CN107392896A - A kind of Wood Defects Testing method and system based on deep learning - Google Patents

A kind of Wood Defects Testing method and system based on deep learning Download PDF

Info

Publication number
CN107392896A
CN107392896A CN201710577252.4A CN201710577252A CN107392896A CN 107392896 A CN107392896 A CN 107392896A CN 201710577252 A CN201710577252 A CN 201710577252A CN 107392896 A CN107392896 A CN 107392896A
Authority
CN
China
Prior art keywords
layer
deep learning
image
defects
wood
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
Application number
CN201710577252.4A
Other languages
Chinese (zh)
Other versions
CN107392896B (en
Inventor
黄坤山
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Foshan Nanhai Guangdong Technology University CNC Equipment Cooperative Innovation Institute
Foshan Guangdong University CNC Equipment Technology Development Co. Ltd
Original Assignee
Foshan Nanhai Guangdong Technology University CNC Equipment Cooperative Innovation Institute
Foshan Guangdong University CNC Equipment Technology Development Co. Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Foshan Nanhai Guangdong Technology University CNC Equipment Cooperative Innovation Institute, Foshan Guangdong University CNC Equipment Technology Development Co. Ltd filed Critical Foshan Nanhai Guangdong Technology University CNC Equipment Cooperative Innovation Institute
Priority to CN201710577252.4A priority Critical patent/CN107392896B/en
Publication of CN107392896A publication Critical patent/CN107392896A/en
Application granted granted Critical
Publication of CN107392896B publication Critical patent/CN107392896B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G06T7/0004—Industrial image inspection
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • 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
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T1/00—General purpose image data processing
    • G06T1/20—Processor architectures; Processor configuration, e.g. pipelining
    • 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
    • G01N2021/8466—Investigation of vegetal material, e.g. leaves, plants, fruits
    • 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
    • G06T2207/20021—Dividing image into blocks, subimages or windows
    • 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
    • G06T2207/30161—Wood; Lumber

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Theoretical Computer Science (AREA)
  • Pathology (AREA)
  • Health & Medical Sciences (AREA)
  • Immunology (AREA)
  • Chemical & Material Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Quality & Reliability (AREA)
  • Signal Processing (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention provides a kind of Wood Defects Testing method based on deep learning, specifically include：Gather image；Segment the image into equirotal image block；The defects of different classes of image block and flawless image block are chosen as training sample set；Utilize training sample set off-line training deep learning algorithm；Using the deep learning algorithm trained, the step such as the defects of on-line checking identification timber image.For the present invention by powerful deep learning algorithm, accurately on-line checking identification solves the insurmountable problem of traditional images Processing Algorithm the different complex texture wood surfaces the defects of.Present invention also offers a kind of Wood Defects Testing system based on deep learning, can effectively accelerate detection speed by the cooperation between image capture module, deep learning algorithm processing module and control execution module and improve practicality.

Description

A kind of Wood Defects Testing method and system based on deep learning
Technical field
The present invention relates to technical field of vision detection, and in particular to a kind of Wood Defects Testing method based on deep learning And system.
Background technology
With the intensivization development of wood-processing industry, the output of Wood products persistently increases substantially.In production, it is right Wood surface crudy is high-caliber overcritical, the especially requirement of uniformity so that traditional manual detection mode is difficult To be competent at, it is badly in need of a kind of efficient detection method and system now to realize the automation of Wood Defects Testing.
In recent years, machine vision technique is developed rapidly, and in the industrial production, people utilize technology success Automation mechanized operation is realized to the detection of various products outward appearance.But wood surface is due to the texture with complexity, and the species of defect It is various, according to national standard can defects in timber be divided into knot, discoloration, rotten, small holes caused by worms, crackle, stem form defect, wood structure Defect, 10 major classes of scar (damage), timber manufacturing deficiency and deformation, the image of timber is gathered by NI Vision Builder for Automated Inspection, is utilized Traditional image processing algorithm detects the defects of timber, can face huge challenge.
The fast development of deep learning algorithm therewith, it has very outstanding performance in computer vision field so that passes The insurmountable problem of image processing algorithm of system is resolved in succession, therefore, can be attempted a variety of using timber Defect image trains deep learning algorithm, and the defects of recycling the deep learning algorithm on-line checking timber trained, this is one The effective solution method of kind.
The content of the invention
It is an object of the invention to by a kind of Wood Defects Testing method and system based on deep learning, utilize timber A variety of defect images train deep learning algorithm, recycle lacking for the deep learning algorithm on-line checking timber trained Fall into, the requirement of real-time of commercial Application can be met.
To realize above-mentioned technical proposal, the invention provides a kind of Wood Defects Testing method based on deep learning, tool Body comprises the following steps：
A kind of 1. Wood Defects Testing method based on deep learning, it is characterised in that comprise the following steps：
Step 1：Image capture module gathers image；
Step 2：Segment the image into equirotal image block；
Step 3：The defects of different classes of image block and flawless image block are chosen as training sample set；
Step 4：Collect training deep learning algorithm offline using training sample；
Step 5：Using the deep learning algorithm trained, the defects of on-line checking identification timber image.
Preferably, in step 2, equirotal image block is segmented the image into：I.e. using s/2 as step-length, what is collected Image is divided into the square image blocks that size is all s*s, and wherein s represents the pixel chi for the square image blocks length of side being divided into It is very little.
Preferably, in step 3, choose the defects of different classes of image block and flawless image block make sample training collection, Training sample set is pre-processed before training, and increases the size of data set by the method for stochastic transformation, specifically Step is as follows：
Step 31：Whitening processing is carried out to training sample set；
Step 32：Enter row stochastic left and right upset to image；
Step 33：The contrast of stochastic transformation image.
Preferably, in step 4, using training sample set off-line training deep learning algorithm, wherein deep learning algorithm is adopted It is specific as follows with multilayer convolutional neural networks vgg16 models, training process：
Step 41：Construct multilayer convolutional neural networks vgg16；
Step 42：Using training sample set, multilayer convolutional neural networks vgg16 error gradient is done using ADAM algorithms Steepest declines optimization, off-line training construction multilayer convolutional neural networks vgg16；
Step 43：Multilayer convolutional neural networks vgg16 is increased income deep learning system after forming using Google TensorFlow is developed, and deep learning algorithm is accelerated up to GPU using tall and handsome.
Preferably, in step 41, construct multilayer convolutional neural networks vgg16, wherein have 16 layers of weight layer, 5 layers of pond layer, 1 layer of input layer and 23 layers altogether of 1 layer of output layer, concrete structure are：1st layer be the image block that size is s*s input layer；2nd Layer is convolutional layer with the 3rd layer, and the size of convolution kernel is all 3*3, and every layer of convolution nuclear volume is all 64；4th layer is most Big value pond layer；5th layer and the 6th layer is convolutional layer, and the size of convolution kernel is all 3*3, and every layer of convolution nuclear volume is all 128；7th layer is maximum pond layer；8th, 9 and 10 layer is convolutional layer, and the size of convolution kernel is all 3*3, and every layer Convolution nuclear volume is all 256；11th layer is maximum pond layer；12nd, 13 and 14 layer is convolutional layer, and the size of convolution kernel is all For 3*3, and every layer of convolution nuclear volume is all 512；15th layer is maximum pond layer；16th, 17 and 18 layer is convolution Layer, the size of convolution kernel is all 3*3, and every layer of convolution nuclear volume is all 512；19th layer is maximum pond layer；The 20,21 and 22 layers are full articulamentum, wherein the 20th, 21 full articulamentum has 4096 neurodes, the 22nd full articulamentum has 1000 Individual neurode；23rd layer is softmax classification layers.
Preferably, in step 42, specifically comprise the following steps：
The preparation of step 421, training sample set：Gather knot, discoloration, rotten, small holes caused by worms, crackle, stem form defect, wood Material construction defect, scar (damage), timber manufacturing deficiency and 10 major class Filtering for Wood Defect Images blocks of deformation and clean timber image For block as training sample set, each of which class defects in timber gather 5000 pieces of image blocks and 5000 pieces of zero defect image blocks, and By carrying out simple pretreatment increase training sample set to the image block of selection；
Step 422, training method：Iteration is trained for 15000 times to multilayer convolutional neural networks vgg16, is changed each time In generation, concentrates from training sample choose 32 pieces of image blocks as input at random, using ADAM algorithms to multilayer convolutional neural networks Vgg16 error gradient does steepest and declines optimization.
Present invention also offers a kind of Wood Defects Testing method based on deep learning, including：
Image capture module, described image acquisition module is the industrial camera using line array CCD, special using machine vision With light source and industrial camera collection surface image；
Image processing module, described image processing module are soft using the host computer of industrial computer carrying image processing module Part, the software includes image procossing interactive interface, deep learning algorithm, database and communication module, using deep learning Algorithm carries out Wood Defects Testing to the image of collection, and defect information is sent to control execution unit, the deep learning Algorithm is developed using the Google deep learning system TensorFlow that increase income, and deep learning is calculated up to GPU using tall and handsome Method is accelerated.；
Control execution module：The defects of being sent according to image processing module information, using programmable logic controller (PLC) Surface defect is marked PLC control marking machines or alarm is prompted surface defect.
Preferably, described image processing interactive interface provide interface allow user's arrange parameter and allow user choose training sample This collection, while the real-time condition that display window allows user to observe surface defects detection is provided.
Preferably, the database is used to surface defects detection result is stored and managed.
Preferably, the communication module uses Ethernet ICP/IP protocol, allows upper computer software and action execution unit mould The slave computer communication of block.
A kind of Wood Defects Testing method beneficial effect based on deep learning provided by the invention is：
1) based on the Wood Defects Testing method of deep learning, by under timber different texture the defects of different type Image block trains deep learning algorithm, recycles the deep learning algorithm trained to detect the defects of identifying timber, solves Traditional images Processing Algorithm can not detect the problem for identifying complex texture defect.
2) can be fallen vacant based on the Wood Defects Testing method of deep learning in the detection of the high-precision surface of complex texture Fall into, and identify the type of defect, then developed using the Google deep learning system TensorFlow that increase income, and utilize It is tall and handsome that deep learning algorithm is accelerated up to GPU, the requirement of real-time of commercial Application can be met.
3) based on the Wood Defects Testing system of deep learning, image capture module, image processing module and control are passed through Cooperation between execution module processed can efficiently realize the detection to different complex texture wood surfaces, and not only accuracy of detection is high, Adaptability is extensive, and strong robustness, and speed is fast.
Brief description of the drawings
Fig. 1 is the flow chart of the Wood Defects Testing method based on deep learning in the present invention.
Fig. 2 is multilayer convolutional neural networks vgg16 training process accuracy detects schematic diagrams in the present invention.
Fig. 3 is multilayer convolutional neural networks vgg16 training process cost function value detects schematic diagrams in the present invention.
Fig. 4 is multilayer convolutional neural networks vgg16 schematic diagrames in the present invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete Whole description, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.Ability The every other embodiment that domain ordinary person is obtained under the premise of creative work is not made, belong to the protection of the present invention Scope.
Embodiment 1：A kind of Wood Defects Testing method based on deep learning.
Shown in reference picture 1, a kind of Wood Defects Testing method based on deep learning, specifically comprise the following steps：
Step 1：Image capture module gathers image, i.e., gathers timber table using machine vision special light source and industrial camera Face image；
Step 2：Equirotal image block is segmented the image into, specific dividing method is：Using s/2 as step-length, adopting The image integrated is divided into size all as s*s square image blocks, and wherein size s can not typically be less than the big of defects in timber Small, the size of image block can take 128*128, wherein the pixel size for the image block length of side that 128 expressions are divided into；
Step 3：The defects of different classes of image block and flawless image block are chosen as training sample set；
Wherein, choose the defects of different classes of image block and flawless image block makees sample training collection, before training Training sample set is simply pre-processed, and by the methods of stochastic transformations a series of come artificial increase training sample The size of collection so that the deep learning algorithm generalization ability after training is stronger, comprises the following steps that：
Step 31：Whitening processing is carried out to training sample set, it is possible to reduce because being influenceed caused by illumination is different；
Step 32：Enter row stochastic left and right upset to image, timber of the deep learning algorithm to different directions can be increased The detectability of defect；
Step 33：The contrast of stochastic transformation image, it is possible to reduce the influence caused by image resolution ratio difference；
Step 4：Utilize training sample set off-line training deep learning algorithm；
Wherein, deep learning algorithm uses multilayer convolutional neural networks vgg16 models, and training process is specific as follows：
Step 41：Construct multilayer convolutional neural networks vgg16；
As shown in figure 3, in step 41, multilayer convolutional neural networks vgg16 is constructed, wherein having 16 layers of weight layer, 5 layers of pond Layer, 1 layer of input layer and 23 layers altogether of 1 layer of output layer, concrete structure are：1st layer be the image block that size is s*s input layer； Layers 2 and 3 is convolutional layer, and the size of convolution kernel is all 3*3, and every layer of convolution nuclear volume is all 64；4th layer is Maximum pond layer；5th layer and the 6th layer is convolutional layer, and the size of convolution kernel is all 3*3, and every layer convolution nuclear volume all For 128；7th layer is maximum pond layer；8th, 9 and 10 layer is convolutional layer, and the size of convolution kernel is all 3*3, and every layer Convolution nuclear volume all be 256；11th layer is maximum pond layer；12nd, 13 and 14 layer is convolutional layer, the size of convolution kernel All it is 3*3, and every layer of convolution nuclear volume is all 512；15th layer is maximum pond layer；16th, 17 and 18 layer is volume Lamination, the size of convolution kernel is all 3*3, and every layer of convolution nuclear volume is all 512；19th layer is maximum pond layer； 20th, 21 and 22 layer is full articulamentum, wherein the 20th, 21 full articulamentum has 4096 neurodes, the 22nd full articulamentum has 1000 neurodes；23rd layer is softmax classification layers；
Convolutional layer is used for the high-level characteristic for extracting image, and the input of maximum pond layer is typically derived from a convolution Layer, main function there is provided very strong robustness, take the maximum in a pocket, if the now other values in this region It is slightly changed, or image slightly translates, Chi Huahou result is still constant, and reduces the quantity of parameter, prevents over-fitting The generation of phenomenon, the typically no parameter of pond layer, so when backpropagation, only need to be to input parameter derivation, it is not necessary to enter Row right value update；And softmax layers are mainly used in classifying, timber has a few class defects, and softmax is with regard to how many node；
Step 42：Using training sample set, multilayer convolutional neural networks vgg16 error gradient is done using ADAM algorithms Steepest declines optimization, off-line training construction multilayer convolutional neural networks vgg16；
Concrete operations are as follows：(1) preparation of training sample set：Gather knot, discoloration, rotten, small holes caused by worms, crackle, trunk shape Shape defect, wood structure defect, scar (damage), timber manufacturing deficiency and 10 major class Filtering for Wood Defect Images blocks of deformation and zero defect Timber image block gathers 5000 pieces of image blocks and 5000 pieces of zero defect figures as training sample set, each of which class defects in timber As block, and by carrying out simple pretreatment increase training sample set to the image block of selection；(2) training method：Iteration 15000 Secondary that multilayer convolutional neural networks vgg16 is trained, iteration is concentrated from training sample choose 32 pieces of image blocks at random each time As input, steepest is done to multilayer convolutional neural networks vgg16 error gradient using ADAM algorithms and declines optimization, as a result as schemed (curve in Fig. 2 and Fig. 3, which has, passes through smoothing processing) shown in 2 and Fig. 3, multilayer convolutional neural networks vgg16 accuracy with The increase of iteration and improve, while cost function value diminishes with the increase of iteration, final multilayer convolutional neural networks Vgg16 Wood Defects Testing recognition correct rate can reach 95% or so, and not only speed is fast, and strong robustness；
Step 5：Using the deep learning algorithm trained, the defects of on-line checking identification timber image, and defect is realized Automatic classification.
Based on a kind of Wood Defects Testing method based on deep learning, pass through the different type under timber different texture The defects of image block training deep learning algorithm, recycle the deep learning algorithm that trains to detect the defects of identifying timber, Solves the problem that traditional images Processing Algorithm None- identified detects complex texture defect.
Embodiment 2：A kind of Wood Defects Testing system based on deep learning
A kind of Wood Defects Testing system based on deep learning, is specifically included：
Image capture module, described image acquisition module is the industrial camera using line array CCD, special using machine vision With light source and industrial camera collection surface image；
Image processing module, described image processing module are soft using the host computer of industrial computer carrying image processing module Part, the software includes image procossing interactive interface, deep learning algorithm, database and communication module, using deep learning Algorithm carries out Wood Defects Testing to the image of collection, and defect information is sent to control execution unit；
Control execution module：The defects of being sent according to image processing module information, using programmable logic controller (PLC) Surface defect is marked PLC control marking machines or alarm is prompted surface defect.
In the present embodiment, image procossing interactive interface, there is provided interface allow user's arrange parameter and allow user choose training Sample set, there is provided display window allow user observe surface defects detection real-time condition；Deep learning algorithm, is opened using Google Depth learning system TensorFlow is developed, and deep learning algorithm is accelerated up to GPU using tall and handsome, is, for example, Industrial computer configures the tall and handsome GPU up to GTX1080 models, tall and handsome up to GPU acceleration cuda7.0 kits by downloading installation, then pacifies Fill the deep neural network storehouse cuDNN that GPU accelerates, you can accelerate the operation of deep learning algorithm using tensorflow frameworks； Image procossing interactive interface, there is provided interface allow user's arrange parameter and allow user choose training sample set, there is provided display window User is allowed to observe the real-time condition of surface defects detection；Database, for surface defects detection result to be stored and managed； Communication module, using Ethernet ICP/IP protocol, upper computer software and the slave computer of action execution unit module is allowed to communicate.
Based on a kind of Wood Defects Testing system based on deep learning, increased income deep learning system using Google TensorFlow is developed, and deep learning algorithm is accelerated up to GPU using tall and handsome, can meet the real-time of commercial Application Property require, and can efficiently be realized pair by the cooperation between image capture module, image processing module and control execution module The detection of different complex texture wood surfaces, not only accuracy of detection is high, and adaptability is extensive, and strong robustness, and speed is fast.
Described above is presently preferred embodiments of the present invention, but the present invention should not be limited to the embodiment and accompanying drawing institute is public The content opened, so every do not depart from the lower equivalent or modification completed of spirit disclosed in this invention, both fall within protection of the present invention Scope.

Claims (10)

  1. A kind of 1. Wood Defects Testing method based on deep learning, it is characterised in that specifically include following steps：
    Step 1：Image capture module gathers image；
    Step 2：Segment the image into equirotal image block；
    Step 3：The defects of different classes of image block and flawless image block are chosen as training sample set；
    Step 4：Utilize training sample set off-line training deep learning algorithm；
    Step 5：Using the deep learning algorithm trained, the defects of on-line checking identification timber image.
  2. 2. the Wood Defects Testing method according to claim 1 based on deep learning, it is characterised in that in step 2, Image is divided into equirotal image block：I.e. using s/2 as step-length, the image collected be divided into size all for s*s just Square image block, wherein s represent the Pixel Dimensions for the square image blocks length of side being divided into.
  3. 3. the Wood Defects Testing method according to claim 1 based on deep learning, it is characterised in that in step 3, choosing The defects of different classes of image block and flawless image block is taken to enter before training to training sample set as sample training collection Row pre-processes, and increases the size of data set by the method for stochastic transformation, comprises the following steps that：
    Step 31：Whitening processing is carried out to training sample set；
    Step 32：Enter row stochastic left and right upset to image；
    Step 33：The contrast of stochastic transformation image.
  4. 4. the Wood Defects Testing method according to claim 1 based on deep learning, it is characterised in that in step 4, profit With training sample set off-line training deep learning algorithm, wherein deep learning algorithm uses multilayer convolutional neural networks vgg16 moulds Type, training process are specific as follows：
    Step 41：Construct multilayer convolutional neural networks vgg16；
    Step 42：Using training sample set, steepest is done to multilayer convolutional neural networks vgg16 error gradient using ADAM algorithms Decline optimization, off-line training construction multilayer convolutional neural networks vgg16；
    Step 43：Multilayer convolutional neural networks vgg16 is increased income deep learning system TensorFlow after forming using Google Developed, and deep learning algorithm is accelerated up to GPU using tall and handsome.
  5. 5. the Wood Defects Testing method according to claim 4 based on deep learning, it is characterised in that in step 41, Multilayer convolutional neural networks vgg16 is constructed, wherein having 16 layers of weight layer, 5 layers of pond layer, 1 layer of input layer and 1 layer of output layer altogether 23 layers, concrete structure is：1st layer be the image block that size is s*s input layer；Layers 2 and 3 is convolutional layer, convolution kernel Size be all 3*3, and every layer of convolution nuclear volume is all 64；4th layer is maximum pond layer；5th layer and the 6th layer is Convolutional layer, the size of convolution kernel is all 3*3, and every layer of convolution nuclear volume is all 128；7th layer is maximum pond layer； 8th, 9 and 10 layer is convolutional layer, and the size of convolution kernel is all 3*3, and every layer of convolution nuclear volume is all 256；11th layer For maximum pond layer；12nd, 13 and 14 layer is convolutional layer, and the size of convolution kernel is all 3*3, and every layer of convolution nuclear volume All it is 512；15th layer is maximum pond layer；16th, 17 and 18 layer is convolutional layer, and the size of convolution kernel is all 3*3, and Every layer of convolution nuclear volume is all 512；19th layer is maximum pond layer；20th, 21 and 22 layer is full articulamentum, wherein the 20,21 full articulamentums have 4096 neurodes, and the 22nd full articulamentum has 1000 neurodes；23rd layer is softmax points Class layer.
  6. 6. the Wood Defects Testing method according to claim 4 based on deep learning, it is characterised in that in step 42, Specifically comprise the following steps：
    The preparation of step 421, training sample set：Gather knot, discoloration, rotten, small holes caused by worms, crackle, stem form defect, timber structure Make defect, scar (damage), timber manufacturing deficiency and 10 major class Filtering for Wood Defect Images blocks of deformation and clean timber image block is made For training sample set, each of which class defects in timber gather 5000 pieces of image blocks and 5000 pieces of zero defect image blocks, and pass through Simple pretreatment increase training sample set is carried out to the image block of selection；
    Step 422, training method：Iteration is trained for 15000 times to multilayer convolutional neural networks vgg16, each time iteration with Machine is concentrated from training sample chooses 32 pieces of image blocks as input, using ADAM algorithms to multilayer convolutional neural networks vgg16's Error gradient does steepest and declines optimization.
  7. 7. a kind of existed based on the system for realizing the Wood Defects Testing method based on deep learning described in claim 6, its feature In, including：
    Image capture module, described image acquisition module is the industrial camera using line array CCD, using machine vision dedicated optical Source and industrial camera collection surface image；
    Image processing module, described image processing module are the upper computer software using industrial computer carrying image processing module, should Software includes image procossing interactive interface, deep learning algorithm, database and communication module, using deep learning algorithm pair The image of collection carries out Wood Defects Testing, and defect information is sent to control execution unit, the deep learning algorithm profit Developed with the Google deep learning system TensorFlow that increase income, and deep learning algorithm is carried out up to GPU using tall and handsome Accelerate；
    Control execution module：The defects of being sent according to image processing module information, controlled using programmable logic controller (PLC) PLC Surface defect is marked marking machine processed or alarm is prompted surface defect.
  8. 8. the Wood Defects Testing system based on deep learning as claimed in claim 7, it is characterised in that described image processing Interactive interface, which provides interface, to be allowed and user's arrange parameter and allows user to choose training sample set, while is provided display window and allowed user Observe the real-time condition of surface defects detection.
  9. 9. the Wood Defects Testing system based on deep learning as claimed in claim 7, it is characterised in that the database is used In surface defects detection result is stored and managed.
  10. 10. the Wood Defects Testing system based on deep learning as claimed in claim 7, it is characterised in that the communication mould Block uses Ethernet ICP/IP protocol, allows upper computer software and the slave computer of action execution unit module to communicate.
CN201710577252.4A 2017-07-14 2017-07-14 A kind of Wood Defects Testing method and system based on deep learning Active CN107392896B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710577252.4A CN107392896B (en) 2017-07-14 2017-07-14 A kind of Wood Defects Testing method and system based on deep learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710577252.4A CN107392896B (en) 2017-07-14 2017-07-14 A kind of Wood Defects Testing method and system based on deep learning

Publications (2)

Publication Number Publication Date
CN107392896A true CN107392896A (en) 2017-11-24
CN107392896B CN107392896B (en) 2019-11-08

Family

ID=60339848

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710577252.4A Active CN107392896B (en) 2017-07-14 2017-07-14 A kind of Wood Defects Testing method and system based on deep learning

Country Status (1)

Country Link
CN (1) CN107392896B (en)

Cited By (76)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108038853A (en) * 2017-12-18 2018-05-15 浙江工业大学 Ceramic tile surface defect identification method based on convolutional neural network and active learning
CN108074231A (en) * 2017-12-18 2018-05-25 浙江工业大学 Magnetic sheet surface defect detection method based on convolutional neural network
CN108154504A (en) * 2017-12-25 2018-06-12 浙江工业大学 Method for detecting surface defects of steel plate based on convolutional neural network
CN108230317A (en) * 2018-01-09 2018-06-29 北京百度网讯科技有限公司 Steel plate defect detection sorting technique, device, equipment and computer-readable medium
CN108230318A (en) * 2018-01-09 2018-06-29 北京百度网讯科技有限公司 Ladle defects detection sorting technique, device, equipment and computer-readable medium
CN108247764A (en) * 2017-12-14 2018-07-06 北京木业邦科技有限公司 Plank cutting method, device, electronic equipment and medium based on machine learning
CN108262809A (en) * 2017-12-14 2018-07-10 北京木业邦科技有限公司 Plank processing method, device, electronic equipment and medium based on artificial intelligence
CN108311411A (en) * 2018-02-07 2018-07-24 陕西科技大学 A kind of cordyceps sinensis intelligence sorting system and its application method
CN108346153A (en) * 2018-03-22 2018-07-31 北京木业邦科技有限公司 The machine learning of defects in timber and restorative procedure, device, system, electronic equipment
CN108362702A (en) * 2017-12-14 2018-08-03 北京木业邦科技有限公司 A kind of defect of veneer detection method, system and equipment based on artificial intelligence
CN108375353A (en) * 2017-12-14 2018-08-07 北京木业邦科技有限公司 A kind of log detection method, apparatus and equipment
CN108389181A (en) * 2018-01-23 2018-08-10 南京大学 It is a kind of based on the part defect of deep learning and normal direction figure detection and localization method
CN108491867A (en) * 2018-03-12 2018-09-04 苏州卓融新能源科技有限公司 Image Matching based on artificial intelligence and recognition methods
CN108776808A (en) * 2018-05-25 2018-11-09 北京百度网讯科技有限公司 A kind of method and apparatus for detecting ladle corrosion defect
CN108789426A (en) * 2017-12-29 2018-11-13 金门工程建设有限公司 The mechanical people of surface treatment
CN108827974A (en) * 2018-06-28 2018-11-16 广东科达洁能股份有限公司 A kind of ceramic tile defect inspection method and system
CN108943179A (en) * 2018-08-22 2018-12-07 慧泉智能科技（苏州）有限公司 A kind of optimal cutling method for arithmetic for lumber surface defects'
CN108961239A (en) * 2018-07-02 2018-12-07 北京百度网讯科技有限公司 Continuous casting billet quality detection method, device, electronic equipment and storage medium
CN108985337A (en) * 2018-06-20 2018-12-11 中科院广州电子技术有限公司 A kind of product surface scratch detection method based on picture depth study
CN109060832A (en) * 2018-07-05 2018-12-21 上海徕木电子股份有限公司 A kind of electric power connector contact pin defective workmanship visible detection method
CN109063713A (en) * 2018-07-20 2018-12-21 中国林业科学研究院木材工业研究所 A kind of timber discrimination method and system based on the study of construction feature picture depth
CN109087281A (en) * 2018-07-02 2018-12-25 北京百度网讯科技有限公司 Display screen peripheral circuit detection method, device, electronic equipment and storage medium
CN109100371A (en) * 2018-10-25 2018-12-28 广东华中科技大学工业技术研究院 A kind of the 3C transparent component defect detecting device and method of laser total reflection formula
CN109145955A (en) * 2018-07-26 2019-01-04 中国林业科学研究院木材工业研究所 A kind of Wood Identification Method and system
CN109142371A (en) * 2018-07-31 2019-01-04 华南理工大学 High density flexible exterior substrate defect detecting system and method based on deep learning
CN109249300A (en) * 2018-10-25 2019-01-22 北京木业邦科技有限公司 Timber processing control method, device, electronic equipment and storage medium
CN109454195A (en) * 2018-10-31 2019-03-12 安徽巨自动化装备有限公司 A kind of aluminium vehicle body self-piercing riveting riveting mould failure visible detection method
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
CN109724543A (en) * 2018-12-29 2019-05-07 南京林业大学 A kind of method of Fast Evaluation wood milling processing performance
CN109829483A (en) * 2019-01-07 2019-05-31 鲁班嫡系机器人（深圳）有限公司 Defect recognition model training method, device, computer equipment and storage medium
CN109884066A (en) * 2019-01-22 2019-06-14 江苏恒力化纤股份有限公司 A kind of method of offline inspection long filament silk volume greasy dirt fault
CN109949286A (en) * 2019-03-12 2019-06-28 北京百度网讯科技有限公司 Method and apparatus for output information
CN109959661A (en) * 2017-12-25 2019-07-02 由田新技股份有限公司 Automatic optical detection method, equipment and its deep learning system
CN109978867A (en) * 2019-03-29 2019-07-05 北京百度网讯科技有限公司 Toy appearance quality determining method and its relevant device
CN109978868A (en) * 2019-03-29 2019-07-05 北京百度网讯科技有限公司 Toy appearance quality determining method and its relevant device
CN110009617A (en) * 2019-04-02 2019-07-12 浙江大学 A kind of shaft forgings online quality control method
CN110059549A (en) * 2019-03-11 2019-07-26 齐鲁工业大学 A kind of thin wood plate categorizing system and algorithm based on deep learning
CN110245642A (en) * 2019-06-21 2019-09-17 北京市市政工程研究院 A kind of radar spectrum recognition method and system based on deep learning
CN110310259A (en) * 2019-06-19 2019-10-08 江南大学 It is a kind of that flaw detection method is tied based on the wood for improving YOLOv3 algorithm
CN110414140A (en) * 2019-07-30 2019-11-05 同济大学 The model training and test method of wood damage, device, equipment and storage medium
CN110458814A (en) * 2019-07-29 2019-11-15 山东艾雷维特智能科技有限公司 A kind of detection method of wood single-plate defect, apparatus and system
CN110458791A (en) * 2018-05-04 2019-11-15 深圳市圆周率软件科技有限责任公司 Mass defect detection method and detection device
CN110689539A (en) * 2019-11-12 2020-01-14 南京大学 Workpiece surface defect detection method based on deep learning
CN110738630A (en) * 2018-07-02 2020-01-31 由田新技股份有限公司 Training method and detection system of recursive deep learning system
CN110930383A (en) * 2019-11-20 2020-03-27 佛山市南海区广工大数控装备协同创新研究院 Injector defect detection method based on deep learning semantic segmentation and image classification
CN111024729A (en) * 2019-12-27 2020-04-17 无锡动视宫原科技有限公司 Method and system for detecting slight indentation on surface of polaroid
CN111036578A (en) * 2019-12-27 2020-04-21 无锡动视宫原科技有限公司 Detection method and detection system for fine concave-convex points on surface of polarizer
CN111079628A (en) * 2019-12-12 2020-04-28 哈尔滨市科佳通用机电股份有限公司 Floor damage fault image identification method
CN111091551A (en) * 2019-12-12 2020-05-01 哈尔滨市科佳通用机电股份有限公司 Method for detecting loss fault of brake beam strut opening pin of railway wagon
CN111091550A (en) * 2019-12-12 2020-05-01 创新奇智(北京)科技有限公司 Multi-size self-adaptive PCB solder paste area detection system and detection method
CN111161246A (en) * 2019-12-30 2020-05-15 歌尔股份有限公司 Product defect detection method, device and system
CN111179263A (en) * 2020-01-06 2020-05-19 宜通世纪物联网研究院(广州)有限公司 Industrial image surface defect detection model, method, system and device
CN111300144A (en) * 2019-11-25 2020-06-19 上海大学 Automatic detection method for tool wear state based on image processing
CN111325713A (en) * 2020-01-21 2020-06-23 浙江省北大信息技术高等研究院 Wood defect detection method, system and storage medium based on neural network
CN111428796A (en) * 2020-03-27 2020-07-17 上海媒智科技有限公司 General article detection method and system based on deep learning
CN111563455A (en) * 2020-05-08 2020-08-21 南昌工程学院 Damage identification method based on time series signal and compressed convolution neural network
US10769774B2 (en) 2018-01-09 2020-09-08 Beijing Baidu Netcom Science And Technology Co., Ltd. Method and device for detecting a defect in a steel plate, as well as apparatus and server therefor
CN111709909A (en) * 2020-05-12 2020-09-25 安徽艾睿思智能科技有限公司 General printing defect detection method based on deep learning and model thereof
CN111751253A (en) * 2020-07-06 2020-10-09 重庆理工大学 Forming method and quality detection method of concrete aggregate detection model
CN111833287A (en) * 2019-04-11 2020-10-27 陕西坤同半导体科技有限公司 System for judging defects of display panel by applying deep learning and judging method thereof
WO2021042367A1 (en) * 2019-09-06 2021-03-11 Intel Corporation Deep learning based distributed machine vision camera system
CN112668409A (en) * 2020-12-14 2021-04-16 合肥富煌君达高科信息技术有限公司 Visual measurement system and method for identifying medicine type by using same
CN112697814A (en) * 2020-12-15 2021-04-23 电子科技大学 Cable surface defect detection system and method based on machine vision
CN112734690A (en) * 2020-12-17 2021-04-30 郑州金惠计算机系统工程有限公司 Surface defect detection method and device and computer readable storage medium
CN113077430A (en) * 2021-03-30 2021-07-06 太原理工大学 Laser chip defect detection and classification method and system based on SSD algorithm
CN113155851A (en) * 2021-04-30 2021-07-23 西安交通大学 Copper-clad plate surface defect visual online detection method and device based on deep learning
CN113240647A (en) * 2021-05-18 2021-08-10 哈尔滨工业大学（深圳） Mobile phone shell rear cover defect detection method and system based on deep learning
CN113379747A (en) * 2021-08-16 2021-09-10 深圳市信润富联数字科技有限公司 Wood defect detection method, electronic device, device and storage medium
CN113409290A (en) * 2021-06-29 2021-09-17 北京兆维电子(集团)有限责任公司 Method and device for detecting appearance defects of liquid crystal display and storage medium
CN113935415A (en) * 2021-10-14 2022-01-14 上海上实龙创智能科技股份有限公司 Quality detection model training method and device, electronic equipment and storage medium
CN114332083A (en) * 2022-03-09 2022-04-12 齐鲁工业大学 PFNet-based industrial product camouflage flaw identification method
US11315231B2 (en) 2018-06-08 2022-04-26 Industrial Technology Research Institute Industrial image inspection method and system and computer readable recording medium
CN116309375A (en) * 2023-02-23 2023-06-23 南京林业大学 Method for detecting double-sided defects of solid wood plate and determining intelligent processing coordinates
CN116559119A (en) * 2023-05-11 2023-08-08 东北林业大学 Deep learning-based wood dyeing color difference detection method, system and medium
CN117152161A (en) * 2023-11-01 2023-12-01 山东迪特智联信息科技有限责任公司 Shaving board quality detection method and system based on image recognition
CN117635619A (en) * 2024-01-26 2024-03-01 南京海关工业产品检测中心 Log volume detection method and system based on machine vision

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105590319A (en) * 2015-12-18 2016-05-18 华南理工大学 Method for detecting image saliency region for deep learning
US20160307072A1 (en) * 2015-04-17 2016-10-20 Nec Laboratories America, Inc. Fine-grained Image Classification by Exploring Bipartite-Graph Labels
CN106127780A (en) * 2016-06-28 2016-11-16 华南理工大学 A kind of curved surface defect automatic testing method and device thereof

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160307072A1 (en) * 2015-04-17 2016-10-20 Nec Laboratories America, Inc. Fine-grained Image Classification by Exploring Bipartite-Graph Labels
CN105590319A (en) * 2015-12-18 2016-05-18 华南理工大学 Method for detecting image saliency region for deep learning
CN106127780A (en) * 2016-06-28 2016-11-16 华南理工大学 A kind of curved surface defect automatic testing method and device thereof

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
李梦园: ""深度学习算法在表面缺陷识别中的应用研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 *
梁海平: ""基于机器视觉的木材表面缺陷的在线检测技术研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 *

Cited By (104)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108362702A (en) * 2017-12-14 2018-08-03 北京木业邦科技有限公司 A kind of defect of veneer detection method, system and equipment based on artificial intelligence
CN108262809A (en) * 2017-12-14 2018-07-10 北京木业邦科技有限公司 Plank processing method, device, electronic equipment and medium based on artificial intelligence
CN108247764B (en) * 2017-12-14 2021-04-27 北京木业邦科技有限公司 Wood board cutting method and device based on machine learning, electronic equipment and medium
CN108375353B (en) * 2017-12-14 2024-04-19 北京木业邦科技有限公司 Wood detection method, device and equipment
CN108262809B (en) * 2017-12-14 2021-06-29 北京木业邦科技有限公司 Wood board processing method and device based on artificial intelligence, electronic equipment and medium
CN108247764A (en) * 2017-12-14 2018-07-06 北京木业邦科技有限公司 Plank cutting method, device, electronic equipment and medium based on machine learning
CN108375353A (en) * 2017-12-14 2018-08-07 北京木业邦科技有限公司 A kind of log detection method, apparatus and equipment
CN108074231A (en) * 2017-12-18 2018-05-25 浙江工业大学 Magnetic sheet surface defect detection method based on convolutional neural network
CN108038853B (en) * 2017-12-18 2020-05-26 浙江工业大学 Ceramic tile surface defect identification method based on convolutional neural network and active learning
CN108038853A (en) * 2017-12-18 2018-05-15 浙江工业大学 Ceramic tile surface defect identification method based on convolutional neural network and active learning
CN108074231B (en) * 2017-12-18 2020-04-21 浙江工业大学 Magnetic sheet surface defect detection method based on convolutional neural network
CN109959661A (en) * 2017-12-25 2019-07-02 由田新技股份有限公司 Automatic optical detection method, equipment and its deep learning system
CN108154504A (en) * 2017-12-25 2018-06-12 浙江工业大学 Method for detecting surface defects of steel plate based on convolutional neural network
CN108789426A (en) * 2017-12-29 2018-11-13 金门工程建设有限公司 The mechanical people of surface treatment
CN108230318A (en) * 2018-01-09 2018-06-29 北京百度网讯科技有限公司 Ladle defects detection sorting technique, device, equipment and computer-readable medium
CN108230318B (en) * 2018-01-09 2019-04-16 北京百度网讯科技有限公司 Ladle defects detection classification method, device, equipment and computer-readable medium
US10769774B2 (en) 2018-01-09 2020-09-08 Beijing Baidu Netcom Science And Technology Co., Ltd. Method and device for detecting a defect in a steel plate, as well as apparatus and server therefor
CN108230317A (en) * 2018-01-09 2018-06-29 北京百度网讯科技有限公司 Steel plate defect detection sorting technique, device, equipment and computer-readable medium
CN108389181A (en) * 2018-01-23 2018-08-10 南京大学 It is a kind of based on the part defect of deep learning and normal direction figure detection and localization method
CN108311411A (en) * 2018-02-07 2018-07-24 陕西科技大学 A kind of cordyceps sinensis intelligence sorting system and its application method
CN108491867A (en) * 2018-03-12 2018-09-04 苏州卓融新能源科技有限公司 Image Matching based on artificial intelligence and recognition methods
CN108346153A (en) * 2018-03-22 2018-07-31 北京木业邦科技有限公司 The machine learning of defects in timber and restorative procedure, device, system, electronic equipment
CN110458791B (en) * 2018-05-04 2023-06-06 圆周率科技(常州)有限公司 Quality defect detection method and detection equipment
CN110458791A (en) * 2018-05-04 2019-11-15 深圳市圆周率软件科技有限责任公司 Mass defect detection method and detection device
US11276158B2 (en) 2018-05-25 2022-03-15 Beijing Baidu Netcom Science And Technology Co., Ltd. Method and apparatus for inspecting corrosion defect of ladle
CN108776808A (en) * 2018-05-25 2018-11-09 北京百度网讯科技有限公司 A kind of method and apparatus for detecting ladle corrosion defect
US11315231B2 (en) 2018-06-08 2022-04-26 Industrial Technology Research Institute Industrial image inspection method and system and computer readable recording medium
CN108985337A (en) * 2018-06-20 2018-12-11 中科院广州电子技术有限公司 A kind of product surface scratch detection method based on picture depth study
CN108827974A (en) * 2018-06-28 2018-11-16 广东科达洁能股份有限公司 A kind of ceramic tile defect inspection method and system
CN108827974B (en) * 2018-06-28 2024-01-09 广东科达洁能股份有限公司 Ceramic tile defect detection method and system
CN108961239A (en) * 2018-07-02 2018-12-07 北京百度网讯科技有限公司 Continuous casting billet quality detection method, device, electronic equipment and storage medium
CN110738630A (en) * 2018-07-02 2020-01-31 由田新技股份有限公司 Training method and detection system of recursive deep learning system
CN109087281A (en) * 2018-07-02 2018-12-25 北京百度网讯科技有限公司 Display screen peripheral circuit detection method, device, electronic equipment and storage medium
CN109060832A (en) * 2018-07-05 2018-12-21 上海徕木电子股份有限公司 A kind of electric power connector contact pin defective workmanship visible detection method
CN109063713A (en) * 2018-07-20 2018-12-21 中国林业科学研究院木材工业研究所 A kind of timber discrimination method and system based on the study of construction feature picture depth
CN109145955A (en) * 2018-07-26 2019-01-04 中国林业科学研究院木材工业研究所 A kind of Wood Identification Method and system
CN109145955B (en) * 2018-07-26 2019-10-22 中国林业科学研究院木材工业研究所 A kind of Wood Identification Method and system
CN109142371A (en) * 2018-07-31 2019-01-04 华南理工大学 High density flexible exterior substrate defect detecting system and method based on deep learning
CN108943179A (en) * 2018-08-22 2018-12-07 慧泉智能科技（苏州）有限公司 A kind of optimal cutling method for arithmetic for lumber surface defects'
CN109100371A (en) * 2018-10-25 2018-12-28 广东华中科技大学工业技术研究院 A kind of the 3C transparent component defect detecting device and method of laser total reflection formula
CN109249300A (en) * 2018-10-25 2019-01-22 北京木业邦科技有限公司 Timber processing control method, device, electronic equipment and storage medium
WO2020087725A1 (en) * 2018-10-31 2020-05-07 安徽巨一自动化装备有限公司 Method for visually detecting failure of self-piercing riveting die in aluminum bodywork
CN109454195A (en) * 2018-10-31 2019-03-12 安徽巨自动化装备有限公司 A kind of aluminium vehicle body self-piercing riveting riveting mould failure visible detection method
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
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
CN109724543A (en) * 2018-12-29 2019-05-07 南京林业大学 A kind of method of Fast Evaluation wood milling processing performance
CN109829483B (en) * 2019-01-07 2021-05-18 鲁班嫡系机器人（深圳）有限公司 Defect recognition model training method and device, computer equipment and storage medium
CN109829483A (en) * 2019-01-07 2019-05-31 鲁班嫡系机器人（深圳）有限公司 Defect recognition model training method, device, computer equipment and storage medium
CN109884066A (en) * 2019-01-22 2019-06-14 江苏恒力化纤股份有限公司 A kind of method of offline inspection long filament silk volume greasy dirt fault
CN110059549A (en) * 2019-03-11 2019-07-26 齐鲁工业大学 A kind of thin wood plate categorizing system and algorithm based on deep learning
CN109949286A (en) * 2019-03-12 2019-06-28 北京百度网讯科技有限公司 Method and apparatus for output information
US11341626B2 (en) 2019-03-12 2022-05-24 Beijing Baidu Netcom Science And Technology Co., Ltd. Method and apparatus for outputting information
CN109978867A (en) * 2019-03-29 2019-07-05 北京百度网讯科技有限公司 Toy appearance quality determining method and its relevant device
CN109978868A (en) * 2019-03-29 2019-07-05 北京百度网讯科技有限公司 Toy appearance quality determining method and its relevant device
CN110009617A (en) * 2019-04-02 2019-07-12 浙江大学 A kind of shaft forgings online quality control method
CN111833287A (en) * 2019-04-11 2020-10-27 陕西坤同半导体科技有限公司 System for judging defects of display panel by applying deep learning and judging method thereof
CN110310259B (en) * 2019-06-19 2021-07-27 江南大学 Improved YOLOv3 algorithm-based knot defect detection method
CN110310259A (en) * 2019-06-19 2019-10-08 江南大学 It is a kind of that flaw detection method is tied based on the wood for improving YOLOv3 algorithm
CN110245642A (en) * 2019-06-21 2019-09-17 北京市市政工程研究院 A kind of radar spectrum recognition method and system based on deep learning
CN110458814A (en) * 2019-07-29 2019-11-15 山东艾雷维特智能科技有限公司 A kind of detection method of wood single-plate defect, apparatus and system
CN110414140A (en) * 2019-07-30 2019-11-05 同济大学 The model training and test method of wood damage, device, equipment and storage medium
WO2021042367A1 (en) * 2019-09-06 2021-03-11 Intel Corporation Deep learning based distributed machine vision camera system
CN110689539A (en) * 2019-11-12 2020-01-14 南京大学 Workpiece surface defect detection method based on deep learning
CN110930383A (en) * 2019-11-20 2020-03-27 佛山市南海区广工大数控装备协同创新研究院 Injector defect detection method based on deep learning semantic segmentation and image classification
CN111300144A (en) * 2019-11-25 2020-06-19 上海大学 Automatic detection method for tool wear state based on image processing
CN111091551A (en) * 2019-12-12 2020-05-01 哈尔滨市科佳通用机电股份有限公司 Method for detecting loss fault of brake beam strut opening pin of railway wagon
CN111079628A (en) * 2019-12-12 2020-04-28 哈尔滨市科佳通用机电股份有限公司 Floor damage fault image identification method
CN111091550A (en) * 2019-12-12 2020-05-01 创新奇智(北京)科技有限公司 Multi-size self-adaptive PCB solder paste area detection system and detection method
CN111036578A (en) * 2019-12-27 2020-04-21 无锡动视宫原科技有限公司 Detection method and detection system for fine concave-convex points on surface of polarizer
CN111024729A (en) * 2019-12-27 2020-04-17 无锡动视宫原科技有限公司 Method and system for detecting slight indentation on surface of polaroid
CN111161246B (en) * 2019-12-30 2024-05-14 歌尔股份有限公司 Product defect detection method, device and system
CN111161246A (en) * 2019-12-30 2020-05-15 歌尔股份有限公司 Product defect detection method, device and system
CN111179263A (en) * 2020-01-06 2020-05-19 宜通世纪物联网研究院(广州)有限公司 Industrial image surface defect detection model, method, system and device
CN111179263B (en) * 2020-01-06 2023-10-13 广东宜通联云智能信息有限公司 Industrial image surface defect detection model, method, system and device
CN111325713A (en) * 2020-01-21 2020-06-23 浙江省北大信息技术高等研究院 Wood defect detection method, system and storage medium based on neural network
CN111428796A (en) * 2020-03-27 2020-07-17 上海媒智科技有限公司 General article detection method and system based on deep learning
CN111428796B (en) * 2020-03-27 2024-04-02 上海媒智科技有限公司 General object detection method and system based on deep learning
CN111563455A (en) * 2020-05-08 2020-08-21 南昌工程学院 Damage identification method based on time series signal and compressed convolution neural network
CN111709909B (en) * 2020-05-12 2024-02-20 苏州科亿信息科技有限公司 General printing defect detection method based on deep learning and model thereof
CN111709909A (en) * 2020-05-12 2020-09-25 安徽艾睿思智能科技有限公司 General printing defect detection method based on deep learning and model thereof
CN111751253A (en) * 2020-07-06 2020-10-09 重庆理工大学 Forming method and quality detection method of concrete aggregate detection model
CN111751253B (en) * 2020-07-06 2022-10-14 重庆理工大学 Forming method and quality detection method of concrete aggregate detection model
CN112668409A (en) * 2020-12-14 2021-04-16 合肥富煌君达高科信息技术有限公司 Visual measurement system and method for identifying medicine type by using same
CN112697814B (en) * 2020-12-15 2022-05-17 电子科技大学 Cable surface defect detection system and method based on machine vision
CN112697814A (en) * 2020-12-15 2021-04-23 电子科技大学 Cable surface defect detection system and method based on machine vision
CN112734690A (en) * 2020-12-17 2021-04-30 郑州金惠计算机系统工程有限公司 Surface defect detection method and device and computer readable storage medium
CN113077430A (en) * 2021-03-30 2021-07-06 太原理工大学 Laser chip defect detection and classification method and system based on SSD algorithm
CN113077430B (en) * 2021-03-30 2023-05-09 太原理工大学 SSD algorithm-based laser chip defect detection and classification method and system
CN113155851A (en) * 2021-04-30 2021-07-23 西安交通大学 Copper-clad plate surface defect visual online detection method and device based on deep learning
CN113240647A (en) * 2021-05-18 2021-08-10 哈尔滨工业大学（深圳） Mobile phone shell rear cover defect detection method and system based on deep learning
CN113409290B (en) * 2021-06-29 2023-12-15 北京兆维电子(集团)有限责任公司 Method and device for detecting appearance defects of liquid crystal display, and storage medium
CN113409290A (en) * 2021-06-29 2021-09-17 北京兆维电子(集团)有限责任公司 Method and device for detecting appearance defects of liquid crystal display and storage medium
CN113379747B (en) * 2021-08-16 2021-12-21 深圳市信润富联数字科技有限公司 Wood defect detection method, electronic device, device and storage medium
CN113379747A (en) * 2021-08-16 2021-09-10 深圳市信润富联数字科技有限公司 Wood defect detection method, electronic device, device and storage medium
CN113935415A (en) * 2021-10-14 2022-01-14 上海上实龙创智能科技股份有限公司 Quality detection model training method and device, electronic equipment and storage medium
CN114332083A (en) * 2022-03-09 2022-04-12 齐鲁工业大学 PFNet-based industrial product camouflage flaw identification method
CN116309375A (en) * 2023-02-23 2023-06-23 南京林业大学 Method for detecting double-sided defects of solid wood plate and determining intelligent processing coordinates
CN116309375B (en) * 2023-02-23 2023-10-24 南京林业大学 Method for detecting double-sided defects of solid wood plate and determining intelligent processing coordinates
CN116559119A (en) * 2023-05-11 2023-08-08 东北林业大学 Deep learning-based wood dyeing color difference detection method, system and medium
CN116559119B (en) * 2023-05-11 2024-01-26 东北林业大学 Deep learning-based wood dyeing color difference detection method, system and medium
CN117152161A (en) * 2023-11-01 2023-12-01 山东迪特智联信息科技有限责任公司 Shaving board quality detection method and system based on image recognition
CN117152161B (en) * 2023-11-01 2024-03-01 山东迪特智联信息科技有限责任公司 Shaving board quality detection method and system based on image recognition
CN117635619B (en) * 2024-01-26 2024-04-05 南京海关工业产品检测中心 Log volume detection method and system based on machine vision
CN117635619A (en) * 2024-01-26 2024-03-01 南京海关工业产品检测中心 Log volume detection method and system based on machine vision

Also Published As

Publication number Publication date
CN107392896B (en) 2019-11-08

Similar Documents

Publication Publication Date Title
CN107392896B (en) A kind of Wood Defects Testing method and system based on deep learning
CN108038853B (en) Ceramic tile surface defect identification method based on convolutional neural network and active learning
CN110992317B (en) PCB defect detection method based on semantic segmentation
CN110274908A (en) Flaw detection apparatus, defect detecting method and computer readable recording medium
CN107123111B (en) Deep residual error network construction method for mobile phone screen defect detection
CN107123114A (en) A kind of cloth defect inspection method and device based on machine learning
CN111311544B (en) Floor defect detection method based on deep learning
CN108562589A (en) A method of magnetic circuit material surface defect is detected
CN108229665A (en) A kind of the System of Sorting Components based on the convolutional neural networks by depth
CN110310259A (en) It is a kind of that flaw detection method is tied based on the wood for improving YOLOv3 algorithm
CN107016406A (en) The pest and disease damage image generating method of network is resisted based on production
CN107862668A (en) A kind of cultural relic images restored method based on GNN
CN106875373A (en) Mobile phone screen MURA defect inspection methods based on convolutional neural networks pruning algorithms
CN113160123B (en) Leather defect detection method, system and device based on YOLOv5
CN113034483B (en) Cigarette defect detection method based on deep migration learning
CN110060251A (en) A kind of building surface crack detecting method based on U-Net
CN113627435B (en) Defect detection and identification method and system for ceramic tile
CN112967271B (en) Casting surface defect identification method based on improved DeepLabv3+ network model
CN112150417A (en) Coating defect detection method based on deep learning
CN107194432B (en) Refrigerator door body identification method and system based on deep convolutional neural network
CN114529484B (en) Deep learning sample enhancement method for direct current component change in imaging
Meng et al. X-DenseNet: deep learning for garbage classification based on visual images
CN107742132A (en) Potato detection method of surface flaw based on convolutional neural networks
CN112070712A (en) Printing defect detection method based on self-encoder network
CN106323985B (en) Solid wood board quality detection method combining computer vision with self-learning behavior

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