CN109978844A - One kind being based on the modular intelligent analysis method of deep learning and system - Google Patents
One kind being based on the modular intelligent analysis method of deep learning and system Download PDFInfo
- Publication number
- CN109978844A CN109978844A CN201910200205.7A CN201910200205A CN109978844A CN 109978844 A CN109978844 A CN 109978844A CN 201910200205 A CN201910200205 A CN 201910200205A CN 109978844 A CN109978844 A CN 109978844A
- Authority
- CN
- China
- Prior art keywords
- target
- deep learning
- image data
- intelligent analysis
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Abstract
The present invention provides one kind based on the modular intelligent analysis method of deep learning, obtains image data from camera or local picture；It is searched in image data according to known target template and obtains target data；Target defect is further tested and analyzed in image data according to target data, or directly tests and analyzes target defect in image data, and the present invention also provides one kind to be based on the modular intelligent analysis system of deep learning, improves detection efficiency.
Description
Technical field
The present invention relates to one kind to be based on the modular intelligent analysis method of deep learning and system.
Background technique
NI Vision Builder for Automated Inspection refers to that will be ingested targeted transformation by camera is data image signal, sends image procossing to
System obtains the form result information of target, such as color, shape, position etc., and then controls scene according to the result of differentiation
Equipment makes action response.Domestic field of machine vision is in the stage of flourishing, and applies in all trades and professions, with all trades and professions
Higher and higher to product quality requirement, the testing requirements of industrial production line are also constantly promoted.Conventional machines vision technique is increasingly
It is unable to satisfy the requirement of high-quality.New technology based on deep learning gradually replaces traditional technology, bears complicated, high-precision
Detection task.Therefore, the convenient application based on deep learning and friendly interaction just there is an urgent need to.
Summary of the invention
The technical problem to be solved in the present invention, be to provide it is a kind of based on the modular intelligent analysis method of deep learning with
And system, promote detection efficiency.
One of present invention is achieved in that a kind of based on the modular intelligent analysis method of deep learning, comprising:
Step 1 obtains image data from camera or local picture；
Step 2 searches in image data according to known target template and obtains target data；
Step 3 further tests and analyzes target defect according to target data in image data, or directly in picture number
According to middle detection and analysis target defect.
Further, further include step 4, carry out target classification in image data, or carry out mesh in target data
Mark classification, or target classification is carried out in target defect.
Further, the target classification is Inception the VGGNet network model based on deep learning come real
The classification of existing target.
Further, the step 2 is further specifically: using geometric match, template matching or based on deep learning
Faster-R-CNN, SSD or YOLO algorithm search in image data the similar target with target template, obtain number of targets
According to.
Further, the step 3 is further specifically: using YOLO the or SSD algorithm based on deep learning and basis
Target data further tests and analyzes target defect in image data, or YOLO or SSD based on deep learning is used to calculate
Method directly tests and analyzes target defect in image data.
The two of the present invention are achieved in that a kind of based on the modular intelligent analysis system of deep learning, comprising:
Image collection module obtains image data from camera or local picture；
Locating module is searched in image data according to known target template and obtains target data；
Detection module further tests and analyzes target defect according to target data in image data, or is directly scheming
As testing and analyzing target defect in data.
Further, further include categorization module, target classification is carried out in image data, or carried out in target data
Target classification, or target classification is carried out in target defect.
Further, the target classification is Inception the VGGNet network model based on deep learning come real
The classification of existing target.
Further, the locating module is further specifically: using geometric match, template matching or is based on deep learning
Faster-R-CNN, SSD or YOLO algorithm search in image data the similar target with target template, obtain number of targets
According to.
Further, the detection module is further specifically: using YOLO the or SSD algorithm based on deep learning and root
Target defect is further tested and analyzed in image data according to target data, or uses YOLO or SSD based on deep learning
Algorithm directly tests and analyzes target defect in image data.
The present invention has the advantage that a kind of be based on the modular intelligent analysis method of deep learning and system, the party
Method can be used in an editing machine editing independent module, and intuitively preview testing process, promotion can detect efficiency of the practice.
Detailed description of the invention
The present invention is further illustrated in conjunction with the embodiments with reference to the accompanying drawings.
Fig. 1 is the method for the present invention execution flow chart.
Specific embodiment
The present invention is based on the modular intelligent analysis methods of deep learning, comprising:
Step 1 obtains image data from camera or local picture；
Step 2 is calculated using geometric match, template matching or Faster-R-CNN, SSD or YOLO based on deep learning
Method searches in image data the similar target with target template, obtains target data；
Step 3, using YOLO the or SSD algorithm based on deep learning and according to target data in image data further
Target defect is tested and analyzed, or is directly tested and analyzed in image data using YOLO the or SSD algorithm based on deep learning
Target defect；
Step 4 carries out target classification in image data, target classification is perhaps carried out in target data or in mesh
Target classification is carried out in mark defect, the target classification is Inception the VGGNet network model based on deep learning
To realize the classification of target.
The present invention is based on the modular intelligent analysis systems of deep learning, comprising:
Image collection module obtains image data from camera or local picture；
Locating module, using geometric match, template matching or Faster-R-CNN, SSD based on deep learning or
YOLO algorithm searches in image data the similar target with target template, obtains target data；
Detection module, using YOLO the or SSD algorithm based on deep learning and according to target data in image data into
One step tests and analyzes target defect, or is directly detected in image data using YOLO the or SSD algorithm based on deep learning
Analyze target defect.
Categorization module carries out target classification in image data, or target classification, Huo Zhe are carried out in target data
Target classification is carried out in target defect, the target classification is Inception the VGGNet network mould based on deep learning
Type realizes the classification of target.
One embodiment of the present invention:
Include four parts based on deep learning module: image obtains, positioning, detection and classification；Detailed step is as follows: step
1, image collection module obtains image data from camera or local picture；Step 2, locating module exists according to known target template
All targets are searched in image data obtains target data, such as position；Step 3, detection module according to target position (to be step
The position of rapid 2 target data, it with step 3 is associated that the present embodiment, which is step 2, if being exactly to scheme to whole without step 2
As data are tested and analyzed to obtain the defect of target as target position) image data further test and analyze target lack
It falls into；Step 4, categorization module (is exactly the target to step 2 if without the target defect of step 3 to the target defect detected
Data carry out target classification and exactly directly carry out target classification to the raw image data of step 1 if skipping step 2 again) into
Row further obtains target classification；Wherein step 1 can be with step 2,3,4 any individually combinations, can also by 1,2,3,4,
Sequence all combinations.
A preferred embodiment method of the invention is following (the 1st column of system principle):
Further, step 2, geometric match, template matching or the Faster- based on deep learning can be used in locating module
R-CNN, SSD or YOLO algorithm search in image data the similar target position P1 with target template；
Further, step 3, the SSD algorithm based on deep learning can be used to find localized target sites in detection module
Thin objects position where P1 in image, for example: the position tiny defect P11, the position tiny defect P12, tiny defect P13
It sets；
Further, step 4, Inception the VGGNet network mould based on deep learning can be used in categorization module
Type realizes the classifications of thin objects, and for example: tiny defect P11 is scratch, and tiny defect P12 is broken hole, and tiny defect P13 is
It is dirty.
Existing technology has that two: 1. do not have self-study mechanism；2. need expert by Heuristics manually to
Feature out；This is two technical advantages of deep learning, so that the theoretical recognition accuracy of deep learning can be increased to 99%, it is general
General family can use foolly.
Although specific embodiments of the present invention have been described above, those familiar with the art should be managed
Solution, we are merely exemplary described specific embodiment, rather than for the restriction to the scope of the present invention, it is familiar with this
The technical staff in field should be covered of the invention according to modification and variation equivalent made by spirit of the invention
In scope of the claimed protection.
Claims (10)
1. one kind is based on the modular intelligent analysis method of deep learning, it is characterised in that: include:
Step 1 obtains image data from camera or local picture；
Step 2 searches in image data according to known target template and obtains target data；
Step 3 further tests and analyzes target defect according to target data in image data, or directly in image data
Test and analyze target defect.
2. according to claim 1 a kind of based on the modular intelligent analysis method of deep learning, it is characterised in that: also wrap
It includes step 4, carry out target classification in image data, target classification is perhaps carried out in target data or in target defect
Middle carry out target classification.
3. according to claim 2 a kind of based on the modular intelligent analysis method of deep learning, it is characterised in that: described
The classification that target classification is Inception the VGGNet network model based on deep learning to realize target.
4. according to claim 1 a kind of based on the modular intelligent analysis method of deep learning, it is characterised in that: described
Step 2 is further specifically: using geometric match, template matching or Faster-R-CNN, SSD based on deep learning or
YOLO algorithm searches in image data the similar target with target template, obtains target data.
5. according to claim 1 a kind of based on the modular intelligent analysis method of deep learning, it is characterised in that: described
Step 3 is further specifically: using YOLO the or SSD algorithm based on deep learning and according to target data in image data into
One step tests and analyzes target defect, or is directly detected in image data using YOLO the or SSD algorithm based on deep learning
Analyze target defect.
6. one kind is based on the modular intelligent analysis system of deep learning, it is characterised in that: include:
Image collection module obtains image data from camera or local picture；
Locating module is searched in image data according to known target template and obtains target data；
Detection module further tests and analyzes target defect according to target data in image data, or directly in picture number
According to middle detection and analysis target defect.
7. according to claim 6 a kind of based on the modular intelligent analysis system of deep learning, it is characterised in that: also wrap
Categorization module is included, target classification is carried out in image data, target classification is perhaps carried out in target data or is lacked in target
Target classification is carried out in falling into.
8. according to claim 7 a kind of based on the modular intelligent analysis system of deep learning, it is characterised in that: described
The classification that target classification is Inception the VGGNet network model based on deep learning to realize target.
9. according to claim 6 a kind of based on the modular intelligent analysis system of deep learning, it is characterised in that: described
Locating module is further specifically: using geometric match, template matching or Faster-R-CNN, SSD based on deep learning or
Person YOLO algorithm searches in image data the similar target with target template, obtains target data.
10. according to claim 6 a kind of based on the modular intelligent analysis system of deep learning, it is characterised in that: institute
It is further to state detection module specifically: using YOLO the or SSD algorithm based on deep learning and according to target data in picture number
Target defect is further tested and analyzed in, or using YOLO the or SSD algorithm based on deep learning directly in image data
Middle detection and analysis target defect.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910200205.7A CN109978844A (en) | 2019-03-15 | 2019-03-15 | One kind being based on the modular intelligent analysis method of deep learning and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910200205.7A CN109978844A (en) | 2019-03-15 | 2019-03-15 | One kind being based on the modular intelligent analysis method of deep learning and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109978844A true CN109978844A (en) | 2019-07-05 |
Family
ID=67079109
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910200205.7A Pending CN109978844A (en) | 2019-03-15 | 2019-03-15 | One kind being based on the modular intelligent analysis method of deep learning and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109978844A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111242924A (en) * | 2020-01-13 | 2020-06-05 | 浙江水利水电学院 | Product quality management system |
| CN113030121A (en) * | 2021-03-11 | 2021-06-25 | 微讯智造(广州)电子有限公司 | Automatic optical detection method, system and equipment for circuit board components |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107123114A (en) * | 2017-04-21 | 2017-09-01 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of cloth defect inspection method and device based on machine learning |
| CN107833220A (en) * | 2017-11-28 | 2018-03-23 | 河海大学常州校区 | Fabric defect detection method based on depth convolutional neural networks and vision significance |
| CN108257114A (en) * | 2017-12-29 | 2018-07-06 | 天津市万贸科技有限公司 | A kind of transmission facility defect inspection method based on deep learning |
| CN108765402A (en) * | 2018-05-30 | 2018-11-06 | 武汉理工大学 | Non-woven fabrics defects detection and sorting technique |
| CN108932713A (en) * | 2018-07-20 | 2018-12-04 | 成都指码科技有限公司 | A kind of weld porosity defect automatic testing method based on deep learning |
| CN108982508A (en) * | 2018-05-23 | 2018-12-11 | 江苏农林职业技术学院 | A kind of plastic-sealed body IC chip defect inspection method based on feature templates matching and deep learning |
| CN109063708A (en) * | 2018-06-21 | 2018-12-21 | 歌尔股份有限公司 | Industrial picture characteristic recognition method and system based on contours extract |
-
2019
- 2019-03-15 CN CN201910200205.7A patent/CN109978844A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107123114A (en) * | 2017-04-21 | 2017-09-01 | 佛山市南海区广工大数控装备协同创新研究院 | A kind of cloth defect inspection method and device based on machine learning |
| CN107833220A (en) * | 2017-11-28 | 2018-03-23 | 河海大学常州校区 | Fabric defect detection method based on depth convolutional neural networks and vision significance |
| CN108257114A (en) * | 2017-12-29 | 2018-07-06 | 天津市万贸科技有限公司 | A kind of transmission facility defect inspection method based on deep learning |
| CN108982508A (en) * | 2018-05-23 | 2018-12-11 | 江苏农林职业技术学院 | A kind of plastic-sealed body IC chip defect inspection method based on feature templates matching and deep learning |
| CN108765402A (en) * | 2018-05-30 | 2018-11-06 | 武汉理工大学 | Non-woven fabrics defects detection and sorting technique |
| CN109063708A (en) * | 2018-06-21 | 2018-12-21 | 歌尔股份有限公司 | Industrial picture characteristic recognition method and system based on contours extract |
| CN108932713A (en) * | 2018-07-20 | 2018-12-04 | 成都指码科技有限公司 | A kind of weld porosity defect automatic testing method based on deep learning |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111242924A (en) * | 2020-01-13 | 2020-06-05 | 浙江水利水电学院 | Product quality management system |
| CN113030121A (en) * | 2021-03-11 | 2021-06-25 | 微讯智造(广州)电子有限公司 | Automatic optical detection method, system and equipment for circuit board components |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11763443B2 (en) | Method for monitoring manufacture of assembly units | |
| CN101576956B (en) | On-line character detection method based on machine vision and system thereof | |
| US5933519A (en) | Cytological slide scoring apparatus | |
| CN108830332A (en) | A kind of vision vehicle checking method and system | |
| JP2011007800A (en) | Method for the automatic analysis of microscope images | |
| CN110992349A (en) | Underground pipeline abnormity automatic positioning and identification method based on deep learning | |
| CN111179250B (en) | Industrial product defect detection system based on multitask learning | |
| CN109389109B (en) | Automatic testing method and device for OCR full-text recognition accuracy | |
| CN116188475B (en) | Intelligent control method, system and medium for automatic optical detection of appearance defects | |
| CN110533654A (en) | The method for detecting abnormality and device of components | |
| CN103593695A (en) | Method for positioning DPM two-dimension code area | |
| JP2013077127A (en) | Image classification device and image classification method | |
| CN102663418B (en) | An image set modeling and matching method based on regression model | |
| CN109978844A (en) | One kind being based on the modular intelligent analysis method of deep learning and system | |
| CN111103307A (en) | Pcb defect detection method based on deep learning | |
| CN103177266A (en) | Intelligent stock pest identification system | |
| CN112288741A (en) | Product surface defect detection method and system based on semantic segmentation | |
| CN113780342A (en) | Intelligent detection method and device based on self-supervision pre-training and robot | |
| CN110263608B (en) | Automatic electronic component identification method based on image feature space variable threshold measurement | |
| CN105426926B (en) | A kind of couple of AMOLED carries out the method and device of detection classification | |
| CN115019294A (en) | Pointer instrument reading identification method and system | |
| CN105224941A (en) | Process identification and localization method | |
| CN112184679A (en) | YOLOv 3-based wine bottle flaw automatic detection method | |
| CN108229586B (en) | The detection method and system of a kind of exceptional data point in data | |
| CN104459516A (en) | Testing method used for SMT first article detection |
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 |