CN106874907B - A kind of method and device for establishing Car license recognition model - Google Patents
A kind of method and device for establishing Car license recognition model Download PDFInfo
- Publication number
- CN106874907B CN106874907B CN201710039025.6A CN201710039025A CN106874907B CN 106874907 B CN106874907 B CN 106874907B CN 201710039025 A CN201710039025 A CN 201710039025A CN 106874907 B CN106874907 B CN 106874907B
- Authority
- CN
- China
- Prior art keywords
- license plate
- plate image
- network model
- license
- neural network
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/63—Scene text, e.g. street names
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
Abstract
A kind of method and device for establishing Car license recognition model, wherein the method for establishing Car license recognition model, comprising: obtain the license plate image of multiple vehicles；The license plate image is expanded to pre-set dimension, obtains license plate image sample；Using the license board information in the multiple license plate image sample and the license plate image sample as training data, neural network model is trained, until the neural network model converges on preset value to the penalty values for the loss function that the discrimination of the license plate image sample license board information is greater than preset threshold or neural network model, then license board information identification is carried out to license plate image to be identified using trained neural network model, avoid cumbersome License Plate Segmentation process, solve the problems, such as that existing licence plate recognition method does not adapt to environment complicated and changeable and identification process is time-consuming serious.
Description
Technical field
The present invention relates to image identification technical fields, and in particular to a kind of method and device for establishing Car license recognition model.
Background technique
In intelligent transportation field, license plate recognition technology is occupied an important position.Traditional license plate recognition technology generally will
Car license recognition is divided into several big modules such as License Plate, License Plate Character Segmentation, Recognition of License Plate Characters.Existing licence plate recognition method is will
License plate picture Character segmentation after positioning is at several single characters, and then Recognition of License Plate Characters carries out the single character after cutting
Identification classification.
But existing licence plate recognition method has some disadvantages, and occurs being stained, incompleteness, be broken, adhesion for example, working as license plate
When, traditional character segmentation method faces huge challenge, and segmentation accuracy sharply declines, and directly results in Car license recognition mistake
It loses.Therefore, traditional Car license recognition does not adapt to environment complicated and changeable, and Car license recognition process is to the single character after cutting
Identification classification is carried out, a large amount of digital image-processing methods are taken, it is time-consuming serious.
Summary of the invention
Therefore, the technical problem to be solved in the present invention is that existing licence plate recognition method does not adapt to environment complicated and changeable
And identification process is time-consuming serious.
In view of this, the present invention provides a kind of method for establishing Car license recognition model, comprising:
Obtain the license plate image of multiple vehicles；
The license plate image is expanded to pre-set dimension, obtains license plate image sample；
Using the license board information in the multiple license plate image sample and the license plate image sample as training data, to mind
It is trained through network model, until the neural network model is greater than the discrimination of the license plate image sample license board information
The penalty values of the loss function of first preset threshold or neural network model converge on preset value.
Preferably, the neural network model is convolutional neural networks model.
Preferably, the neural network model is provided with multiple character result output layers, the multiple character result output
Layer respectively corresponds the recognition result for exporting multiple characters in the license plate image sample；
The license board information using in the multiple license plate image sample and the license plate image sample as training data,
Neural network model is trained, until the neural network model is big to the discrimination of license board information in license plate image sample
Preset value is converged in the penalty values of preset threshold or the loss function of neural network model, comprising:
Using the license board information in the multiple license plate image sample and the license plate image sample as training data, to mind
It is trained through network model, until multiple character result output layers of the neural network model are to more in license plate image sample
The penalty values that the discrimination of the recognition result of a character is greater than the loss function of preset threshold or neural network model converge on
Preset value.
Preferably, the neural network model is provided with length sequences output layer, for exporting the license plate image sample
Middle character length recognition result.
Preferably, described that the license plate image is expanded to pre-set dimension, obtain license plate image sample, comprising:
Obtain the coordinate and size of the license plate image；
According to the coordinate and size of the license plate image, the image center of the license plate image is determined；
Centered on described image central point, the license plate image extension is carried out according to the pre-set dimension, is obtained described
License plate image sample.
Correspondingly, the present invention also provides a kind of devices for establishing Car license recognition model, comprising:
Acquiring unit, for obtaining the license plate image of multiple vehicles；
License plate image sample acquisition unit obtains license plate image sample for the license plate image to be expanded to pre-set dimension
This；
Training unit, for using the license board information in the multiple license plate image sample and the license plate image sample as
Training data is trained neural network model, until the neural network model is to license plate in the license plate image sample
The penalty values that the discrimination of information is greater than the loss function of the first preset threshold or neural network model converge on preset value.
Preferably, the neural network model is convolutional neural networks model.
Preferably, the neural network model is provided with multiple character result output layers, the multiple character result output
Layer respectively corresponds the recognition result for exporting multiple characters in the license plate image sample；
The training unit, comprising: training subelement is used for
Using the license board information in the multiple license plate image sample and the license plate image sample as training data, to mind
It is trained through network model, until multiple character result output layers of the neural network model are to more in license plate image sample
The penalty values that the discrimination of the recognition result of a character is greater than the loss function of preset threshold or neural network model converge on
Preset value.
Preferably, the neural network model is provided with length sequences output layer, for exporting the license plate image sample
Middle character length recognition result.
Preferably, the license plate image sample acquisition unit includes:
Coordinate and size acquiring unit, for obtaining the coordinate and size of the license plate image；
Image center determination unit determines the license plate for the coordinate and size according to the license plate image
The image center of image；
License plate image sample determination unit, for being carried out according to the pre-set dimension centered on described image central point
The license plate image extension, obtains the license plate image sample.
Technical solution of the present invention has the advantage that
By the license plate image of the multiple vehicles of acquisition, license plate image is expanded to pre-set dimension, obtains license plate image sample,
And using the license board information in multiple license plate image samples and license plate image sample as training data, neural network model is carried out
Training, until neural network model is greater than the first preset threshold or nerve net to the discrimination of license plate image sample license board information
The penalty values of the loss function of network model converge on preset value, then using trained neural network model to license plate to be identified
Image carries out license board information identification, solves existing licence plate recognition method and does not adapt to environment complicated and changeable and identification process consumption
The problem of Shi Yanchong.
Detailed description of the invention
It, below will be to specific in order to illustrate more clearly of the specific embodiment of the invention or technical solution in the prior art
Embodiment or attached drawing needed to be used in the description of the prior art be briefly described, it should be apparent that, it is described below
Attached drawing is some embodiments of the present invention, for those of ordinary skill in the art, before not making the creative labor
It puts, is also possible to obtain other drawings based on these drawings.
Fig. 1 is a kind of flow chart for method for establishing Car license recognition model that the embodiment of the present invention 1 provides；
Fig. 2 is a kind of structural schematic diagram for device for establishing Car license recognition model that the embodiment of the present invention 2 provides；
Fig. 3 is a kind of flow chart for license board information recognition methods that the embodiment of the present invention 3 provides；
Fig. 4 is a kind of structural schematic diagram for license board information identification device that the embodiment of the present invention 4 provides.
Specific embodiment
Technical solution of the present invention is clearly and completely described below in conjunction with attached drawing, it is clear that described implementation
Example is a part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill
Personnel's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Embodiment 1
The embodiment of the present invention provides a kind of method for establishing Car license recognition model, as shown in Figure 1, comprising:
S11 obtains the license plate image of multiple vehicles.Plurality of license plate image sample includes positive sample and negative sample, just
Sample be true license plate image, negative sample be non-genuine license plate image, i.e., negative sample can be other positions of vehicle image or
The license plate image of person's incompleteness, is labeled Massive Sample, and marked content includes license plate number and/or characters on license plate length, vacation
All characters of board are all labeled as 0.All samples are finally zoomed to the input layer size of neural network model, the present embodiment mind
Size through network model input layer is 128x128.The acquisition of license plate image carries out License Plate using ACF detection algorithm.
License plate image is expanded to pre-set dimension by S12, obtains license plate image sample.The license plate that ACF detection algorithm detects
There may be the small probability event situations for not including complete license plate area, so that license board information identification is inaccurate, in order to
Training result accuracy is improved, the license plate image region that will acquire amplifies, specifically includes the following steps:
S121 obtains the coordinate and size of license plate image；License plate image is wherein obtained by ACF detection algorithm, benefit
The coordinate and picture size size of license plate image are obtained with minimum circumscribed rectangle method.
S122 determines the image center of license plate image according to the coordinate and size of license plate image；
S123 is carried out license plate image extension according to pre-set dimension, is obtained license plate image sample centered on image center
This.Wherein the size of the license plate image sample is greater than license plate image size, and pre-set dimension can be the license plate image of acquisition
Maximum value in length and width, or according to obtaining the actual size statistical result of license plate image as propagation size, wherein
Statistical data can be the average value of random license plate image dimension data.
S13, using the license board information in multiple license plate image samples and license plate image sample as training data, to nerve net
Network model is trained, until neural network model is greater than the first preset threshold to the discrimination of license plate image sample license board information
Or the penalty values of the loss function of neural network model converge on preset value.The penalty values are for calculating neural network model
Convergence rate, when the penalty values of loss function change in a certain range of preset value, i.e., it is believed that loss function value not
Decline again, wherein neural network model is convolutional neural networks model, and obtained license plate image sample is input to input layer, is passed through
Several layers convolutional layer and pond layer are crossed, the embodiment of the present invention uses 4 layers of convolutional layer and 4 layers of pond layer alternating action, obtains one
The feature vector H of 4096 dimensions.The convolution kernel of all convolutional layers is 3, and step-length 1, pond layer takes maximum value pond mode, pond
Change window size is 2x2, step-length 2.The port number of convolutional layer and full articulamentum be respectively (32,64,128,256,500,
4096) feature vector of 4096 dimensions, is obtained.
In order to improve license board information recognition speed, neural network model is provided with multiple character result output layers simultaneously, more
A character result output layer respectively corresponds character identification result in output license plate image sample.
Step S13 is specifically included:
Using the license board information in multiple license plate image samples and license plate image sample as training data, to neural network mould
Type is trained, until each character result output layer of neural network model is to character recognition in license plate image sample license board information
As a result the penalty values that discrimination is greater than the loss function of the first preset threshold or neural network model converge on preset value.Its
Character identification result includes the recognition result of character in the length recognition result and license plate of license board information in middle license board information；
As an alternative embodiment, the neural network model is provided with length sequences output layer, for exporting
Character length recognition result in the license plate image sample.
Specific training method is repeated no more with the character result training process of above-mentioned steps S12.
Specifically, the feature vector of 4096 dimensions of the present embodiment can be set multiple output layers and be connected, and obtain multiple outputs
As a result, work as output layer settable 6 output layers of number, for identification license plate image letter and digital information or 7 it is defeated
Layer out, for identification the Chinese character information of license plate and letter with digital information either 8 output layers, identification license plate Chinese character
Information and letter and digital information can also identify the length of license plate.Characters on license plate length is generally 7 or 8, because
This, character length output result is set as tri- classification task of CNN, and the license plate digit of the first branching representation identification is 0, indicates license plate
Identify that mistake, the license plate digit of the second branching representation identification are 7, the license plate digit of third branching representation identification is 8；Output
Layer is that first branches into the 1st chinese character recognition result of output license plate, chinese character identification to the output result of character
Branch exports result integer value (representing 34 region Chinese characters) between 0-34；Rear the 6 of second branch output license plate or 7 numbers
The recognition result of code, license plate number identification branch output result arbitrary value between 0-36 (represent 0-9, A-Z totally 36 numeric words
Alphabetic character and false-trademark), when its output can be expressed as ZSi=WSiH+bSi, (i=1,2,3 ... 6,7,8), wherein W indicates connection
Weight, bsi be biasing, wherein the convolutional neural networks model use classification corresponding to the output valve of maximum a posteriori probability for
Final output value, it may be assumed that
S=(l, s1, s2, s3, s4, s5, s6, s7, s8)=arg maxL,S1,S2,S3,...,S8log P(S|X)
The method provided in an embodiment of the present invention for establishing Car license recognition model, by obtaining the license plate image of multiple vehicles,
License plate image is expanded to pre-set dimension, obtains license plate image sample, and by multiple license plate image samples and license plate image sample
In license board information as training data, neural network model is trained, until neural network model is to license plate image sample
The penalty values that the discrimination of this license board information is greater than the loss function of the first preset threshold or neural network model converge on pre-
If value, license board information identification then is carried out to license plate image to be identified using trained neural network model, is solved existing
Licence plate recognition method does not adapt to environment complicated and changeable and the time-consuming serious problem of identification process.
Embodiment 2
The embodiment of the present invention provides a kind of device for establishing Car license recognition model, as shown in Figure 2, comprising:
Acquiring unit 21, for obtaining the license plate image of multiple vehicles；
License plate image sample acquisition unit 22 obtains license plate image sample for license plate image to be expanded to pre-set dimension；
Training unit 23, for using the license board information in multiple license plate image samples and license plate image sample as training number
According to being trained to neural network model, until neural network model is big to the discrimination of license board information in license plate image sample
Preset value is converged in the penalty values of the first preset threshold or the loss function of neural network model.
Preferably, neural network model is convolutional neural networks model.
Preferably,
The neural network model is provided with multiple character result output layers, and the multiple character result output layer is right respectively
The recognition result of multiple characters in the license plate image sample should be exported；
The training unit, comprising: training subelement is used for
Using the license board information in the multiple license plate image sample and the license plate image sample as training data, to mind
It is trained through network model, until multiple character result output layers of the neural network model are to more in license plate image sample
The penalty values that the discrimination of the recognition result of a character is greater than the loss function of preset threshold or neural network model converge on
Preset value.
Preferably, the neural network model is provided with length sequences output layer, for exporting the license plate image sample
Middle character length recognition result.
Preferably, license plate image sample acquisition unit includes:
Coordinate and size acquiring unit, for obtaining the coordinate and size of license plate image；
Image center determination unit determines the license plate image for the coordinate and size according to license plate image
Image center；
License plate image sample determination unit, for carrying out license plate image according to pre-set dimension centered on image center
Extension, obtains license plate image sample.
The device provided in an embodiment of the present invention for establishing Car license recognition model, the vehicle of multiple vehicles is obtained by acquiring unit
License plate image is expanded to pre-set dimension by board image, obtains license plate image sample, and by multiple license plate image samples and license plate figure
License board information in decent is trained neural network model as training data, until neural network model is to license plate
The penalty values that the discrimination of image pattern license board information is greater than the loss function of the first preset threshold or neural network model are received
It holds back in preset value, license board information identification then is carried out to license plate image to be identified using trained neural network model, is solved
Existing licence plate recognition method does not adapt to environment complicated and changeable and the time-consuming serious problem of identification process.
Embodiment 3
The embodiment of the present invention provides a kind of license board information recognition methods, as shown in Figure 3, comprising:
S31 obtains license plate image to be identified；
License plate image to be identified is input in the model that method as described in Example 1 is established, determines to be identified by S32
The license board information of license plate image.For example, a license board information is capital AAB123, the machine training identification of the 6th character " 1 "
In the process, poll identifies number 0-9, obtained maximum a posteriori probability is respectively 0.001,0.001,0.99,0.001,0.001,
0.001,0.001,0.001,0.001,0.001,0.001,0.001, then when identifying that 0.99 corresponding digital 1 is license plate
6th is character, similarly obtains the information of other characters of license plate by the method for maximum a posteriori probability, and images to be recognized is each
The recognition result of position is 0.99,0.97,0.98,0.96,0.99,0.99,0.99,0.98,0.95, will then be obtained on each
Result even multiply, when even multiplying result greater than the second preset threshold, then images to be recognized is license plate image, and the present embodiment is excellent
Selecting the second preset threshold is 0.3；When the recognition result of a certain position is far smaller than the recognition result on other, at this time it can be concluded that
Images to be recognized identification mistake is recognized verifying.When license plate length recognition result is to have in 0 and character identification result
Output result greater than 4 is 0, then determines images to be recognized for false-trademark.
License board information recognition methods provided in an embodiment of the present invention, by trained machine learning model to vehicle to be identified
Board image improves the accuracy rate and recognition speed of the identification of license plate image information.
Embodiment 4
The embodiment of the present invention provides a kind of license board information identification device, as shown in Figure 4, comprising:
License plate image acquiring unit 41 to be identified, for obtaining license plate image to be identified；
License board information determination unit 42, for the license plate image to be identified to be input to method as described in Example 1
In the model of foundation, the license board information of license plate image to be identified is determined.
License board information identification device provided in an embodiment of the present invention, by trained machine learning model to vehicle to be identified
Board image improves the accuracy rate and recognition speed of the identification of license plate image information.
Obviously, the above embodiments are merely examples for clarifying the description, and does not limit the embodiments.It is right
For those of ordinary skill in the art, can also make on the basis of the above description it is other it is various forms of variation or
It changes.There is no necessity and possibility to exhaust all the enbodiments.And it is extended from this it is obvious variation or
It changes still within the protection scope of the invention.
Claims (8)
1. a kind of method for establishing Car license recognition model characterized by comprising
Obtain the license plate image of multiple vehicles；
The license plate image is expanded to pre-set dimension, obtains license plate image sample；
Using the license board information in the multiple license plate image sample and the license plate image sample as training data, to nerve net
Network model is trained, and is preset until the neural network model is greater than the discrimination of the license plate image sample license board information
The penalty values of the loss function of threshold value or neural network model converge on preset value；
Wherein, the neural network model is provided with multiple character result output layers, the multiple character result output layer difference
The corresponding recognition result for exporting multiple characters in the license plate image sample；
The license board information using in the multiple license plate image sample and the license plate image sample is as training data, to mind
It is trained through network model, until the neural network model is greater than in advance the discrimination of license board information in license plate image sample
If the penalty values of the loss function of threshold value or neural network model converge on preset value, comprising:
Using the license board information in the multiple license plate image sample and the license plate image sample as training data, to nerve net
Network model is trained, until multiple character result output layers of the neural network model are to multiple words in license plate image sample
The penalty values that the discrimination of the recognition result of symbol is greater than the loss function of preset threshold or neural network model converge on default
Value.
2. the method according to claim 1, wherein the neural network model is convolutional neural networks model.
3. the method according to claim 1, wherein the neural network model is provided with length sequences output
Layer, for exporting character length recognition result in the license plate image sample.
4. being obtained the method according to claim 1, wherein described be expanded to pre-set dimension for the license plate image
To license plate image sample, comprising:
Obtain the coordinate and size of the license plate image；
According to the coordinate and size of the license plate image, the image center of the license plate image is determined；
Centered on described image central point, the license plate image extension is carried out according to the pre-set dimension, obtains the license plate
Image pattern.
5. a kind of device for establishing Car license recognition model characterized by comprising
Acquiring unit, for obtaining the license plate image of multiple vehicles；
License plate image sample acquisition unit obtains license plate image sample for the license plate image to be expanded to pre-set dimension；
Training unit, for using the license board information in the multiple license plate image sample and the license plate image sample as training
Data are trained neural network model, until the neural network model is to license board information in the license plate image sample
The penalty values of the discrimination loss function that is greater than preset threshold or neural network model converge on preset value；
The neural network model is provided with multiple character result output layers, and the multiple character result output layer respectively corresponds defeated
Out in the license plate image sample multiple characters recognition result；
The training unit, comprising:
Training subelement, for using the license board information in the multiple license plate image sample and the license plate image sample as instruction
Practice data, neural network model is trained, until multiple character result output layers of the neural network model are to license plate
The discrimination of the recognition result of multiple characters is greater than the loss function of preset threshold or neural network model in image pattern
Penalty values converge on preset value.
6. device according to claim 5, which is characterized in that the neural network model is convolutional neural networks model.
7. device according to claim 5, which is characterized in that the neural network model is provided with length sequences output
Layer, for exporting character length recognition result in the license plate image sample.
8. device according to claim 5, which is characterized in that the license plate image sample acquisition unit includes:
Coordinate and size acquiring unit, for obtaining the coordinate and size of the license plate image；
Image center determination unit determines the license plate image for the coordinate and size according to the license plate image
Image center；
License plate image sample determination unit is used for centered on described image central point, according to described in pre-set dimension progress
License plate image extension, obtains the license plate image sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710039025.6A CN106874907B (en) | 2017-01-19 | 2017-01-19 | A kind of method and device for establishing Car license recognition model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710039025.6A CN106874907B (en) | 2017-01-19 | 2017-01-19 | A kind of method and device for establishing Car license recognition model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106874907A CN106874907A (en) | 2017-06-20 |
| CN106874907B true CN106874907B (en) | 2019-08-27 |
Family
ID=59157949
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710039025.6A Active CN106874907B (en) | 2017-01-19 | 2017-01-19 | A kind of method and device for establishing Car license recognition model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106874907B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107220675A (en) * | 2017-06-21 | 2017-09-29 | 苏州卡睿知光电科技有限公司 | A kind of method and device for setting up eyeglass identification model and eyeglass identification |
| CN107330878A (en) * | 2017-06-21 | 2017-11-07 | 苏州卡睿知光电科技有限公司 | A kind of method and device for setting up eyeglass identification model |
| CN108334881B (en) * | 2018-03-12 | 2022-04-29 | 南京云创大数据科技股份有限公司 | License plate recognition method based on deep learning |
| WO2019175686A1 (en) | 2018-03-12 | 2019-09-19 | Ratti Jayant | On-demand artificial intelligence and roadway stewardship system |
| CN108510084B (en) * | 2018-04-04 | 2022-08-23 | 百度在线网络技术（北京）有限公司 | Method and apparatus for generating information |
| CN111046891A (en) * | 2018-10-11 | 2020-04-21 | 杭州海康威视数字技术股份有限公司 | Training method of license plate recognition model, and license plate recognition method and device |
| CN110443245B (en) * | 2019-08-14 | 2022-02-15 | 上海世茂物联网科技有限公司 | License plate region positioning method, device and equipment in non-limited scene |
| CN112580643A (en) * | 2020-12-09 | 2021-03-30 | 浙江智慧视频安防创新中心有限公司 | License plate recognition method and device based on deep learning and storage medium |
| CN112801093A (en) * | 2021-01-11 | 2021-05-14 | 爱泊车美好科技有限公司 | License plate information calibration method and device |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104298976A (en) * | 2014-10-16 | 2015-01-21 | 电子科技大学 | License plate detection method based on convolutional neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI384408B (en) * | 2009-04-30 | 2013-02-01 | Ind Tech Res Inst | Method and system for identifying image and outputting identification result |
-
2017
- 2017-01-19 CN CN201710039025.6A patent/CN106874907B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104298976A (en) * | 2014-10-16 | 2015-01-21 | 电子科技大学 | License plate detection method based on convolutional neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106874907A (en) | 2017-06-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106874907B (en) | A kind of method and device for establishing Car license recognition model | |
| CN106874902A (en) | A kind of license board information recognition methods and device | |
| CN109543606B (en) | Human face recognition method with attention mechanism | |
| CN110298266B (en) | Deep neural network target detection method based on multiscale receptive field feature fusion | |
| CN110309880B (en) | Method for classifying images of 5-day and 9-day incubated egg embryos based on attention mechanism CNN | |
| CN110927706B (en) | Convolutional neural network-based radar interference detection and identification method | |
| CN103971102B (en) | Static gesture recognition method based on finger contour and decision-making trees | |
| CN110032954B (en) | Intelligent identification and counting method and system for reinforcing steel bars | |
| CN104850858B (en) | A kind of injection-molded item defects detection recognition methods | |
| CN105913087B (en) | Object identification method based on optimal pond convolutional neural networks | |
| CN109271960A (en) | A kind of demographic method based on convolutional neural networks | |
| CN108595585B (en) | Sample data classification method, model training method, electronic equipment and storage medium | |
| CN104463195B (en) | Printing digit recognizing method based on template matches | |
| CN107657249A (en) | Method, apparatus, storage medium and the processor that Analysis On Multi-scale Features pedestrian identifies again | |
| CN108615034A (en) | A kind of licence plate recognition method that template matches are combined with neural network algorithm | |
| CN108562589A (en) | A method of magnetic circuit material surface defect is detected | |
| CN111860459B (en) | Gramineae plant leaf pore index measurement method based on microscopic image | |
| CN106447020A (en) | Intelligent bacterium colony counting method | |
| CN109583357A (en) | A kind of improvement LBP and the cascade face identification method of light weight convolutional neural networks | |
| CN106680775A (en) | Method and system for automatically identifying radar signal modulation modes | |
| CN108805833A (en) | Miscellaneous minimizing technology of copybook binaryzation ambient noise of network is fought based on condition | |
| CN110232318A (en) | Acupuncture point recognition methods, device, electronic equipment and storage medium | |
| CN109886391A (en) | A kind of neural network compression method based on the positive and negative diagonal convolution in space | |
| CN109255336A (en) | Arrester recognition methods based on crusing robot | |
| CN110599463A (en) | Tongue image detection and positioning algorithm based on lightweight cascade neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |