CN116309343A - Defect detection method and device based on deep learning and storage medium - Google Patents
Defect detection method and device based on deep learning and storage medium Download PDFInfo
- Publication number
- CN116309343A CN116309343A CN202310110150.7A CN202310110150A CN116309343A CN 116309343 A CN116309343 A CN 116309343A CN 202310110150 A CN202310110150 A CN 202310110150A CN 116309343 A CN116309343 A CN 116309343A
- Authority
- CN
- China
- Prior art keywords
- defect
- deep learning
- product
- image
- file
- 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.)
- Withdrawn
Links
- 230000007547 defect Effects 0.000 title claims abstract description 160
- 238000001514 detection method Methods 0.000 title claims abstract description 46
- 238000013135 deep learning Methods 0.000 title claims abstract description 37
- 238000012549 training Methods 0.000 claims abstract description 27
- 238000000034 method Methods 0.000 claims abstract description 18
- 238000002372 labelling Methods 0.000 claims abstract description 11
- 238000003062 neural network model Methods 0.000 claims abstract description 8
- 239000000047 product Substances 0.000 claims description 49
- 238000013527 convolutional neural network Methods 0.000 claims description 19
- 230000003287 optical effect Effects 0.000 claims description 12
- 230000006870 function Effects 0.000 claims description 11
- 230000004913 activation Effects 0.000 claims description 3
- 239000007795 chemical reaction product Substances 0.000 claims description 3
- 238000004590 computer program Methods 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000011176 pooling Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 4
- 238000007789 sealing Methods 0.000 description 5
- 230000037303 wrinkles Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 239000012535 impurity Substances 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 239000002699 waste material Substances 0.000 description 3
- -1 broken filaments Substances 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
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
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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
- G06T2207/30164—Workpiece; Machine component
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Multimedia (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a defect detection method, device and storage medium based on deep learning, which comprises the following specific steps: step one: collecting product images with defect information and classifying the product images according to defect types; step two: labeling the defect image and generating a label file with defect information; step three: leading the label file and the training image into a deep learning neural network model, and training to obtain a model file; step four: and loading the model file and transmitting the product image on line, so that various defects of the product can be accurately identified. The method is suitable for defect detection, solves the problem that the accuracy and the efficiency of manual detection in the prior art are low, and achieves the effects of automatically detecting according to the obtained product pictures and further improving the detection accuracy and the efficiency.
Description
Technical Field
The present invention relates to the field of product quality detection technologies, and in particular, to a method, an apparatus, and a storage medium for detecting defects based on deep learning.
Background
Along with the rapid iteration of 3C electronic products and new energy automobile parts, the appearance detection requirements of the products are increased, the yield requirements are difficult to meet by traditional intensive manual detection, meanwhile, the accuracy of product defect detection is difficult to ensure by manual detection under high-pressure and long-time operation environments, and automatic appearance detection equipment is urgently needed to replace manual detection. However, the conventional detection method on the existing automatic detection equipment is poor in generalization, and once new defect characteristics appear after product iteration, new method programs and even rewriting methods are needed to be added, so that the robustness is poor.
In order to meet the appearance detection requirements of various products, a set of appearance detection methods compatible with multiple products and multiple defects needs to be developed, the defect detection technology based on deep learning is mature day by day, the defects which cannot be identified by the conventional method can be effectively overcome, and meanwhile, the technology has the characteristics of strong universality and good robustness and gradually becomes a mainstream method for solving the difficult points of complex defect detection.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provides a defect detection method, device and storage medium based on deep learning.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
the invention discloses a defect detection method, device and storage medium based on deep learning, which comprises the following specific steps:
step one: collecting product images with defect information and classifying the product images according to defect types;
step two: labeling the defect image and generating a label file with defect information;
step three: leading the label file and the training image into a deep learning neural network model, and training to obtain a model file;
step four: and loading the model file and transmitting the product image on line, so that various defects of the product can be accurately identified.
Preferably, in the first step, the product images with defect information are collected and classified according to defect types, and the specific steps are as follows:
collecting various defect images of the target product through an optical device, wherein the optical device can meet clear collection of various defects, and each defect can collect a certain number of images;
classifying the acquired defect images according to defect types, and respectively screening out proper defect images aiming at defects of different types, wherein the proper defect images are used for objectively reflecting details of various defects.
Preferably, the marking of the defect image in the second step and the generation of the tag file with defect information are performed, and the specific steps are as follows:
and labeling the various defect images by using a LabelImg labeling tool to generate an xml tag file with defect information, wherein the defect information comprises the tag name of the defect, the pixel position of the defect, the size of the defect image and the channel number.
Preferably, the step three is to import the tag file and the training image into a deep learning neural network model, train to obtain a model file, and specifically comprises the following steps:
building a convolutional neural network model through an open-source deep learning framework, and constructing the convolutional neural network model with a convolutional layer, a pooling layer, an activation function and a full-connection layer structure;
and importing the images of all the defect samples and the corresponding label files into a constructed convolutional neural network model, and training to obtain a model file.
Preferably, the images of all the defect samples and the corresponding label files are imported into a constructed convolutional neural network model, and the model files can be obtained after training, and the specific steps are as follows:
the method comprises the steps of importing an image of a defect sample and a label file corresponding to the image into a constructed convolutional neural network model, adding a data enhancement algorithm to the input end of the convolutional neural network for enriching the defect sample and improving the robustness of a training network, and splicing the defect sample in a random scaling, random cutting and random arrangement mode so as to achieve the aim of enhancing a data set;
the Backbone network of the convolutional neural network is responsible for extracting characteristic information of an input image, and each defect map contains true values of defects: characteristic information and location; meanwhile, dividing each input image into areas, distributing initialized multi-scale prediction frames in the areas, wherein part of the prediction frames contain predicted values of defects, constructing a loss function through the true values and the predicted values of the defects, and converting the defect detection problem into a regression problem;
the learning rate self-adaptive optimization algorithm is added to achieve that the loss function value is close to 0 rapidly, so that the predicted value is close to the true value rapidly and accurately, and a large number of node parameter values generated in the model training process form a model file.
Preferably, in the fourth step, loading a model file and transmitting a product image on line, so as to accurately identify various defects of the product, and the specific steps are as follows:
the camera is controlled to acquire the product image on line through the motion module, the optical device is designed to be capable of clearly and comprehensively acquiring the area to be detected, and the details of the defects of the reaction product are clearly shown;
the defect type and the position on the product can be accurately identified through loading the trained model file and the product image acquired on line and through the feature extraction and regression algorithm of deep learning.
In a second aspect of the embodiment of the present invention, a defect detection device based on deep learning is provided, where the device includes a memory and a processor, where at least one program instruction is stored in the memory, and the processor loads and executes the at least one program instruction to implement defect detection.
In a third aspect of embodiments of the present invention, there is provided a readable storage medium for deep learning based defect detection, the readable storage medium having stored therein a computer program for enabling defect detection when executed by a processor.
Compared with the prior art, the invention has the following beneficial effects:
by collecting the product images with defect information and classifying according to defect types; labeling the defect image and generating a label file with defect information; leading the label file and the training image into a deep learning neural network model, and training to obtain a model file; and loading the model file and transmitting the product image on line, so that various defects of the product can be accurately identified. The problem of the manual detection's in the prior art rate of accuracy and efficiency are all lower is solved, the effect that can be according to the product picture automated inspection who obtains has been reached, and then detection accuracy and efficiency are improved.
Drawings
FIG. 1 is a flow chart of a method for detecting defects based on deep learning according to an embodiment of the present invention;
FIG. 2 is a representation of bubble, impurity, hairline, scrap residue, and wrinkle defects of a 3C electronic watch seal ring according to one embodiment of the present invention;
FIG. 3 is a schematic diagram of an xml tag file with defect information generated and marking a defect of a product according to one embodiment of the present invention;
FIG. 4 is a schematic diagram showing the relationship between a predicted frame and a real frame of a defect of a product according to an embodiment of the present invention;
fig. 5 is a diagram showing the identification result of bubbles, impurities, broken filaments, residue, and wrinkles of a 3C electronic watch sealing rubber ring according to an embodiment of the present invention.
Detailed Description
Specific embodiments of a deep learning-based defect detection method, apparatus and storage medium according to the present invention are further described below with reference to fig. 1 to 5. The deep learning-based defect detection method, apparatus and storage medium of the present invention are not limited to the descriptions of the following embodiments.
Referring to fig. 1, a method flowchart of a defect detection method based on deep learning according to an embodiment of the present application is shown, and as shown in fig. 1, the steps are as follows:
step one, collecting product images with defect information and classifying the product images according to defect types, wherein the specific steps are as follows:
collecting various defect images of a sealing rubber ring of the 3C electronic watch through an optical device;
the optical device adopts a 2000 ten thousand pixel color area array camera with MV-CE200-10 UC. In addition, in practical implementation, in order to improve the quality of the acquired product pictures, an industrial telecentric lens with the model of DTCM111-80H-AL and a light source matched with the lens can be used, wherein the light source comprises coaxial light, annular light and backlight.
For example, please refer to fig. 2, which shows one possible schematic diagram of an optical system used in the present application. The optical system shown in fig. 2 is stable and reliable, and the optical device should be able to meet clear collection of various defects, each of which should collect a certain number of images. In the figure: 1 is a camera component, 2 is a lens component, 3 is a coaxial light source, 4 is a 75-degree annular light source, 5 is a product and a carrier, and 6 is a backlight.
Classifying the acquired defect images according to the defect types.
Taking a sealing rubber ring of a 3C electronic watch as an example, the defect types of the sealing rubber ring include bubbles, impurities, broken filaments, waste residues and wrinkles, and proper defect images are respectively screened out according to different types of defects, wherein the proper defect images should objectively reflect the details of various defects. Referring to fig. 3, there is shown an image of bubbles, foreign matter, hairlines, residue of waste, and wrinkles of a 3C electronic watch seal ring.
Marking the defect image and generating a label file with defect information;
the method comprises the following steps:
and labeling various defect images by using a LabelImg labeling tool to generate an xml label file with defect information, wherein the defect information comprises the label name of the defect, the pixel position of the defect, the size and the channel number of the defect image and the like. Referring to FIG. 4, an xml tag file is shown that marks a defect in a product and that is generated with defect information.
Step three, importing the label file and the training image into a deep learning neural network model, and training to obtain a model file;
building a convolutional neural network model through an open-source deep learning framework, and constructing the convolutional neural network model with structures such as a convolutional layer, a pooling layer, an activation function, a full-connection layer and the like;
the Backbone network of the convolutional neural network is responsible for extracting characteristic information of an input image, and each defect map contains true values of defects: characteristic information and location; meanwhile, each input image is divided into areas and initialized multi-scale prediction frames are distributed in the areas, part of the prediction frames contain predicted values of defects, a loss function total_loss is constructed through the true values and the predicted values of the defects, and the total_loss loss consists of three parts: the bounding box regression loss ciou_loss, confidence loss conf_loss, classification loss cls_loss, three loss functions are described below in connection with the examples:
1) The bounding box regresses the loss ciou_loss;
referring to FIG. 5, a relationship between a predicted frame and a real frame is shown, where iou represents the intersection of the predicted frame and the real frame divided by the union of the two, D 2 Represents the distance between the center points of the predicted frame and the real frame, D C And v represents a parameter for measuring the consistency of the aspect ratio of the prediction frame and the real frame, and is expressed as follows:
wherein w is gt 、h gt Respectively the width and the height of a real frame, w p 、h p The width and height of the prediction box, respectively.
2) Confidence loss conf_loss;
the formula is the binary cross entropy of confidence, wherein K multiplied by K is the number of grids of the image to be trained, M is the number of candidate frames in each grid, the anchor points of each candidate frame are the center points of the grids,representing that when the center of the real frame is within a certain grid, the +.>All are 1->Representing +.f. of a candidate box within a grid when the center of the real box is not within the grid>Are all 1, C i For the confidence target value of the real box, +.>Is a confidence predictor for the candidate box.
3) Classification loss cls_loss;
the formula is the binary cross entropy of the classification loss, in which,indicating that when the centre of the real frame is within the i grid, the grid is responsible for the prediction of the real frame class, then +.>p i (c) For the probability target value of the category to which the real frame belongs,a predicted probability value for the category for the mesh.
In summary, the total loss function total_loss is composed of bounding box regression loss ciou_loss, confidence loss conf_loss, and classification loss cls_loss:
total_loss=ciou_loss+conf_loss+cls_loss
the parameters in total_loss are continuously and iteratively optimized by an Adam optimizer, so that the total_loss can be quickly converged to 0, and a predicted value is quicker and more accurate to be close to a true value, namely, accurate positioning and classification of a defect target are conveniently realized in the training process;
in order to enrich the defect samples and improve the robustness of the training network, a data enhancement algorithm is added at the input end of the convolutional neural network, and the defect samples are spliced in a random scaling, random cutting and random arrangement mode, so that the aim of enhancing the data set is fulfilled;
and importing the images of all the defect samples and the corresponding label files into a constructed convolutional neural network model, and training to obtain a model file.
The learning rate self-adaptive optimization algorithm is added to achieve that the loss function value is close to 0 rapidly, so that the predicted value is close to the true value rapidly and accurately, and a large number of node parameter values generated in the iterative process of model training form a model file.
Loading a model file and transmitting a product image on line, so that various defects of the product can be accurately identified, and the method comprises the following specific steps:
the camera is controlled to acquire the product image on line through the motion module, the optical device is designed to be capable of clearly and comprehensively acquiring the area to be detected, and the details of the defects of the reaction product are clearly shown;
the defect type and position on the product can be accurately identified through loading the trained model file and the product image acquired on line by the algorithms of deep learning feature extraction, bounding box regression, classification and the like. Please refer to fig. 5, which illustrates the identification effect of bubbles, impurities, broken filaments, waste residues and wrinkles of the 3C electronic watch sealing rubber ring, and accurately identifies the position, confidence and category of each defect.
In summary, by collecting the product images with defect information and classifying according to defect types; labeling the defect image and generating a label file with defect information; leading the label file and the training image into a deep learning neural network model, and training to obtain a model file; and loading the model file and transmitting the product image on line, so that various defects of the product can be accurately identified. The problem of the manual detection's in the prior art rate of accuracy and efficiency are all lower is solved, the effect that can be according to the product picture automated inspection who obtains has been reached, and then detection accuracy and efficiency are improved.
The invention also provides a defect detection device based on deep learning, which comprises a memory and a processor, wherein at least one program instruction is stored in the memory, and the processor realizes defect detection by loading and executing the at least one program instruction.
The invention also provides a readable storage medium for deep learning-based defect detection, wherein the readable storage medium stores a computer program which is used for realizing the defect detection when being executed by a processor.
The foregoing has shown and described the basic principles and main features of the present invention and the advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made without departing from the spirit and scope of the invention, which is defined in the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (8)
1. A defect detection method based on deep learning is characterized in that: the method comprises the following specific steps:
step one: collecting product images with defect information and classifying the product images according to defect types;
step two: labeling the defect image and generating a label file with defect information;
step three: leading the label file and the training image into a deep learning neural network model, and training to obtain a model file;
step four: and loading the model file and transmitting the product image on line, so that various defects of the product can be accurately identified.
2. The deep learning-based defect detection method of claim 1, wherein: in the first step, product images with defect information are collected and classified according to defect types, and the specific steps are as follows:
collecting various defect images of the target product through an optical device, wherein the optical device can meet clear collection of various defects, and each defect can collect a certain number of images;
classifying the acquired defect images according to defect types, and respectively screening out proper defect images aiming at defects of different types, wherein the proper defect images are used for objectively reflecting details of various defects.
3. The deep learning-based defect detection method of claim 1, wherein: in the second step, the defect image is marked, and a label file with defect information is generated, and the specific steps are as follows:
and labeling the various defect images by using a LabelImg labeling tool to generate an xml tag file with defect information, wherein the defect information comprises the tag name of the defect, the pixel position of the defect, the size of the defect image and the channel number.
4. The deep learning-based defect detection method of claim 1, wherein: the label file and the training image are imported into a deep learning neural network model, and the model file is obtained through training, and the method specifically comprises the following steps:
building a convolutional neural network model through an open-source deep learning framework, and constructing the convolutional neural network model with a convolutional layer, a pooling layer, an activation function and a full-connection layer structure;
and importing the images of all the defect samples and the corresponding label files into a constructed convolutional neural network model, and training to obtain a model file.
5. The deep learning-based defect detection method of claim 4, wherein: the images of all the defect samples and the corresponding label files are imported into a constructed convolutional neural network model, and the model files can be obtained after training, and the specific steps are as follows:
the method comprises the steps of importing an image of a defect sample and a label file corresponding to the image into a constructed convolutional neural network model, adding a data enhancement algorithm to the input end of the convolutional neural network for enriching the defect sample and improving the robustness of a training network, and splicing the defect sample in a random scaling, random cutting and random arrangement mode so as to achieve the aim of enhancing a data set;
the Backbone network of the convolutional neural network is responsible for extracting characteristic information of an input image, and each defect map contains true values of defects: characteristic information and location; meanwhile, dividing each input image into areas, distributing initialized multi-scale prediction frames in the areas, wherein part of the prediction frames contain predicted values of defects, constructing a loss function through the true values and the predicted values of the defects, and converting the defect detection problem into a regression problem;
the learning rate self-adaptive optimization algorithm is added to achieve that the loss function value is close to 0 rapidly, so that the predicted value is close to the true value rapidly and accurately, and a large number of node parameter values generated in the model training process form a model file.
6. The deep learning-based defect detection method according to any one of claims 1 to 5, wherein: in the fourth step, loading a model file and transmitting a product image on line, so that various defects of the product can be accurately identified, and the specific steps are as follows:
the camera is controlled to acquire the product image on line through the motion module, the optical device is designed to be capable of clearly and comprehensively acquiring the area to be detected, and the details of the defects of the reaction product are clearly shown;
the defect type and the position on the product can be accurately identified through loading the trained model file and the product image acquired on line and through the feature extraction and regression algorithm of deep learning.
7. The deep learning-based defect detection apparatus of claim 1, wherein: the apparatus comprises a memory having stored therein at least one program instruction and a processor that implements the method of any of claims 1 to 6 by loading and executing the at least one program instruction.
8. The deep learning based defect detection readable storage medium of claim 1, wherein: the readable storage medium has stored therein a computer program for implementing the method of any of claims 1 to 6 when executed by a processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310110150.7A CN116309343A (en) | 2023-02-14 | 2023-02-14 | Defect detection method and device based on deep learning and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310110150.7A CN116309343A (en) | 2023-02-14 | 2023-02-14 | Defect detection method and device based on deep learning and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116309343A true CN116309343A (en) | 2023-06-23 |
Family
ID=86789746
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310110150.7A Withdrawn CN116309343A (en) | 2023-02-14 | 2023-02-14 | Defect detection method and device based on deep learning and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116309343A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117710310A (en) * | 2023-12-14 | 2024-03-15 | 上海千映智能科技有限公司 | System and method for detecting appearance defects of chip based on artificial intelligence |
-
2023
- 2023-02-14 CN CN202310110150.7A patent/CN116309343A/en not_active Withdrawn
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117710310A (en) * | 2023-12-14 | 2024-03-15 | 上海千映智能科技有限公司 | System and method for detecting appearance defects of chip based on artificial intelligence |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108537269B (en) | Weak interactive object detection deep learning method and system thereof | |
CN115829999A (en) | Insulator defect detection model generation method, device, equipment and storage medium | |
CN111767927A (en) | Lightweight license plate recognition method and system based on full convolution network | |
CN111444939A (en) | Small-scale equipment component detection method based on weak supervision cooperative learning in open scene of power field | |
CN113609896A (en) | Object-level remote sensing change detection method and system based on dual-correlation attention | |
CN113688665A (en) | Remote sensing image target detection method and system based on semi-supervised iterative learning | |
CN110796018A (en) | Hand motion recognition method based on depth image and color image | |
CN115439458A (en) | Industrial image defect target detection algorithm based on depth map attention | |
CN113344888A (en) | Surface defect detection method and device based on combined model | |
CN112651989A (en) | SEM image molecular sieve particle size statistical method and system based on Mask RCNN example segmentation | |
Chen et al. | Exchange means change: An unsupervised single-temporal change detection framework based on intra-and inter-image patch exchange | |
CN116309343A (en) | Defect detection method and device based on deep learning and storage medium | |
CN116597436A (en) | Method and device for recognizing characters of nameplate of switch cabinet of power distribution room | |
CN111696079A (en) | Surface defect detection method based on multi-task learning | |
CN114782967B (en) | Software defect prediction method based on code visual chemistry | |
CN116665054A (en) | Remote sensing image small target detection method based on improved YOLOv3 | |
CN115272252A (en) | Improved YOLOX-based carbon fiber defect detection method, device and system | |
CN117437647B (en) | Oracle character detection method based on deep learning and computer vision | |
CN114972246A (en) | Die-cutting product surface defect detection method based on deep learning | |
CN111612803B (en) | Vehicle image semantic segmentation method based on image definition | |
CN113139507A (en) | Automatic capturing method and system for drainage pipeline defect photos | |
CN117496426A (en) | Precast beam procedure identification method and device based on mutual learning | |
CN115909493A (en) | Teacher improper gesture detection method and system for classroom real-time recorded video | |
Das et al. | Object Detection on Scene Images: A Novel Approach | |
CN113688789B (en) | Online learning input degree identification method and system based on deep learning |
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 | ||
WW01 | Invention patent application withdrawn after publication | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20230623 |