CN110033052A - A kind of the self-training method and self-training platform of AI identification hand-written script - Google Patents
A kind of the self-training method and self-training platform of AI identification hand-written script Download PDFInfo
- Publication number
- CN110033052A CN110033052A CN201910317078.9A CN201910317078A CN110033052A CN 110033052 A CN110033052 A CN 110033052A CN 201910317078 A CN201910317078 A CN 201910317078A CN 110033052 A CN110033052 A CN 110033052A
- Authority
- CN
- China
- Prior art keywords
- hand
- written script
- written
- identification
- data set
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Telephone Function (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention discloses a kind of self-training method of AI identification hand-written script, is related to image identification technical field.Personalized hand-written script aiming at the problem that practical application is difficult to well be identified.General hand-written script data set using technical solution based on cloud and storage beyond the clouds, general hand-written script identification model is constructed using general hand-written script data set beyond the clouds first, then acquire and handle the hand-written script image of user, and the general hand-written script identification model of training, until when being not less than threshold value set by user using the accuracy rate of general hand-written script identification model verifying hand-written script image, complete the training of the dedicated hand-written script identification model of user, utilize the dedicated hand-written script identification model of user, it can be to all hand-written notes of the user, the images such as document are digitally converted and identify.Invention additionally discloses a kind of self-training platforms of AI identification hand-written script, are installed on mobile terminal, combine with above-mentioned self-training method.
Description
Technical field
The present invention relates to image identification technical field, the self-training method of specifically a kind of AI identification hand-written script and
Self-training platform.
Background technique
Hand-written script identification be OCR technique one kind, obtained quick development in recent years, also certain some industry into
Popularization and application of having gone carry out the training of model, in limited conditions range largely by the way of machine learning, deep learning
Interior discrimination can achieve higher level.But since hand-written script has personal personality, so by the end of at present still
Fail above have good recognition effect in application, also fails in market wide popularization and application.Hand-written script identification product is not present
Technology barrier, but due to its limited training data, training area, application model etc., become the maximum bottleneck of its development.With
The development of science and technology, digital at and use demand it is more urgent, during development, how by traditional hand-written data with
Electronic data combines conversion, is the problem for needing to break through.
Summary of the invention
The present invention is directed to the demand and shortcoming of current technology development, provides a kind of self-training of AI identification hand-written script
Method and self-training platform are difficult to reach asking for good recognition effect in practical applications to solve personalized hand-written script
Topic.
Firstly, the present invention protects a kind of self-training method of AI identification hand-written script, this method is based on cloud and is stored in
The general hand-written script data set in cloud, the realization step of this method include:
1) general hand-written script identification model is constructed using general hand-written script data set；
2) building hand-written script identifies validation data set and dedicated hand-written script data set beyond the clouds；
3) the hand-written script image of user is acquired, and handles the specification number that can be identified for general hand-written script identification model
According to, be subsequently stored in hand-written script identification validation data set；
4) data of hand-written script identification validation data set are inputted general hand-written script identification model to verify, user
Voluntarily check verification result, when verification result is not consistent with input data, user voluntarily corrects verification result, then will input
Verification result after data and correction is stored in dedicated hand-written script data set；
5) circulation executes step 3), 4), and given threshold acquires and handles the new hand-written script image of user and is iterated
Verifying, until the verifying accuracy rate of general hand-written script identification model is not less than threshold value set by user；
6) at this point, the general hand-written script identification model of output is the dedicated hand-written script identification model of user.
Specifically, the concrete operations of processing hand-written script image include:
Processing is split to the hand-written script image of acquisition, forms single font image；
Single font image is pre-processed, noise reduction, digitlization conversion, being converted to general hand-written script identification model can
With the authority data of identification, it is subsequently stored in hand-written script identification validation data set.
Specifically, needing the hand-written script image of multi collect user, and the whole hand-written scripts acquired in step 3)
Image preferably covers all texts that general hand-written script identification model can identify.
Specifically, involved hand-written script identification validation data set includes multiple hand-written script identification verifying Sub Data Sets,
All hand-written scripts of the same text are stored in the same hand-written script identification verifying Sub Data Set, the handwritten word of different literals
Body is stored in different hand-written script identification verifying Sub Data Sets.
Specifically, involved dedicated hand-written script data set includes multiple dedicated hand-written script Sub Data Sets, the same text
All hand-written scripts of word are stored in the same dedicated hand-written script Sub Data Set, and the hand-written script of different literals is stored in difference
Dedicated hand-written script Sub Data Set.
Secondly, the present invention also protects a kind of self-training platform of AI identification hand-written script, which is based on passing through
The mobile terminal and cloud that network is communicated, cloud are stored with general hand-written script data set, and structure includes:
Building module one beyond the clouds is set, is identified for constructing general hand-written script using general hand-written script data set
Model；
Building module two beyond the clouds is set, for constructing hand-written script identification validation data set and dedicated hand-written script number
According to collection；
The acquisition module of mobile terminal is set, for acquiring the hand-written script image of user；
The processing module of mobile terminal is set, and the hand-written script image procossing for that will acquire is the identification of general hand-written script
The authority data that model can identify, is also used to that authority data is sent to cloud by treated, and is stored in the hand-written of cloud
Character Font Recognition validation data set；
Extraction input module beyond the clouds is set, for extracting the data of hand-written script identification validation data set, and is inputted
General hand-written script identification model, general hand-written script identification model verify input data, artificial decision verification result；
The correction module of mobile terminal is set, and when artificial decision verification result is incorrect, user is entangled by correcting module
Positive verification result, then the verification result by input data and after correcting is stored in dedicated hand-written script data set；
The threshold module of mobile terminal is set, the threshold for user's setting general-purpose hand-written script identification model verifying accuracy rate
Value；
Calculating judgment module beyond the clouds is set, for calculating the verifying accuracy rate of general hand-written script identification model, and
Judge whether the verifying accuracy rate of general hand-written script identification model is not less than threshold value set by user, if so, by general hand
Body identification model of writing is exported as the dedicated hand-written script identification model of user to mobile terminal, if it is not, then continuing to acquire user
Hand-written script image.
Specifically, involved processing module includes:
Divide submodule, for being split processing to the hand-written script image of acquisition to form single font image；
Submodule is pre-processed, for pre-processing to single font image；
Noise reduction submodule, for carrying out noise reduction process to single font image；
Transform subblock is digitized, is identified for the single font image after noise reduction process to be converted to general hand-written script
The authority data that model can identify；
Sending submodule, the authority data for converting single font image is sent to cloud, and is stored in cloud
Hand-written script identifies validation data set；
Specifically, whole hand-written script images of acquisition module acquisition cover general handwritten word for the same user
All texts that body identification model can identify.
Specifically, involved hand-written script identification validation data set includes multiple hand-written script identification verifying Sub Data Sets,
All hand-written scripts of the same text are stored in the same hand-written script identification verifying Sub Data Set, the handwritten word of different literals
Body is stored in different hand-written script identification verifying Sub Data Sets；
Involved dedicated hand-written script data set includes multiple dedicated hand-written script Sub Data Sets, and the same text owns
Hand-written script is stored in the same dedicated hand-written script Sub Data Set, and the hand-written script of different literals is stored in different dedicated hands
Write body Sub Data Set.
Optionally, involved mobile terminal can be mobile phone terminal, the end PC, the end PAD；Acquisition module passes through the side taken pictures or downloaded
Formula acquires hand-written script image.
The self-training method and self-training platform of a kind of AI identification hand-written script of the invention, have compared with prior art
Beneficial effect be:
1) self-training method of the invention and self-training platform, the general hand-written script number based on cloud and storage beyond the clouds
According to collection, general hand-written script identification model is constructed, followed by the general handwritten word of hand-written script image training for acquiring and handling
Body identification model, until when the verifying accuracy rate using general hand-written script identification model is not less than threshold value set by user, i.e.,
The training for completing the dedicated hand-written script identification model of user can be to this using the dedicated hand-written script identification model of user
The images such as all hand-written notes of user, document are digitally converted and identify, solve personalized handwritten word from the root
Body is difficult to reach the problem of good recognition effect in practical applications；
2) self-training method of the invention and self-training platform can be convenient all users and carry out self-training, self-training mistake
The verification result that journey needs user to acquire the hand-written script image of oneself, correct mistake, in the hand-written script image of user's acquisition
When the verifying accuracy rate of enough and general hand-written script identification model is not less than threshold value set by user, so that it may obtain user
Dedicated hand-written script identification model, whole process is simple and convenient, is able to satisfy the quick knowledge of different user personalization hand-written script
Not.
Detailed description of the invention
Attached drawing 1 is the connection block diagram of the embodiment of the present invention two.
Each label information indicates in attached drawing:
A, cloud, B, mobile terminal；
A, general hand-written script data set, b, hand-written script identify validation data set,
C, dedicated hand-written script data set；
10, building module one, 20, building module two, 30, acquisition module, 40, processing module,
50, extraction input module, 60, correction module, 70, calculating judgment module,
80, general hand-written script identification model, 90, dedicated hand-written script identification model,
100, threshold module；
41, segmentation submodule, 42, pretreatment submodule, 43, noise reduction submodule,
44, transform subblock, 45, sending submodule are digitized.
Specific embodiment
The technical issues of to make technical solution of the present invention, solving and technical effect are more clearly understood, below in conjunction with tool
Body embodiment is checked technical solution of the present invention, is completely described, it is clear that described embodiment is only this hair
Bright a part of the embodiment, instead of all the embodiments.Based on the embodiment of the present invention, those skilled in the art are not doing
All embodiments obtained under the premise of creative work out, all within protection scope of the present invention.
Embodiment one:
The present embodiment proposes that a kind of self-training method of AI identification hand-written script, this method are based on cloud A and are stored in cloud
The general hand-written script data set a of A is held, the realization step of this method includes:
1) general hand-written script identification model 80 is constructed using general hand-written script data set a；
2) A constructs hand-written script and identifies validation data set b and dedicated hand-written script data set c beyond the clouds；
3) the hand-written script image of user is acquired, and handles the specification that can be identified for general hand-written script identification model 80
Data are subsequently stored in hand-written script identification validation data set b；
4) data of hand-written script identification validation data set b general hand-written script identification model 80 is inputted to verify,
User voluntarily checks verification result, and when verification result is not consistent with input data, user voluntarily corrects verification result, then will
Verification result after input data and correction is stored in dedicated hand-written script data set c；
5) circulation executes step 3), 4), and given threshold acquires and handles the new hand-written script image of user and is iterated
Verifying, until the verifying accuracy rate of general hand-written script identification model 80 is not less than threshold value set by user；
6) at this point, the general hand-written script identification model 80 of output is the dedicated hand-written script identification model 90 of user.
In the present embodiment, handle hand-written script image concrete operations include:
Processing is split to the hand-written script image of acquisition, forms single font image；
Single font image is pre-processed, noise reduction, digitlization conversion, is converted to general hand-written script identification model 80
The authority data that can be identified is subsequently stored in hand-written script identification validation data set b.
In the present embodiment, when executing step 3), the hand-written script image of multi collect user, and the whole acquired are needed
Hand-written script image preferably covers all texts that general hand-written script identification model 80 can identify.
In the present embodiment, involved hand-written script identification validation data set b includes multiple hand-written script identification verifying
Data set, all hand-written scripts of the same text are stored in the same hand-written script identification verifying Sub Data Set, different literals
Hand-written script be stored in different hand-written script identification verifying Sub Data Sets.
In the present embodiment, involved dedicated hand-written script data set c includes multiple dedicated hand-written script Sub Data Sets, together
All hand-written scripts of one text are stored in the same dedicated hand-written script Sub Data Set, the hand-written script storage of different literals
In different dedicated hand-written script Sub Data Sets.
Embodiment two:
In conjunction with attached drawing 1, the present embodiment proposes that a kind of self-training platform of AI identification hand-written script, the self-training platform are based on
The mobile terminal B and cloud A communicated by network, cloud A are stored with general hand-written script data set a, and structure includes:
The building module 1 of A beyond the clouds is set, for constructing general hand-written script using general hand-written script data set a
Identification model 80；
The building module 2 20 of A beyond the clouds is set, for constructing hand-written script identification validation data set b and dedicated handwritten word
Volumetric data set c；
The acquisition module 30 of mobile terminal B is set, for acquiring the hand-written script image of user；
The processing module 40 of mobile terminal B is set, and the hand-written script image procossing for that will acquire is general hand-written script
The authority data that identification model 80 can identify, is also used to that authority data is sent to cloud A by treated, and is stored in cloud
The hand-written script of A identifies validation data set b；
The extraction input module 50 of A beyond the clouds is set, for extracting the data of hand-written script identification validation data set b, and
General hand-written script identification model 80 is inputted, general hand-written script identification model 80 verifies input data, artificial decision verification knot
Fruit；
The correction module 60 of mobile terminal B is set, and when artificial decision verification result is incorrect, user is by correcting module
60 correct verification result, and then the verification result by input data and after correcting is stored in dedicated hand-written script data set c；
The threshold module 100 of mobile terminal B is set, is verified accurately for user's setting general-purpose hand-written script identification model 80
The threshold value of rate；
The calculating judgment module 70 of A beyond the clouds is set, and the verifying for calculating general hand-written script identification model 80 is accurate
Rate, and judge whether the verifying accuracy rate of general hand-written script identification model 80 is not less than threshold value set by user, if so, will
General hand-written script identification model 80 as user the output of dedicated hand-written script identification model 90 to mobile terminal B, if it is not, then after
The hand-written script image of continuous acquisition user.
In the present embodiment, involved processing module 40 includes:
Divide submodule 41, for being split processing to the hand-written script image of acquisition to form single font image；
Submodule 42 is pre-processed, for pre-processing to single font image；
Noise reduction submodule 43, for carrying out noise reduction process to single font image；
Transform subblock 44 is digitized, is known for the single font image after noise reduction process to be converted to general hand-written script
The authority data that other model 80 can identify；
Sending submodule 45, the authority data for converting single font image is sent to cloud A, and is stored in cloud
The hand-written script of A identifies validation data set b；
In the present embodiment, for the same user, whole hand-written script images covering that acquisition module 30 acquires is logical
All texts that can be identified with hand-written script identification model 80.
In the present embodiment, involved hand-written script identification validation data set b includes multiple hand-written script identification verifying
Data set, all hand-written scripts of the same text are stored in the same hand-written script identification verifying Sub Data Set, different literals
Hand-written script be stored in different hand-written script identification verifying Sub Data Sets；
Involved dedicated hand-written script data set c includes multiple dedicated hand-written script Sub Data Sets, the institute of the same text
There is hand-written script to be stored in the same dedicated hand-written script Sub Data Set, the hand-written script of different literals is stored in different dedicated
Hand-written script Sub Data Set.
In the present embodiment, involved mobile terminal B can be mobile phone terminal, the end PC, the end PAD；Acquisition module 30 can pass through
The mode taken pictures or downloaded acquires hand-written script image.
The self-training stage+module of the present embodiment is communicated in mobile terminal B, mobile terminal B and cloud A by network.Specifically
When use:
1) firstly, the general hand-written script data set a based on cloud A, user are constructed by the building module 1 of cloud A
General hand-written script identification model 80；
2) secondly, user constructs hand-written script identification validation data set b and dedicated hand using the building module 2 20 of cloud A
Write volumetric data set c；
3) after completing construction work, user acquires the hand-written script image of oneself by the acquisition module 30 of mobile terminal B；
4) it is general hand-written script that user passes through the processing module 40 of mobile terminal B for the hand-written script image procossing of acquisition again
The authority data that identification model 80 can identify, then by treated, authority data is sent to cloud A, and is stored in cloud A
Hand-written script identify validation data set b；
5) at this point, the extraction input module 50 of cloud A extracts the data of hand-written script identification validation data set b one by one, and
It inputs general hand-written script identification model 80 to be verified, export verification result and is sent to cloud A and is shown, it is artificial to determine
Whether verification result is correct:
When 5a) artificial decision verification result is correct, extracts input module 50 and continue to extract hand-written script identification verify data
Collect next data of b；
When 5b) artificial decision verification result is incorrect, user corrects verification result by the correction module 60 of mobile terminal B,
Then the verification result by input data and after correcting is stored in dedicated hand-written script data set c, gos to step 6)；
6) user can be verified accurately by the 100 setting general-purpose hand-written script identification model 80 of threshold module of mobile terminal B
The threshold value of rate, then, the calculating judgment module 70 of cloud A are accurate by the verifying for calculating general hand-written script identification model 80
Rate, and judge whether the verifying accuracy rate of general hand-written script identification model 80 is not less than threshold value set by user, if so, will
General hand-written script identification model 80 is exported as the dedicated hand-written script identification model 90 of user to mobile terminal B, if it is not, then needing
User is wanted to continue to acquire the hand-written script image of oneself, until the verifying accuracy rate of general hand-written script identification model 80 is not less than
Threshold value set by user.
After executing the step 6), the dedicated hand-written script identification model 90 for being output to mobile terminal is suitable for acquired handwritten word
The user of body image, the dedicated hand-written script identification model 90 can the images such as all hand-written notes, document to the user carry out
Digitlization conversion and identification solve personalized hand-written script from the root and are difficult to reach good recognition effect in practical applications
The problem of.
Other people equally can use above-mentioned steps and train dedicated hand-written script identification model 90 suitable for oneself.
Claims (10)
1. a kind of self-training method of AI identification hand-written script, which is characterized in that this method is based on cloud and stores beyond the clouds
The realization step of general hand-written script data set, this method includes:
1) general hand-written script identification model is constructed using general hand-written script data set；
2) building hand-written script identifies validation data set and dedicated hand-written script data set beyond the clouds；
3) the hand-written script image of user is acquired, and handles the authority data that can be identified for general hand-written script identification model,
It is subsequently stored in hand-written script identification validation data set；
4) data of hand-written script identification validation data set are inputted general hand-written script identification model to verify, user is voluntarily
Check verification result, when verification result is not consistent with input data, user voluntarily corrects verification result, then by input data
Dedicated hand-written script data set is stored in the verification result after correction；
5) circulation executes step 3), 4), and given threshold acquires and handles the new hand-written script image of user and be iterated and tests
Card, until the verifying accuracy rate of general hand-written script identification model is not less than threshold value set by user；
6) at this point, the general hand-written script identification model of output is the dedicated hand-written script identification model of user.
2. a kind of self-training method of AI identification hand-written script according to claim 1, which is characterized in that processing handwritten word
The concrete operations of body image include:
Processing is split to the hand-written script image of acquisition, forms single font image；
Single font image is pre-processed, noise reduction, digitlization conversion, being converted to general hand-written script identification model can know
Other authority data is subsequently stored in hand-written script identification validation data set.
3. a kind of self-training method of AI identification hand-written script according to claim 1, which is characterized in that in step 3)
In, the hand-written script image of multi collect user is needed, and the whole hand-written script images acquired preferably cover general handwritten word
All texts that body identification model can identify.
4. a kind of self-training method of AI identification hand-written script according to claim 1, which is characterized in that the handwritten word
Body identification validation data set includes multiple hand-written script identification verifying Sub Data Sets, all hand-written scripts storage of the same text
Identify that verifying Sub Data Set, the hand-written script of different literals are stored in different hand-written script identification and test in the same hand-written script
Demonstrate,prove Sub Data Set.
5. a kind of self-training method of AI identification hand-written script according to claim 1, which is characterized in that the dedicated hand
The volumetric data set that writes includes multiple dedicated hand-written script Sub Data Sets, and all hand-written scripts of the same text are stored in same
Dedicated hand-written script Sub Data Set, the hand-written script of different literals are stored in different dedicated hand-written script Sub Data Sets.
6. a kind of self-training platform of AI identification hand-written script, which is characterized in that the self-training platform is based on carrying out by network
The mobile terminal and cloud of communication, cloud are stored with general hand-written script data set, and structure includes:
Building module one beyond the clouds is set, identifies mould for constructing general hand-written script using general hand-written script data set
Type；
Building module two beyond the clouds is set, for constructing hand-written script identification validation data set and dedicated hand-written script data
Collection；
The acquisition module of mobile terminal is set, for acquiring the hand-written script image of user；
The processing module of mobile terminal is set, and the hand-written script image procossing for that will acquire is general hand-written script identification model
The authority data that can be identified, is also used to that authority data is sent to cloud by treated, and is stored in the hand-written script in cloud
Identify validation data set；
Extraction input module beyond the clouds is set, for extracting the data of hand-written script identification validation data set, and is inputted general
Hand-written script identification model, general hand-written script identification model verify input data, artificial decision verification result；
The correction module of mobile terminal is set, and when artificial decision verification result is incorrect, user is tested by correcting module correction
Card is as a result, then the verification result by input data and after correcting is stored in dedicated hand-written script data set；
The threshold module of mobile terminal is set, the threshold value for user's setting general-purpose hand-written script identification model verifying accuracy rate；
Calculating judgment module beyond the clouds is set, for calculating the verifying accuracy rate of general hand-written script identification model, and is judged
Whether the verifying accuracy rate of general hand-written script identification model is not less than threshold value set by user, if so, by general handwritten word
Body identification model is exported as the dedicated hand-written script identification model of user to mobile terminal, if it is not, then continuing the hand of acquisition user
Write font image.
7. a kind of self-training platform of AI identification hand-written script according to claim 6, which is characterized in that the processing mould
Block includes:
Divide submodule, for being split processing to the hand-written script image of acquisition to form single font image；
Submodule is pre-processed, for pre-processing to single font image；
Noise reduction submodule, for carrying out noise reduction process to single font image；
Transform subblock is digitized, for the single font image after noise reduction process to be converted to general hand-written script identification model
The authority data that can be identified；
Sending submodule, the authority data for converting single font image are sent to cloud, and are stored in the hand-written of cloud
Character Font Recognition validation data set.
8. a kind of self-training platform of AI identification hand-written script according to claim 6, which is characterized in that the same use
For family, whole hand-written script images of the acquisition module acquisition cover the institute that general hand-written script identification model can identify
There is text.
9. a kind of self-training platform of AI identification hand-written script according to claim 6, which is characterized in that the handwritten word
Body identification validation data set includes multiple hand-written script identification verifying Sub Data Sets, all hand-written scripts storage of the same text
Identify that verifying Sub Data Set, the hand-written script of different literals are stored in different hand-written script identification and test in the same hand-written script
Demonstrate,prove Sub Data Set；
The dedicated hand-written script data set includes multiple dedicated hand-written script Sub Data Sets, all handwritten words of the same text
Body is stored in the same dedicated hand-written script Sub Data Set, and the hand-written script of different literals is stored in different dedicated hand-written scripts
Sub Data Set.
10. a kind of self-training platform of AI identification hand-written script according to claim 6, which is characterized in that the movement
End can be mobile phone terminal, the end PC, the end PAD；The acquisition module acquires hand-written script image by way of taking pictures or downloading.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910317078.9A CN110033052A (en) | 2019-04-19 | 2019-04-19 | A kind of the self-training method and self-training platform of AI identification hand-written script |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910317078.9A CN110033052A (en) | 2019-04-19 | 2019-04-19 | A kind of the self-training method and self-training platform of AI identification hand-written script |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110033052A true CN110033052A (en) | 2019-07-19 |
Family
ID=67239307
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910317078.9A Pending CN110033052A (en) | 2019-04-19 | 2019-04-19 | A kind of the self-training method and self-training platform of AI identification hand-written script |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110033052A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705533A (en) * | 2019-09-09 | 2020-01-17 | 武汉联析医疗技术有限公司 | AI recognition and grabbing system for inspection report |
| CN113408373A (en) * | 2021-06-02 | 2021-09-17 | 中金金融认证中心有限公司 | Handwriting recognition method, system, client and server |
| US11755687B2 (en) | 2021-10-05 | 2023-09-12 | Prince Mohammad Bin Fahd University | Text independent writer verification method and system |
| CN113408373B (en) * | 2021-06-02 | 2024-06-07 | 中金金融认证中心有限公司 | Handwriting recognition method, handwriting recognition system, client and server |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105184329A (en) * | 2015-08-27 | 2015-12-23 | 鲁东大学 | Cloud-platform-based off-line handwriting recognition method |
| US20180293435A1 (en) * | 2017-04-10 | 2018-10-11 | Pearson Education, Inc. | Electronic handwriting processor with convolutional neural networks |
| CN108710866A (en) * | 2018-06-04 | 2018-10-26 | 平安科技（深圳）有限公司 | Chinese mold training method, Chinese characters recognition method, device, equipment and medium |
| CN108921031A (en) * | 2018-06-04 | 2018-11-30 | 平安科技（深圳）有限公司 | Chinese mold training method, hand-written character recognizing method, device, equipment and medium |
| CN108985297A (en) * | 2018-06-04 | 2018-12-11 | 平安科技（深圳）有限公司 | Handwriting model training, hand-written image recognition methods, device, equipment and medium |
-
2019
- 2019-04-19 CN CN201910317078.9A patent/CN110033052A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105184329A (en) * | 2015-08-27 | 2015-12-23 | 鲁东大学 | Cloud-platform-based off-line handwriting recognition method |
| US20180293435A1 (en) * | 2017-04-10 | 2018-10-11 | Pearson Education, Inc. | Electronic handwriting processor with convolutional neural networks |
| CN108710866A (en) * | 2018-06-04 | 2018-10-26 | 平安科技（深圳）有限公司 | Chinese mold training method, Chinese characters recognition method, device, equipment and medium |
| CN108921031A (en) * | 2018-06-04 | 2018-11-30 | 平安科技（深圳）有限公司 | Chinese mold training method, hand-written character recognizing method, device, equipment and medium |
| CN108985297A (en) * | 2018-06-04 | 2018-12-11 | 平安科技（深圳）有限公司 | Handwriting model training, hand-written image recognition methods, device, equipment and medium |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110705533A (en) * | 2019-09-09 | 2020-01-17 | 武汉联析医疗技术有限公司 | AI recognition and grabbing system for inspection report |
| CN113408373A (en) * | 2021-06-02 | 2021-09-17 | 中金金融认证中心有限公司 | Handwriting recognition method, system, client and server |
| CN113408373B (en) * | 2021-06-02 | 2024-06-07 | 中金金融认证中心有限公司 | Handwriting recognition method, handwriting recognition system, client and server |
| US11755687B2 (en) | 2021-10-05 | 2023-09-12 | Prince Mohammad Bin Fahd University | Text independent writer verification method and system |
| US11914673B2 (en) | 2021-10-05 | 2024-02-27 | Prince Mohammad Bin Fahd University | System to identify authorship of handwritten text based on individual alphabets |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6484333B2 (en) | Intelligent scoring method and system for descriptive problems | |
| CN109325464A (en) | A kind of finger point reading character recognition method and interpretation method based on artificial intelligence | |
| CN103971690A (en) | Voiceprint recognition method and device | |
| CN106372581A (en) | Method for constructing and training human face identification feature extraction network | |
| CN103927532B (en) | Person's handwriting method for registering based on stroke feature | |
| CN103258157B (en) | A kind of online handwriting authentication method based on finger information and system | |
| CN106980856A (en) | Formula identification method and system and symbolic reasoning computational methods and system | |
| CN109993164A (en) | A kind of natural scene character recognition method based on RCRNN neural network | |
| CN103680493A (en) | Voice data recognition method and device for distinguishing regional accents | |
| CN109190579B (en) | Generation type countermeasure network SIGAN signature handwriting identification method based on dual learning | |
| CN104732226A (en) | Character recognition method and device | |
| CN103984948A (en) | Soft double-layer age estimation method based on facial image fusion features | |
| CN112966685B (en) | Attack network training method and device for scene text recognition and related equipment | |
| CN105117740A (en) | Font identification method and device | |
| CN110299132B (en) | Voice digital recognition method and device | |
| CN103839042A (en) | Human face recognition method and human face recognition system | |
| CN110033052A (en) | A kind of the self-training method and self-training platform of AI identification hand-written script | |
| CN112257437A (en) | Voice recognition error correction method and device, electronic equipment and storage medium | |
| CN113886792A (en) | Application method and system of print control instrument combining voiceprint recognition and face recognition | |
| CN106384587B (en) | A kind of audio recognition method and system | |
| CN114170468B (en) | Text recognition method, storage medium and computer terminal | |
| CN109478229B (en) | Training device for classification network for character recognition, character recognition device and method | |
| CN114429636A (en) | Image scanning identification method and device and electronic equipment | |
| CN104901807A (en) | Vocal print password method available for low-end chip | |
| CN112417918B (en) | Two-dimensional code identification method and device, storage medium and electronic equipment |
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: 20190719 |
|
| RJ01 | Rejection of invention patent application after publication |