26 | 欧姆龙株式会社 | Image processing system |
| CN109784129A (en) * | 2017-11-14 | 2019-05-21 | 北京京东尚科信息技术有限公司 | Information output method and device |
| CN107742294A (en) * | 2017-11-27 | 2018-02-27 | 歌尔股份有限公司 | Scratch detection method, apparatus and electronic equipment |
| CN108171707A (en) * | 2018-01-23 | 2018-06-15 | 武汉精测电子集团股份有限公司 | A kind of Mura defects level evaluation method and device 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 |
| 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 |
| CN108802041A (en) * | 2018-03-16 | 2018-11-13 | 浙江大学 | A kind of method that the small sample set of screen detection is quickly remodeled |
| CN108844966A (en) * | 2018-07-09 | 2018-11-20 | 广东速美达自动化股份有限公司 | A kind of screen detection method and detection system |
| CN109063834A (en) * | 2018-07-12 | 2018-12-21 | 浙江工业大学 | A kind of neural networks pruning method based on convolution characteristic response figure |
| CN109063834B (en) * | 2018-07-12 | 2021-07-20 | 浙江工业大学 | Neural network pruning method based on convolution characteristic response graph |
| CN109509172A (en) * | 2018-09-25 | 2019-03-22 | 无锡动视宫原科技有限公司 | A kind of liquid crystal display flaw detection method and system based on deep learning |
| US11830240B2 (en) | 2018-12-20 | 2023-11-28 | Samsung Display Co., Ltd. | Adversarial training method for noisy labels |
| US11568324B2 (en) | 2018-12-20 | 2023-01-31 | Samsung Display Co., Ltd. | Adversarial training method for noisy labels |
| CN109829895A (en) * | 2019-01-09 | 2019-05-31 | 武汉精立电子技术有限公司 | A kind of AOI defect inspection method based on GAN |
| CN109840901A (en) * | 2019-01-09 | 2019-06-04 | 武汉精立电子技术有限公司 | A kind of quick judgment method of split screen Mura |
| CN109840901B (en) * | 2019-01-09 | 2020-08-25 | 武汉精立电子技术有限公司 | Quick judgment method for split screen Mura |
| CN109858613A (en) * | 2019-01-22 | 2019-06-07 | 鹏城实验室 | A kind of compression method of deep neural network, system and terminal device |
| CN109858613B (en) * | 2019-01-22 | 2021-02-19 | 鹏城实验室 | Compression method and system of deep neural network and terminal equipment |
| CN109829914A (en) * | 2019-02-26 | 2019-05-31 | 视睿（杭州）信息科技有限公司 | The method and apparatus of testing product defect |
| CN109829914B (en) * | 2019-02-26 | 2021-09-03 | 视睿（杭州）信息科技有限公司 | Method and device for detecting product defects |
| CN110097084A (en) * | 2019-04-03 | 2019-08-06 | 浙江大学 | Pass through the knowledge fusion method of projection feature training multitask student network |
| CN110097084B (en) * | 2019-04-03 | 2021-08-31 | 浙江大学 | Knowledge fusion method for training multitask student network through projection characteristics |
| CN110211121A (en) * | 2019-06-10 | 2019-09-06 | 北京百度网讯科技有限公司 | Method and apparatus for pushing model |
| CN110211121B (en) * | 2019-06-10 | 2021-07-16 | 北京百度网讯科技有限公司 | Method and device for pushing model |
| CN110532859A (en) * | 2019-07-18 | 2019-12-03 | 西安电子科技大学 | Remote Sensing Target detection method based on depth evolution beta pruning convolution net |
| CN110532859B (en) * | 2019-07-18 | 2021-01-22 | 西安电子科技大学 | Remote sensing image target detection method based on deep evolution pruning convolution net |
| CN110619391B (en) * | 2019-09-19 | 2023-04-18 | 华南理工大学 | Detection model compression method and device and computer readable storage medium |
| CN110619391A (en) * | 2019-09-19 | 2019-12-27 | 华南理工大学 | Detection model compression method and device and computer readable storage medium |
| CN110728681B (en) * | 2019-12-19 | 2020-05-22 | 武汉精立电子技术有限公司 | Mura defect detection method and device |
| CN110728681A (en) * | 2019-12-19 | 2020-01-24 | 武汉精立电子技术有限公司 | Mura defect detection method and device |
| CN111080633A (en) * | 2019-12-20 | 2020-04-28 | Oppo（重庆）智能科技有限公司 | Screen defect detection method and device, terminal equipment and storage medium |
| CN111476759B (en) * | 2020-03-13 | 2022-03-25 | 深圳市鑫信腾机器人科技有限公司 | Screen surface detection method and device, terminal and storage medium |
| CN111476759A (en) * | 2020-03-13 | 2020-07-31 | 深圳市鑫信腾机器人科技有限公司 | Screen surface detection method and device, terminal and storage medium |
| CN111583129A (en) * | 2020-04-09 | 2020-08-25 | 天津大学 | Screen shot image moire removing method based on convolutional neural network AMNet |
| CN111563883B (en) * | 2020-04-22 | 2023-05-05 | 惠州旭鑫智能技术有限公司 | Screen vision positioning method, positioning equipment and storage medium |
| CN111563883A (en) * | 2020-04-22 | 2020-08-21 | 惠州旭鑫智能技术有限公司 | Screen visual positioning method, positioning device and storage medium |
| 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 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106875373B (en) | 2019-12-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106875373A (en) | Mobile phone screen MURA defect inspection methods based on convolutional neural networks pruning algorithms | |
| CN107123111B (en) | Deep residual error network construction method for mobile phone screen defect detection | |
| CN108961217B (en) | Surface defect detection method based on regular training | |
| CN110992317B (en) | PCB defect detection method based on semantic segmentation | |
| CN109636772A (en) | The defect inspection method on the irregular shape intermetallic composite coating surface based on deep learning | |
| CN109584227A (en) | A kind of quality of welding spot detection method and its realization system based on deep learning algorithm of target detection | |
| CN109711474A (en) | A kind of aluminium material surface defects detection algorithm based on deep learning | |
| CN111402226A (en) | Surface defect detection method based on cascade convolution neural network | |
| CN109472769A (en) | A kind of bad image defect detection method and system | |
| CN110473173A (en) | A kind of defect inspection method based on deep learning semantic segmentation | |
| CN107016413B (en) | A kind of online stage division of tobacco leaf based on deep learning algorithm | |
| CN108898610A (en) | A kind of object contour extraction method based on mask-RCNN | |
| CN106709936A (en) | Single target tracking method based on convolution neural network | |
| CN106875381A (en) | A kind of phone housing defect inspection method based on deep learning | |
| CN108492291B (en) | CNN segmentation-based solar photovoltaic silicon wafer defect detection system and method | |
| CN107064170A (en) | One kind detection phone housing profile tolerance defect method | |
| CN109544522A (en) | A kind of Surface Defects in Steel Plate detection method and system | |
| CN109829895A (en) | A kind of AOI defect inspection method based on GAN | |
| CN113554631B (en) | Chip surface defect detection method based on improved network | |
| CN108802041A (en) | A kind of method that the small sample set of screen detection is quickly remodeled | |
| JP2022027473A (en) | Generation of training data usable for inspection of semiconductor sample | |
| CN112686894A (en) | FPCB (flexible printed circuit board) defect detection method and device based on generative countermeasure network | |
| Arikan et al. | Surface defect classification in real-time using convolutional neural networks | |
| CN109145846A (en) | Material microstructure intelligent recognition analysis system and analysis method | |
| CN109685030A (en) | A kind of mug rim of a cup defects detection classification method based on convolutional neural networks |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20180418 Address after: 516006 Huizhou, Zhongkai hi tech Zone, Guangdong, Chen Jiang Street, 36 North (two) fifth. Applicant after: Huizhou Xu Xin Intelligent Technology Co., Ltd. Address before: 310027 Hangzhou, Zhejiang Province, Xihu District, Zhejiang Road, No. 38, No. Applicant before: Zhejiang University |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |