CN111179241A - Panel defect detection and classification method and system - Google Patents
Panel defect detection and classification method and system Download PDFInfo
- Publication number
- CN111179241A CN111179241A CN201911356807.8A CN201911356807A CN111179241A CN 111179241 A CN111179241 A CN 111179241A CN 201911356807 A CN201911356807 A CN 201911356807A CN 111179241 A CN111179241 A CN 111179241A
- Authority
- CN
- China
- Prior art keywords
- defect
- panel
- panel defect
- data
- picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- 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
- G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- 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
- G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
- G01N2021/9513—Liquid crystal panels
-
- 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
- G06T2207/20—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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Quality & Reliability (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a panel defect detection and classification method and a system, wherein the method comprises the following steps: collecting panel defect picture data from historical data and storing the panel defect picture data according to the panel defect type; cleaning the stored panel defect picture data; optimizing the cleaned panel defect picture data; marking optimized panel defect picture data to obtain a training set; training a panel defect detection and classification model by using a training set; and inputting the preset panel defect picture into the trained panel defect detection and classification model, and outputting the detection and classification result of the panel defect. The method and the system can improve the efficiency and the accuracy of defect identification and detection and can work continuously and stably.
Description
Technical Field
The invention relates to the field of intelligent manufacturing, in particular to a panel defect detection and classification method and system.
Background
Along with the rapid upgrade of the intelligent terminal industry in China, the market demand of the liquid crystal display panel is huge. The panel defect detection method of the current liquid crystal display panel production factory is to initially position the defects through AOI equipment and then manually classify the defects. The detection and classification of the defects of human eyes have strong subjectivity, are not beneficial to strictly classifying the defect types, and the detection accuracy is reduced due to the exhaustion of human eyes. As the productivity of the factory assembly line rises, the low efficiency of human visual inspection cannot be matched with the high-speed production.
The current panel defect method of a factory cannot meet the requirements of the factory on efficiency and precision, and a more effective panel defect detection method is urgently needed to assist and even replace the traditional detection method.
Disclosure of Invention
In order to solve the above problems, the present invention provides a method and a system for detecting and classifying defects of a panel, and more particularly, to a method and a system for detecting and classifying defects of a panel (combined picture) suitable for a CELL process (CELL) production process. The combination is that the cameras in the production line cannot cover the whole panel one by one, so that the images are spliced and combined after being shot for multiple times. The method and the system use the deep convolutional neural network, improve the detection rate of the unobvious defects, and define classification defects by combining the defects of a factory; the GPU is used for accelerating operation, and the time consumed by single detection reaches the speed of millisecond level; the mechanical detection classification state is more stable; the accuracy and the efficiency of panel defect detection can be effectively improved.
To achieve the above object, one aspect of the present invention provides a panel defect detecting and classifying method, including:
collecting panel defect picture data from historical data and storing the panel defect picture data according to the panel defect type;
cleaning the stored panel defect picture data;
optimizing the cleaned panel defect picture data;
marking optimized panel defect picture data to obtain a training set;
training a panel defect detection and classification model by using a training set;
and inputting the preset panel defect picture into the trained panel defect detection and classification model, and outputting a panel defect detection and classification result.
Preferably, the filtering removes abnormal data in the data, and the abnormal data includes: overexposure, over-dark, blurred, virtual focus, excessive defects, and no defect data.
Preferably, the optimizing the cleaned panel defect picture data includes: and optimizing the panel defect picture data by adopting a layered random sampling method, and performing sample equalization processing on the panel defect picture data.
Preferably, the sample equalization processing is performed on the panel defect picture data by a method of adding disturbance or inversion or gaussian noise in the panel defect picture data.
Preferably, the model is trained under the mmdetection detection framework using the net50 network, with the anchor _ rates field of the model configuration file set to [0.1,0.33,1.0,3.0,10 ].
Preferably, a learning rate that decreases stepwise is set during model training, so that the model can converge quickly.
Preferably, the method further comprises: and removing redundant description information in the output result of the panel defect detection and classification model, converting the digital identification in the output result into a defect name, combining the output results corresponding to a plurality of defect pictures shot by a plurality of light sources corresponding to the same panel to form data consisting of the structure of the panel ID, the picture ID, the defect name, the defect confidence coefficient and the defect position, and taking the data as feedback data received by the rear end.
Preferably, the method uses a labelImage marking tool to mark the defects in the optimized panel defect picture data by using a rectangular frame so as to be used for panel defect detection and classification model training.
Preferably, the method includes a front-end executing part and a back-end executing part, the front-end executing part includes the method content, and the back-end executing part includes:
processing a front-end request: the front end calls a back end service in an http request mode, and after the back end receives the request, the back end analyzes according to an agreed format, assembles the path information of the defect picture, acquires the panel ID of the defect picture and analyzes the channel ID of the defect picture;
calling model operation: the analyzed request is forwarded to a model server, a model and a related configuration file used by the model server are designated, a model detection defect picture path is set, and the model is called; receiving a detection result fed back by the model, carrying out logic operation on the fed-back result according to the definition of the defect picture to obtain the defect type of the detection panel, and counting the defect number and the defect size information;
and (3) result feedback: and sorting the logical operation output result format and returning to the front end in a key value pair mode.
In another aspect, the present invention further provides a panel defect detecting and classifying system, including:
the data acquisition and storage unit is used for acquiring panel defect picture data from historical data and storing the panel defect picture data according to panel defect types;
the data cleaning unit is used for cleaning the stored panel defect picture data;
the data optimization unit is used for optimizing the cleaned panel defect picture data;
the data marking unit is used for marking the optimized panel defect picture data to obtain a training set;
the model training unit is used for training the panel defect detection and classification model by using a training set;
and the defect detection and classification unit is used for inputting the preset panel defect picture into the trained panel defect detection and classification model and outputting the panel defect detection and classification result.
One or more technical schemes provided by the invention at least have the following technical effects or advantages:
1. the method and the system can improve the defect identification and detection efficiency, the GPU is used for high-speed operation, the processing efficiency of a single picture reaches millisecond level, one-time complete detection and classification is controlled within 1-2 seconds, and the method and the system are efficient and rapid.
2. The method and the system can improve the accuracy of defect identification and detection, and can better learn the characteristics of the small defects and reduce the missing detection of the small defects by using the convolutional neural network.
3. The method and the system can work continuously and stably, because the model detection result is used, subjectivity does not exist, defect misclassification cannot occur, efficiency and accuracy cannot be reduced due to long-time operation, and the method and the system are more suitable for a production line operation mode of a factory.
Drawings
The accompanying drawings, which are included to provide a further understanding of the embodiments of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention;
FIG. 1 is a schematic flow chart of a panel defect detection and classification method according to the present invention;
FIG. 2 is a schematic diagram of a panel defect detection and classification system according to the present invention.
Detailed Description
In order that the above objects, features and advantages of the present invention can be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings. It should be noted that the embodiments of the present invention and features of the embodiments may be combined with each other without conflicting with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those specifically described and thus the scope of the present invention is not limited by the specific embodiments disclosed below.
Referring to fig. 1, an embodiment of the present invention provides a panel defect detecting and classifying method, and in this embodiment, the method includes two parts, namely model training and back-end control.
The model training comprises the following steps: 1. data preparation, 2, model iteration, 3 and result sorting.
The data preparation comprises the following steps: data acquisition, data cleaning, data optimization and data annotation;
the data acquisition comprises the following steps: and summarizing defect distribution according to business requirements, acquiring historical data from a production line under the assistance of factory personnel, and storing the historical data in a fixed-level folder according to defect types.
The data cleaning comprises the following steps: and (4) re-inspecting the data, and filtering and removing abnormal data in the data, wherein the abnormal data comprises data with excessive exposure, excessive darkness, blurred picture, virtual focus of the picture, excessive defect and no defect of the picture.
Optimizing data: the method mainly optimizes data distribution, and has the serious problem of uneven data in a factory production line, namely the proportion of some defect data is extremely high, and the sample amount of some defect data is insufficient; aiming at the defect samples with excessive quantity, the method adopts a layered random sampling method to extract the samples, so that the number of the samples is reduced; aiming at data with insufficient samples, the method generates a plurality of samples by adding disturbance, inversion and Gaussian noise to enrich the total amount of the samples, thereby balancing the distribution of the samples.
Data annotation: labeling defects of acquired picture data with a rectangular box by means of a labelImage labeling tool for model training
Model iteration: the model is trained under the mmdetection framework by using the resnet50 network, and because the aspect ratio of the line defects is large, the anchor _ rates field of the model configuration file is set to [0.1,0.33,1.0,3.0 and 10], so that the aspect ratio of the anchor is more beneficial to learning the line defects by the model, and meanwhile, the step-type descending learning rate is set, so that the model can be rapidly converged.
And (4) result finishing: removing redundant description information of the model detection result, replacing the classification result of the model, converting the original digital identification into a more readable defect name, combining the detection results of multiple pictures shot by multiple light sources under the same panel, and finally taking data consisting of the structure of the panel ID, the picture ID, the defect name, the defect confidence coefficient and the defect position as feedback data received by a back end
The back-end control includes: 1. processing a front-end request, 2, calling model operation, 3 and feeding back a result;
processing a front-end request: and the factory side calls a back-end service in an http request mode, and after receiving the request, the back end analyzes according to an agreed format, assembles the picture path information, acquires the picture panel ID and analyzes the channel ID of the picture.
Calling model operation: the analyzed request is forwarded to a model server, a model and a related configuration file used by the model server are designated, a model detection picture path is set, and the model is called; and receiving the detection result fed back by the model, carrying out logic operation on the fed-back result according to the definition of the factory defects to obtain the defect types of the detection panel, and counting the defect quantity and the defect size information.
And (3) result feedback: and (4) sorting the output result format of the logic operation and returning the output result format to a factory end (front end) in a key value pair mode.
Referring to fig. 2, an embodiment of the present invention provides a panel defect detecting and classifying system, including:
the data acquisition and storage unit is used for acquiring panel defect picture data from historical data and storing the panel defect picture data according to panel defect types;
the data cleaning unit is used for cleaning the stored panel defect picture data;
the data optimization unit is used for optimizing the cleaned panel defect picture data;
the data marking unit is used for marking the optimized panel defect picture data to obtain a training set;
the model training unit is used for training the panel defect detection and classification model by using a training set;
and the defect detection and classification unit is used for inputting the preset panel defect picture into the trained panel defect detection and classification model and outputting the detection and classification result of the panel defect.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A method of panel defect detection and classification, the method comprising:
collecting panel defect picture data from historical data and storing the panel defect picture data according to the panel defect type;
cleaning the stored panel defect picture data;
optimizing the cleaned panel defect picture data;
marking optimized panel defect picture data to obtain a training set;
training a panel defect detection and classification model by using a training set;
and inputting the preset panel defect picture into the trained panel defect detection and classification model, and outputting a panel defect detection and classification result.
2. The method of claim 1, wherein the abnormal data in the panel defect picture data is filtered and removed, and the abnormal data comprises: picture overexposure, picture over-darkness, picture blurring, picture virtual focus, picture over-defect, and picture defect free pictures.
3. The method of claim 1, wherein optimizing the cleaned panel defect picture data comprises: and optimizing the panel defect picture data by adopting a layered random sampling method, and performing sample equalization processing on the panel defect picture data.
4. The method of claim 3, wherein the panel defect image data is processed by sample equalization by adding disturbance or inversion or Gaussian noise to the panel defect image data.
5. The panel defect detecting and classifying method according to claim 1, wherein a model is trained under the mmdetection detection framework using a resnet50 network, and the anchor _ rates field of the model configuration file is set to [0.1,0.33,1.0,3.0,10 ].
6. The method of claim 5, wherein a learning rate that decreases stepwise is set during model training.
7. The method of claim 1, further comprising: and removing redundant description information of the panel defect detection and classification model output result, converting the digital identifier of the output result into a defect name, combining the output results corresponding to a plurality of defect pictures shot by multiple light sources corresponding to the same panel to form data consisting of a panel ID, a picture ID, the defect name, a defect confidence coefficient and a defect position structure, and taking the data as feedback data received by the rear end.
8. The method for detecting and classifying panel defects according to claim 1, wherein the method labels defects in the optimized panel defect picture data with a rectangular frame by means of a labelImage labeling tool for training a panel defect detection and classification model.
9. The method of claim 1, wherein the method comprises a front-end execution part and a back-end execution part, the front-end execution part comprising the method content of any one of claims 1 to 8, and the back-end execution part comprising:
processing a front-end request: the front end calls a back end service in an http request mode, and after the back end receives the request, the back end analyzes according to an agreed format, assembles the path information of the defect picture, acquires the panel ID of the defect picture and analyzes the channel ID of the defect picture;
calling model operation: the analyzed request is forwarded to a model server, a model and a related configuration file used by the model server are designated, a model detection defect picture path is set, and the model is called; receiving a detection result fed back by the model, carrying out logic operation on the fed-back result according to the definition of the defect picture to obtain the defect type of the detection panel, and counting the defect number and the defect size information;
and (3) result feedback: and sorting the logical operation output result format and returning to the front end in a key value pair mode.
10. A panel defect detection and classification system, the system comprising:
the data acquisition and storage unit is used for acquiring panel defect picture data from historical data and storing the panel defect picture data according to panel defect types;
the data cleaning unit is used for cleaning the stored panel defect picture data;
the data optimization unit is used for optimizing the cleaned panel defect picture data;
the data marking unit is used for marking the optimized panel defect picture data to obtain a training set;
the model training unit is used for training the panel defect detection and classification model by using a training set;
and the defect detection and classification unit is used for inputting the preset panel defect picture into the trained panel defect detection and classification model and outputting the panel defect detection and classification result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911356807.8A CN111179241A (en) | 2019-12-25 | 2019-12-25 | Panel defect detection and classification method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911356807.8A CN111179241A (en) | 2019-12-25 | 2019-12-25 | Panel defect detection and classification method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111179241A true CN111179241A (en) | 2020-05-19 |
Family
ID=70652107
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911356807.8A Pending CN111179241A (en) | 2019-12-25 | 2019-12-25 | Panel defect detection and classification method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111179241A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111709951A (en) * | 2020-08-20 | 2020-09-25 | 成都数之联科技有限公司 | Target detection network training method and system, network, device and medium |
| CN113313679A (en) * | 2021-05-21 | 2021-08-27 | 浙江大学 | Bearing surface defect detection method based on multi-source domain depth migration multi-light source integration |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108229561A (en) * | 2018-01-03 | 2018-06-29 | 北京先见科技有限公司 | Particle product defect detection method based on deep learning |
| US20180232601A1 (en) * | 2017-02-16 | 2018-08-16 | Mitsubishi Electric Research Laboratories, Inc. | Deep Active Learning Method for Civil Infrastructure Defect Detection |
| CN109754392A (en) * | 2018-12-20 | 2019-05-14 | 上海集成电路研发中心有限公司 | A kind of device and method that defect image automatically analyzes |
| CN110084313A (en) * | 2019-05-05 | 2019-08-02 | 厦门美图之家科技有限公司 | A method of generating object detection model |
| CN110111331A (en) * | 2019-05-20 | 2019-08-09 | 中南大学 | Honeycomb paper core defect inspection method based on machine vision |
| CN110455822A (en) * | 2019-07-10 | 2019-11-15 | 苏州卓融新能源科技有限公司 | A kind of detection method of pcb board defect |
| CN110554047A (en) * | 2019-09-06 | 2019-12-10 | 腾讯科技（深圳）有限公司 | method, device, system and equipment for processing product defect detection data |
| CN110570393A (en) * | 2019-07-31 | 2019-12-13 | 华南理工大学 | mobile phone glass cover plate window area defect detection method based on machine vision |
-
2019
- 2019-12-25 CN CN201911356807.8A patent/CN111179241A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180232601A1 (en) * | 2017-02-16 | 2018-08-16 | Mitsubishi Electric Research Laboratories, Inc. | Deep Active Learning Method for Civil Infrastructure Defect Detection |
| CN108229561A (en) * | 2018-01-03 | 2018-06-29 | 北京先见科技有限公司 | Particle product defect detection method based on deep learning |
| CN109754392A (en) * | 2018-12-20 | 2019-05-14 | 上海集成电路研发中心有限公司 | A kind of device and method that defect image automatically analyzes |
| CN110084313A (en) * | 2019-05-05 | 2019-08-02 | 厦门美图之家科技有限公司 | A method of generating object detection model |
| CN110111331A (en) * | 2019-05-20 | 2019-08-09 | 中南大学 | Honeycomb paper core defect inspection method based on machine vision |
| CN110455822A (en) * | 2019-07-10 | 2019-11-15 | 苏州卓融新能源科技有限公司 | A kind of detection method of pcb board defect |
| CN110570393A (en) * | 2019-07-31 | 2019-12-13 | 华南理工大学 | mobile phone glass cover plate window area defect detection method based on machine vision |
| CN110554047A (en) * | 2019-09-06 | 2019-12-10 | 腾讯科技（深圳）有限公司 | method, device, system and equipment for processing product defect detection data |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111709951A (en) * | 2020-08-20 | 2020-09-25 | 成都数之联科技有限公司 | Target detection network training method and system, network, device and medium |
| CN113313679A (en) * | 2021-05-21 | 2021-08-27 | 浙江大学 | Bearing surface defect detection method based on multi-source domain depth migration multi-light source integration |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kumar et al. | Colorization transformer | |
| Song et al. | Enhancement of underwater images with statistical model of background light and optimization of transmission map | |
| CN106920229B (en) | Automatic detection method and system for image fuzzy area | |
| CN109714519B (en) | Method and system for automatically adjusting image frame | |
| Rijal et al. | Ensemble of deep neural networks for estimating particulate matter from images | |
| CN110298819A (en) | A kind of quality detecting method and edge calculations equipment of target object | |
| CN111179241A (en) | Panel defect detection and classification method and system | |
| CN112766334B (en) | Cross-domain image classification method based on pseudo label domain adaptation | |
| CN111626279B (en) | Negative sample labeling training method and highly-automatic bill identification method | |
| CN111062961A (en) | Contact lens edge defect detection method based on deep learning | |
| CN110599453A (en) | Panel defect detection method and device based on image fusion and equipment terminal | |
| CN112116582A (en) | Cigarette detection and identification method under stock or display scene | |
| CN115170550A (en) | Deep learning-based battery defect detection method and system | |
| CN113505780A (en) | Two-dimensional code-based intelligent detection maintenance method and equipment | |
| CN114359235A (en) | Wood surface defect detection method based on improved YOLOv5l network | |
| CN111507249A (en) | Transformer substation nest identification method based on target detection | |
| CN110019953A (en) | A kind of real-time quick look system of payload image data | |
| CN116521917A (en) | Picture screening method and device | |
| CN114399487A (en) | Video definition and brightness detection method based on data annotation learning mechanism | |
| CN114913312A (en) | Deep learning model data image information acquisition method | |
| CN112418362A (en) | Target detection training sample screening method | |
| CN113657310A (en) | Photovoltaic module infrared hot spot fault identification method based on multilayer convolutional neural network | |
| CN111474181A (en) | Panel defect detecting and classifying system in box forming process stage | |
| WO2020185432A1 (en) | Pre-processing image frames based on camera statistics | |
| CN115512023A (en) | Film and television image post three-dimensional processing system |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 610000 No. 270, floor 2, No. 8, Jinxiu street, Wuhou District, Chengdu, Sichuan Applicant after: Chengdu shuzhilian Technology Co.,Ltd. Address before: 610000 No.2, 4th floor, building 1, Jule Road intersection, West 1st section of 1st ring road, Wuhou District, Chengdu City, Sichuan Province Applicant before: CHENGDU SHUZHILIAN TECHNOLOGY Co.,Ltd. |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200519 |