CN108764365A - A kind of device signboard detection method - Google Patents
A kind of device signboard detection method Download PDFInfo
- Publication number
- CN108764365A CN108764365A CN201810576027.3A CN201810576027A CN108764365A CN 108764365 A CN108764365 A CN 108764365A CN 201810576027 A CN201810576027 A CN 201810576027A CN 108764365 A CN108764365 A CN 108764365A
- Authority
- CN
- China
- Prior art keywords
- device signboard
- sample image
- image
- region
- files
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/09—Recognition of logos
Abstract
The present invention relates to a kind of device signboard detection methods, using the corresponding tab area of device signboard as the feature of the conspicuousness the most of difference device signboard and ambient background, device signboard is identified by this main feature, to exclude interference of other extraneous features to Detection accuracy, so that more neurons concentrate on the study of device signboard feature, device signboard and background information are more easily distinguished, to improve the Detection accuracy of grid equipment Sign Board.
Description
Technical field
The present invention relates to target identification technology fields, more specifically to a kind of device signboard detection method.
Background technology
Target detection is not difficult for the mankind, passes through the perception to different colours module in picture, it is easy to which positioning is simultaneously
Sort out wherein target object.But for computer, what is faced is rgb pixels matrix, it is difficult to from image directly
To abstract concept as dog and cat and its position is determined, along with sometimes multiple objects and mixed and disorderly background are mingled in one
It rises, target detection is just more difficult.
Traditional target detection adds characteristic value matching to be detected using sliding window, and basic step is exactly to utilize different rulers
Very little sliding window, certain part for framing image extract the relevant visual signature in candidate region as candidate region,
For example the common Harr features of Face datection, pedestrian detection and general goals detect common HOG features etc., finally utilize classification
Device carries out Classification and Identification.
Although traditional method can detect target, there are problems that two very serious：
First, the regional choice strategy based on sliding window does not have specific aim, time complexity is high, window redundancy；
Second is that the feature of hand-designed is for multifarious variation, there is no good robustness.
Invention content
It is an object of the invention to overcome the deficiencies of the prior art and provide a kind of detections that grid equipment Sign Board can be improved
The device signboard detection method of accuracy rate and detection efficiency.
Technical scheme is as follows：
A kind of device signboard detection method, steps are as follows：
1) in the sample image to be trained comprising device signboard, device signboard is set to tab area, root
According to preset neural network model and tab area, sample image is trained, obtains weight model；
2) image to be detected is obtained, and image to be detected is divided into multiple regions；
3) according to weight model, each region of image to be detected is detected respectively, obtains corresponding to setting for each region
Reliability；
4) region that confidence level is greater than or equal to predetermined threshold value is returned to.
Preferably, in step 1), sample image is included under Different climate condition, different lightness environment through different angles
Spend the original image shot to device signboard.
Preferably, in step 1), the training step of weight model is specific as follows：
1.1) sample image to be trained and the corresponding tab area of device signboard are obtained；
1.2) sample image to be trained is divided into multiple regions comprising the region of tab area is mesh
Region, other regions be background area；
1.3) study is trained to the pixel value in target area and background area by preset neural network model,
Obtain weight model.
Preferably, in step 1.3), network object is first generated according to preset neural network configuration file, passes through generation
Network object study is trained to the pixel value in all areas of sample image, iterative process is optimized after training
Parameter preserve to weight model.
Preferably, in step 1), by manually carrying out position mark to device signboard on sample image, marked
Note region.
Preferably, in step 1), sample image to be trained is built into training data by VOC standard data set formats
Collection, specifically, steps are as follows：
First, tri- files of Annotation, ImageSets and JPEGImages are created, under ImageSets files
Including Main files；
Then, sample image is named according to unified naming rule, is then stored in sample image
Under JPEGImages files；
Then, position mark is carried out to the device signboard in sample image using LabelImg tools, and generates xml texts
Part；Wherein, xml document corresponds to a sample image, the content of storage include annotation marks object,
Floder objects, fliename objects, path objects, size objects, the multiple object objects for including in annotation；
Finally, the xml document of all generations is divided into two parts, the sample image title write-in in a part of xml document
For training in trainval.txt files, in the sample image title write-in test.txt files in another part xml document
For the verification in training process；
Trainval.txt files and test.txt files are stored under the Main files in ImageSet files；
All xml documents are stored under Annotation files.
Preferably, in step 4), returns to location information of the confidence level more than or equal to the region of predetermined threshold value and its set
Reliability, the location information include the x-axis coordinate, y-axis coordinate and peak width and height at a certain angle in region.
Preferably, by the location information of return area, so that it is carried out ring box in image to be detected and show.
Preferably, the region that confidence level is greater than or equal to predetermined threshold value includes device signboard.
Preferably, weight model is integrated to background service, called using picture detection script, and back office interface is provided
To client；When client obtains image to be detected, and it is uploaded to back office interface, back office interface calls picture to detect script, will
Image to be detected is divided into multiple regions, and each region is detected and is given a mark according to the parameter in weight model, obtains each area
The confidence level in domain.
Beneficial effects of the present invention are as follows：
In device signboard detection method of the present invention, set the corresponding tab area of device signboard as difference
The feature of the conspicuousness the most of standby Sign Board and ambient background, device signboard is identified by this main feature, to exclude
Interference of other extraneous features to Detection accuracy so that more neurons concentrate on the study of device signboard feature, more
Add easily distinguishable device signboard and background information, to improve the Detection accuracy of grid equipment Sign Board.
Description of the drawings
Fig. 1 is the flow diagram of the present invention；
Fig. 2 is the schematic network structure of YOLO neural networks.
Specific implementation mode
The present invention is further described in detail with reference to the accompanying drawings and embodiments.
The present invention does not have specific aim to solve regional choice strategy of the existing technology, and time complexity is high, window
The feature of redundancy, hand-designed changes the deficiencies of there is no good robustness for multifarious, provides a kind of device identification
Board detection method, this notable feature of capture apparatus Sign Board are distinguished with background, exclude other extraneous features to detection
The Detection accuracy and detection efficiency of grid equipment Sign Board can be improved in the interference of accuracy rate.
As shown in Figure 1, the device signboard detection method, steps are as follows：
1) in the sample image to be trained comprising device signboard, device signboard is set to tab area, root
According to preset neural network model and tab area, sample image is trained, obtains weight model；
2) image to be detected is obtained, and image to be detected is divided into multiple regions；
3) according to weight model, each region of image to be detected is detected respectively, obtains corresponding to setting for each region
Reliability；
4) region that confidence level is greater than or equal to predetermined threshold value is returned to, wherein confidence level is greater than or equal to predetermined threshold value
Region includes device signboard.
In order to improve the robustness of subsequent detection, the present invention is used for by acquiring the device signboard photo under varying environment
Training, that is, sample image, which is included under Different climate condition, different lightness environment, carries out device signboard by different angle
The original image of shooting.
In the present invention, the tab area that tagging equipment Sign Board part is obtained is as target area so that more god
The study of device signboard feature is concentrated on through member, more easily distinguishes device signboard and background information.Then in step 1), power
The training step of molality type is specific as follows：
1.1) sample image to be trained and the corresponding tab area of device signboard are obtained；When it is implemented, can pass through
Position mark manually is carried out to device signboard on sample image, obtains tab area；In the present embodiment, the figure increased income is utilized
Piece annotation tool LabelImg carries out position mark to the device signboard part in every sample image manually；
1.2) sample image to be trained is divided into multiple regions comprising the region of tab area is mesh
Region, other regions be background area；
1.3) study is trained to the pixel value in target area and background area by preset neural network model,
Obtain weight model；Specifically, network object is first generated (in exploitation in object-oriented according to preset neural network configuration file
Class, including the level attributes of network and training method later), own to sample image by the network object of generation
Pixel value in region is trained study, and the parameter after training by iterative process optimization is preserved to weight model.
For the sample image for training weight model, in the present embodiment, by sample image to be trained by VOC standards
Data set format is built into training dataset, and specifically, steps are as follows：
First, tri- files of Annotation, ImageSets and JPEGImages are created, under ImageSets files
Including Main files；
Then, sample image is named according to unified naming rule, is then stored in sample image
Under JPEGImages files；
Then, position mark is carried out to the device signboard in sample image using LabelImg tools, and generates xml texts
Part；Wherein, xml document corresponds to a sample image, the content of storage include annotation marks object,
The floder objects (sample image where file) that include in annotation, fliename objects (sample image title),
Path objects (the specific path of sample image), size objects (sample image size), multiple object objects (include target object
Regional location in sample image of tag along sort title, target object)；
Finally, the xml document of all generations is divided into two parts, the sample image title in a part of xml document is (without after
Sew no path) it is written for training in trainval.txt files, the sample image title in another part xml document is (without after
Sew no path) it is written in test.txt files for the verification in training process；By trainval.txt files and test.txt
File is stored under the Main files in ImageSet files；Meanwhile all xml documents are stored in Annotation
Under file；It further, can be consistent for the title of corresponding sample image by the name nominating of xml document.
According to training dataset, the training of neural network model is carried out, specifically, steps are as follows：
First, neural network model configuration file is generated, the present embodiment is with YOLO (You Only Look Once) nerves
Based on network model, neural network model .cfg files are configured.
YOLO neural networks are based on individual end-to-end (end-to-end) neural network, complete from sample image
It is input to the output of object space and classification, network structure includes 24 convolutional layers and 2 full articulamentums, as shown in Figure 2.Its
In, convolutional layer is used for extracting characteristics of image, and full articulamentum is used for future position and class probability value.What the present embodiment used
Neural network has used for reference GoogLeNet sorter network structures.Unlike, inception module (GoogLeNet are not used
In wherein one layer), but simply substituted using+3 × 3 convolutional layer of 1 × 1 convolutional layer, wherein the presence of 1 × 1 convolutional layer herein
It is to be integrated across channel information.Meanwhile using mean square error (each data deviate the average of the square distance sum of actual value)
Carry out Optimized model parameter as loss functions (loss function), i.e., S × S × (B × 5+C) dimensional vector of neural network output with it is true
The side of correspondence S × S of real image × (B × 5+C) dimensional vector and error.
For configuring neural network model .cfg files, specifically, by detection device Sign Board one in this present embodiment
A classification, therefore the value of class is changed to 1；Filters values are repaiied according to only 1 class in region area levels
Change, due to filters=(classes+coords+1) × NUM, wherein classes indicates categorical measure, is in the present embodiment
4 coordinates tx, ty, tw, th that BoundingBox (bounding box) is indicated for 1, coords, coords=4, NUM in the present embodiment
Indicate BoundingBox number of each grid cell (grid cell) prediction, NUM=5 in the present embodiment；Therefore, this implementation
In example, the value of filters=(1+4+1) × 5=30, i.e., modified filters are 30.Further, names is changed to
Card (card signs for device signboard contingency table), it is voc_card.cfg then to preserve neural network model configuration file.
Then training script reads ImageSet/ according to the voc_card.cfg file generated network objects configured
Trainval.txt and test.txt under Main files, it is all according to the name acquiring of the sample image preserved in text document
The corresponding xml document of sample image reads the sample graph under JPEGImages files according to the information preserved in xml document
Picture, at the same from xml document read sample image tab area；Sample image is divided into multiple regions, then to sample
Pixel value in image in all areas is trained study.It waits after training, the parameter that script optimizes iterative process is protected
It deposits to weight model voc_card_8000.weigh.
In the present invention, network frame can be built, is used for far call, local detection.Weight model is integrated to backstage
Service is called using picture detection script, and provides back office interface to client；When client obtains image to be detected, and on
Back office interface is reached, back office interface calls picture to detect script, image to be detected is divided into multiple regions, according to weight model
In parameter each region is detected and is given a mark, obtain the confidence level in each region.
Specifically, the weight model voc_card_8000.weight files that training terminates to generate backstage is integrated to take
Business, user can be used as client by mobile phone terminal, then shooting grid equipment scene photograph is uploaded to as image to be detected
Back office interface, back office interface calls picture detection script that photo is divided into multiple regions, according to weight model voc_card_
Parameter in 8000.weight is detected and gives a mark to each region, obtains the confidence level in each region；Then successively by each area
The confidence level in domain is compared with preset threshold value, if being greater than or equal to threshold value, then it represents that the region includes device signboard, will
The location information and its confidence level in the region return to mobile phone terminal with json formats, wherein the location information in region includes region
Upper left corner x-axis coordinate, region upper left corner y-axis coordinate, peak width and region height.Preferably, threshold value 0.3.
In turn, in step 4), location information and its confidence level that confidence level is greater than or equal to the region of predetermined threshold value are returned,
The location information includes the x-axis coordinate, y-axis coordinate and peak width and height at a certain angle in region.Pass through return area
Location information, so that it is carried out ring box in image to be detected and show.
In the present embodiment, device signboard region is carried out ring box by mobile phone terminal according to obtained data in photo, and
Display Category label information and confidence level.
Verified, the present embodiment uses YOLO neural network models, and detection speed is up to 45fps, to realize in real time
Detection；It is distinguished with background by capture apparatus Sign Board this notable feature, further improves Detection accuracy；Together
When, it is attained by higher Detection accuracy and detection speed, robustness under complicated background, under strong light, weak light environment
It is good.
Above-described embodiment is intended merely to illustrate the present invention, and is not used as limitation of the invention.As long as according to this hair
Bright technical spirit is changed above-described embodiment, modification etc. will all be fallen in the scope of the claims of the present invention.
Claims (10)
1. a kind of device signboard detection method, which is characterized in that steps are as follows：
1) in the sample image to be trained comprising device signboard, device signboard is set to tab area, according to pre-
If neural network model and tab area, sample image is trained, weight model is obtained；
2) image to be detected is obtained, and image to be detected is divided into multiple regions；
3) according to weight model, each region of image to be detected is detected respectively, obtains the confidence level for corresponding to each region；
4) region that confidence level is greater than or equal to predetermined threshold value is returned to.
2. device signboard detection method according to claim 1, which is characterized in that in step 1), sample image includes
The original image that device signboard is shot by different angle under Different climate condition, different lightness environment.
3. device signboard detection method according to claim 1, which is characterized in that in step 1), the instruction of weight model
It is specific as follows to practice step：
1.1) sample image to be trained and the corresponding tab area of device signboard are obtained；
1.2) sample image to be trained is divided into multiple regions comprising the region of tab area is purpose area
Domain, other regions are background area；
1.3) study is trained to the pixel value in target area and background area by preset neural network model, obtained
Weight model.
4. device signboard detection method according to claim 3, which is characterized in that in step 1.3), first according to default
Neural network configuration file generate network object, by the network object of generation to the pixel in all areas of sample image
Value is trained study, and the parameter after training by iterative process optimization is preserved to weight model.
5. device signboard detection method according to claim 1, which is characterized in that in step 1), by manually in sample
Position mark is carried out to device signboard on this image, obtains tab area.
6. device signboard detection method according to claim 5, which is characterized in that in step 1), by sample to be trained
This image is built into training dataset by VOC standard data set formats, and specifically, steps are as follows：
First, tri- files of Annotation, ImageSets and JPEGImages are created, include under ImageSets files
Main files；
Then, sample image is named according to unified naming rule, sample image is then stored in JPEGImages
Under file；
Then, position mark is carried out to the device signboard in sample image using LabelImg tools, and generates xml document；
Wherein, an xml document corresponds to a sample image, and the content of storage includes annotation marks object, annotation
In include floder objects, fliename objects, path objects, size objects, multiple object objects；
Finally, the xml document of all generations is divided into two parts, the sample image title write-in in a part of xml document
For training in trainval.txt files, in the sample image title write-in test.txt files in another part xml document
For the verification in training process；
Trainval.txt files and test.txt files are stored under the Main files in ImageSet files；By institute
Some xml documents are stored under Annotation files.
7. device signboard detection method according to claim 1, which is characterized in that in step 4), it is big to return to confidence level
In or equal to predetermined threshold value region location information and its confidence level, the location information includes the x-axis at a certain angle in region
Coordinate, y-axis coordinate and peak width and height.
8. device signboard detection method according to claim 7, which is characterized in that believed by the position of return area
Breath, makes it carry out ring box in image to be detected and shows.
9. device signboard detection method according to claim 7, which is characterized in that confidence level is greater than or equal to default threshold
The region of value includes device signboard.
10. according to claim 1 to 9 any one of them device signboard detection method, which is characterized in that by weight model collection
At to background service, called using picture detection script, and provide back office interface to client；When client obtains mapping to be checked
Picture, and it is uploaded to back office interface, back office interface calls picture to detect script, and image to be detected is divided into multiple regions, according to
Parameter in weight model is detected and gives a mark to each region, obtains the confidence level in each region.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810576027.3A CN108764365A (en) | 2018-06-06 | 2018-06-06 | A kind of device signboard detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810576027.3A CN108764365A (en) | 2018-06-06 | 2018-06-06 | A kind of device signboard detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108764365A true CN108764365A (en) | 2018-11-06 |
Family
ID=63999251
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810576027.3A Pending CN108764365A (en) | 2018-06-06 | 2018-06-06 | A kind of device signboard detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108764365A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109711285A (en) * | 2018-12-11 | 2019-05-03 | 百度在线网络技术（北京）有限公司 | Training, test method and the device of identification model |
| CN110119718A (en) * | 2019-05-15 | 2019-08-13 | 燕山大学 | A kind of overboard detection and Survivable Control System based on deep learning |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120045132A1 (en) * | 2010-08-23 | 2012-02-23 | Sony Corporation | Method and apparatus for localizing an object within an image |
| CN102779267A (en) * | 2011-05-12 | 2012-11-14 | 株式会社理光 | Method and device for detection of specific object region in image |
| CN105447529A (en) * | 2015-12-30 | 2016-03-30 | 商汤集团有限公司 | Costume detection and attribute value identification method and system |
| CN106971185A (en) * | 2017-03-27 | 2017-07-21 | 深圳市丰巨泰科电子有限公司 | A kind of license plate locating method and device based on full convolutional network |
| CN107194398A (en) * | 2017-05-10 | 2017-09-22 | 平安科技（深圳）有限公司 | Car damages recognition methods and the system at position |
| CN107247955A (en) * | 2016-07-29 | 2017-10-13 | 竹间智能科技（上海）有限公司 | Accessory recognition methods and device |
| CN107305635A (en) * | 2016-04-15 | 2017-10-31 | 株式会社理光 | Object identifying method, object recognition equipment and classifier training method |
| CN107977686A (en) * | 2017-12-22 | 2018-05-01 | 浙江大学 | A kind of sorting technique of industrial X-ray image |
-
2018
- 2018-06-06 CN CN201810576027.3A patent/CN108764365A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120045132A1 (en) * | 2010-08-23 | 2012-02-23 | Sony Corporation | Method and apparatus for localizing an object within an image |
| CN102779267A (en) * | 2011-05-12 | 2012-11-14 | 株式会社理光 | Method and device for detection of specific object region in image |
| CN105447529A (en) * | 2015-12-30 | 2016-03-30 | 商汤集团有限公司 | Costume detection and attribute value identification method and system |
| CN107305635A (en) * | 2016-04-15 | 2017-10-31 | 株式会社理光 | Object identifying method, object recognition equipment and classifier training method |
| CN107247955A (en) * | 2016-07-29 | 2017-10-13 | 竹间智能科技（上海）有限公司 | Accessory recognition methods and device |
| CN106971185A (en) * | 2017-03-27 | 2017-07-21 | 深圳市丰巨泰科电子有限公司 | A kind of license plate locating method and device based on full convolutional network |
| CN107194398A (en) * | 2017-05-10 | 2017-09-22 | 平安科技（深圳）有限公司 | Car damages recognition methods and the system at position |
| CN107977686A (en) * | 2017-12-22 | 2018-05-01 | 浙江大学 | A kind of sorting technique of industrial X-ray image |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109711285A (en) * | 2018-12-11 | 2019-05-03 | 百度在线网络技术（北京）有限公司 | Training, test method and the device of identification model |
| CN110119718A (en) * | 2019-05-15 | 2019-08-13 | 燕山大学 | A kind of overboard detection and Survivable Control System based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107862303B (en) | Information identifying method, electronic device and the readable storage medium storing program for executing of form class diagram picture | |
| CN110852347B (en) | Fire detection method using improved YOLO v3 | |
| CN107818322A (en) | A kind of vehicle VIN code tampering detection system and methods for vehicle annual test | |
| CN105574550A (en) | Vehicle identification method and device | |
| CN107003977A (en) | System, method and apparatus for organizing the photo of storage on a mobile computing device | |
| CN108009515A (en) | A kind of power transmission line positioning identifying method of the unmanned plane image based on FCN | |
| CN105447522A (en) | Complex image character identification system | |
| CN105608454A (en) | Text structure part detection neural network based text detection method and system | |
| CN109800698A (en) | Icon detection method based on depth network | |
| CN113762269B (en) | Chinese character OCR recognition method, system and medium based on neural network | |
| CN110019912A (en) | Graphic searching based on shape | |
| CN110136130A (en) | A kind of method and device of testing product defect | |
| CN110222694A (en) | Image processing method, device, electronic equipment and computer-readable medium | |
| CN110288612A (en) | Nameplate positioning and bearing calibration and equipment | |
| CN113239753A (en) | Improved traffic sign detection and identification method based on YOLOv4 | |
| CN114677596A (en) | Remote sensing image ship detection method and device based on attention model | |
| CN108764365A (en) | A kind of device signboard detection method | |
| CN110390228A (en) | The recognition methods of traffic sign picture, device and storage medium neural network based | |
| CN109657728B (en) | Sample production method and model training method | |
| CN110263608A (en) | Electronic component automatic identifying method based on image feature space variable threshold value metric | |
| CN113657377A (en) | Structured recognition method for airplane ticket printing data image | |
| CN110766001B (en) | Bank card number positioning and end-to-end identification method based on CNN and RNN | |
| CN113065559B (en) | Image comparison method and device, electronic equipment and storage medium | |
| CN112949630B (en) | Weak supervision target detection method based on frame hierarchical screening | |
| Peng et al. | Application of deep residual neural network to water meter reading recognition |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181106 |
|
| RJ01 | Rejection of invention patent application after publication |