CN109101495A - A kind of fashion world document creation method based on image recognition and knowledge mapping - Google Patents
A kind of fashion world document creation method based on image recognition and knowledge mapping Download PDFInfo
- Publication number
- CN109101495A CN109101495A CN201810982554.4A CN201810982554A CN109101495A CN 109101495 A CN109101495 A CN 109101495A CN 201810982554 A CN201810982554 A CN 201810982554A CN 109101495 A CN109101495 A CN 109101495A
- Authority
- CN
- China
- Prior art keywords
- fashion
- knowledge mapping
- information
- world
- document creation
- 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
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/258—Heading extraction; Automatic titling; Numbering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/189—Automatic justification
Abstract
The present invention relates to a kind of fashion world document creation method based on image recognition and knowledge mapping, step S1, using the method for deep learning, basic picture video data are carried out with the extraction of basic information, and the basic information for all fashion is decomposed and arranged, the algorithm of training deep learning, extracts image basis information；Step S2, the knowledge mapping for preparing fashion world in advance is defined fashion elements on the market, basic information and fashion information is associated；Step S3, using above-mentioned essential information and the Higher Order Abstract label of reasoning as search key, Similar Text is searched in existing fashion world professional article library, according to sequencing of similarity, generates abstract text.Its advantage is shown: being able to solve the problem of needing to write the very big field profession manuscript of a large amount of and picture character correlation in fashion world, is provided fashion editor's textual materials, to excite the inspiration of fashion editor, raising fashion is contributed to newspapers, magazines,etc. efficiency.
Description
Technical field
The present invention relates to internet big data field, fashion world, field of computer technology, specifically, being a kind of base
In the fashion world document creation method of image recognition and knowledge mapping.
Background technique
Text automatically generate be natural language processing field an important research direction, realize text automatically generate be also
The important symbol that artificial intelligence moves to maturity.
The mode that current text automatically generates is concentrated mainly on two kinds of aspects:
The first is often to use internet public data based on full-text search field, several out by search engine inquiry
Piece article close and similar with theme, carries out abstract selection by certain algorithm, ultimately generates the text being understood that
Chapter.
The similar article that this method needs largely to increase income, and text generation can not be carried out to information such as pictures.
Second is the text Auto-Generation Tool based on deep learning, a kind of using similar s2s or neuraltalk
Algorithm expands keyword or picture, to generate an article being understood that.
The difficult point of this technology is that the training of deep learning is complicated, and popularization difficulty is big, it is simple to generate text, and to not
Know that image accuracy rate remains a challenge.
In addition, needing to write a large amount of picture and text in fashion world, the variation of fashion world is often prior to all
Field, so-called top of the waves tide-person, so leading to the text information of fashion world, there are many difficult points.
1, a large amount of relations all are being generated daily；
2, there are very big variations for the main style of relation and internal information.
3, data type is sufficiently complex, is all the target data handled to video from text to image.
For above several difficult points, current method, which fine can not all obtain, is suitable for these situations.Traditional information retrieval
Mode can rapidly adapt to change, but processing capacity is but lacked to multimedia messages；The method of deep learning can handle
Media information, but a large amount of labeled data is needed, and mark and need a large amount of artificial informations, it can not often accomplish well
Adapt to trend.
Chinese patent literature CN201410032479.7, the applying date 20140123, patent name are as follows: text is carried out automatic
The method and apparatus of phonetic notation processing and display disclose method and dress that a kind of pair of text carries out automatic phonetic notation processing and display
Set, wherein to text carry out automatic phonetic notation processing method include: based on hidden code model by the text conversion of original language be target
The phonetic notation text of language；The identification information of acquisition demand user；Receive active user is needed by the current text that client is sent
Seek information, according to identification information determine active user be demand user, according to current text demand information to client send pair
The phonetic notation text answered, for showing to active user.
The method that above patent document carries out automatic phonetic notation processing to text can give birth to automatically according to foreign language source language text
Chinese character phonetic alphabet text is supplied to demand user at Chinese character phonetic alphabet text that is intelligible or easily pronouncing, and actively, is convenient for demand
User easily learns the pronunciation of foreign language original language song, and user experience is high, and at low cost.But it is solved about a kind of
The problem of needing to write the very big field profession manuscript of a large amount of and picture character correlation in fashion world, provides fashion editor
Textual materials improve the contribute to newspapers, magazines,etc. technical solution of efficiency of fashion and are not disclosed accordingly then.
In conclusion it is very big to need a kind of solution to need to write a large amount of and picture character correlation in fashion world
It the problem of field profession manuscript, provides fashion editor's textual materials, improves fashion and contribute to newspapers, magazines,etc. the document creation method of efficiency, and about
This document creation method yet there are no report.
Summary of the invention
The purpose of the present invention is aiming at the shortcomings in the prior art, provide a kind of solution need in fashion world to write it is big
The problem of measuring field profession manuscript very big with picture character correlation, provides fashion editor's textual materials, improves fashion and writes
The document creation method of original text efficiency.
To achieve the above object, the technical solution adopted by the present invention is that:
A kind of fashion world document creation method based on image recognition and knowledge mapping, the method includes following steps
It is rapid:
Step S1, using the method for deep learning, basic picture video data are carried out with the extraction of basic information, and right
It is decomposed and is arranged for the basic information of all fashion, the algorithm of training deep learning carries out image basis information
It extracts；
Step S2, the knowledge mapping for preparing fashion world in advance is defined fashion elements on the market, and basis is believed
Breath and fashion information are associated；
Step S3, it is closed the Higher Order Abstract label of the essential information and reasoning that are obtained in step S1 and step S2 as search
Key word searches for Similar Text in existing fashion world professional article library, according to sequencing of similarity, generates abstract text.
As a kind of perferred technical scheme, the method provides label in the form of the knowledge mapping of fashion world
In logical relation storage and expression.
As a kind of perferred technical scheme, the method is formed by mode using algorithm and data structure to provide mark
The storage and expression of logical relation in label.
As a kind of perferred technical scheme, the method expands fashion in such a way that deep learning cooperates reasoning
The system configurations for the label information for including in the image of field.
As a kind of perferred technical scheme, the method is matched using fashion world picture tag algorithm and fashion tag library
It closes, reaches the multifarious design architecture of label.
As a kind of perferred technical scheme, it is to become that the extraction object in the step S1, which is the basic information of fashion world,
Change slow, and has multimedia feature.
As a kind of perferred technical scheme, the Higher Order Abstract element based on basic element is generated in institute step S2, thus
Generate Higher Order Abstract label.
The invention has the advantages that:
1, a kind of fashion world document creation method based on image recognition and knowledge mapping of the invention, is able to solve
The problem of needing to write the very big field profession manuscript of a large amount of and picture character correlation in fashion world, provides fashion editor
Textual materials improve fashion and contribute to newspapers, magazines,etc. efficiency to excite the inspiration of fashion editor.
2, the learning cost for reducing deep learning, improves the availability of model.
3, the update of fast-changing relation is provided.
4, the knowledge base of manual maintenance is capable of providing more domain knowledges, avoids information out-of-date and inaccurate situation.
Detailed description of the invention
Attached drawing 1 is a kind of process of fashion world document creation method based on image recognition and knowledge mapping of the invention
Schematic diagram.
Specific embodiment
It elaborates with reference to the accompanying drawing to specific embodiment provided by the invention.
Fig. 1 is please referred to, Fig. 1 is a kind of fashion world text generation side based on image recognition and knowledge mapping of the invention
The flow diagram of method.A kind of fashion world document creation method based on image recognition and knowledge mapping, the method includes
Following steps:
Step S1, using the method for deep learning, basic picture video data are carried out with the extraction of basic information, and right
It is decomposed and is arranged for the basic information of all fashion, the algorithm of training deep learning carries out image basis information
It extracts, such as the information such as the main body of image, material, texture, quality, such as the main body in figure is one " skirt ", skirt
Material be " silk ", the color of skirt is " blue " and " red ", and the quality of skirt is the equal basic informations of " smooth ".
Step S2, the knowledge mapping for preparing fashion world in advance is defined fashion elements on the market, and basis is believed
Breath and fashion information are associated, such as: " one-piece dress " and " lace " can produce " pseudo-classic "；" blue " and " red " can produce
" hitting color "；" skirt " and " plate shoes " can release " street corner ".Thus Higher Order Abstract label is generated.
Step S3, it is closed the Higher Order Abstract label of the essential information and reasoning that are obtained in step S1 and step S2 as search
Key word searches for Similar Text in existing fashion world professional article library, according to sequencing of similarity, generates abstract text.
The embodiment it should be understood that
Logical relation in label is provided in the form of the knowledge mapping of fashion world in the example, but is not limited only to
Which realizes the storage and expression of the logical relation between label, other algorithms and data structure are formed by technical solution and also answer
Cover under the inventive concept.
Expand the label letter for including in fashion world image in the example in such a way that deep learning cooperates reasoning
The system configurations of breath, but it is not limited to the specific combination in examples detailed above.Other fashion world picture tag algorithms and fashion
Tag library cooperation, the design architecture for the purpose of reaching label diversity should all be covered under design of the invention.
A kind of fashion world document creation method based on image recognition and knowledge mapping of the invention, be able to solve when
The problem of still needing to write the very big field profession manuscript of a large amount of and picture character correlation in field, it is literary to provide fashion editor
Grapheme material improves fashion and contributes to newspapers, magazines,etc. efficiency to excite the inspiration of fashion editor；The learning cost of deep learning is reduced, model is improved
Availability；The update of fast-changing relation is provided；The knowledge base of manual maintenance is capable of providing more domain knowledges,
Avoid information out-of-date and inaccurate situation.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
Member, under the premise of not departing from the method for the present invention, can also make several improvement and supplement, these are improved and supplement also should be regarded as
Protection scope of the present invention.
Claims (7)
1. a kind of fashion world document creation method based on image recognition and knowledge mapping, which is characterized in that the method packet
Include following steps:
Step S1, using the method for deep learning, basic picture video data are carried out with the extraction of basic information, and to being directed to
The basic information of all fashion is decomposed and is arranged, and the algorithm of training deep learning extracts image basis information；
Step S2, the knowledge mapping for preparing fashion world in advance is defined fashion elements on the market, by basic information and
Fashion information is associated；
Step S3, crucial using the Higher Order Abstract label of the essential information and reasoning that are obtained in step S1 and step S2 as search
Word searches for Similar Text in existing fashion world professional article library, according to sequencing of similarity, generates abstract text.
2. the fashion world document creation method according to claim 1 based on image recognition and knowledge mapping, feature
It is, the method provides the storage and expression of the logical relation in label in the form of the knowledge mapping of fashion world.
3. the fashion world document creation method according to claim 1 based on image recognition and knowledge mapping, feature
It is, the method is formed by mode using algorithm and data structure to provide the storage of the logical relation in label and table
Show.
4. the fashion world document creation method according to claim 1 based on image recognition and knowledge mapping, feature
It is, the method expands the label information for including in fashion world image in such a way that deep learning cooperates reasoning
System configurations.
5. the fashion world document creation method according to claim 1 based on image recognition and knowledge mapping, feature
It is, the method is cooperated using fashion world picture tag algorithm and fashion tag library, reaches the multifarious designed holder of label
Structure.
6. the fashion world document creation method according to claim 1 based on image recognition and knowledge mapping, feature
It is, the extraction object in the step S1 is that the basic information of fashion world is that variation is slow, and has multimedia
Feature.
7. the fashion world document creation method according to claim 1 based on image recognition and knowledge mapping, feature
It is, the Higher Order Abstract element based on basic element is generated in step S2, thus generates Higher Order Abstract label.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810982554.4A CN109101495A (en) | 2018-08-27 | 2018-08-27 | A kind of fashion world document creation method based on image recognition and knowledge mapping |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810982554.4A CN109101495A (en) | 2018-08-27 | 2018-08-27 | A kind of fashion world document creation method based on image recognition and knowledge mapping |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109101495A true CN109101495A (en) | 2018-12-28 |
Family
ID=64851372
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810982554.4A Pending CN109101495A (en) | 2018-08-27 | 2018-08-27 | A kind of fashion world document creation method based on image recognition and knowledge mapping |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109101495A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109961091A (en) * | 2019-03-01 | 2019-07-02 | 杭州叙简科技股份有限公司 | A kind of accident word tag of self study and summarization generation system and method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0751469A1 (en) * | 1995-06-28 | 1997-01-02 | Xerox Corporation | Automatic method of extracting summarization using feature probabilities |
| CN106095762A (en) * | 2016-02-05 | 2016-11-09 | 中科鼎富（北京）科技发展有限公司 | A kind of news based on ontology model storehouse recommends method and device |
| CN106204165A (en) * | 2016-08-11 | 2016-12-07 | 广州出益信息科技有限公司 | A kind of advertisement placement method and device |
| CN107908671A (en) * | 2017-10-25 | 2018-04-13 | 南京擎盾信息科技有限公司 | Knowledge mapping construction method and system based on law data |
| CN108038096A (en) * | 2017-11-10 | 2018-05-15 | 平安科技（深圳）有限公司 | Knowledge database documents method for quickly retrieving, application server computer readable storage medium storing program for executing |
-
2018
- 2018-08-27 CN CN201810982554.4A patent/CN109101495A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0751469A1 (en) * | 1995-06-28 | 1997-01-02 | Xerox Corporation | Automatic method of extracting summarization using feature probabilities |
| CN106095762A (en) * | 2016-02-05 | 2016-11-09 | 中科鼎富（北京）科技发展有限公司 | A kind of news based on ontology model storehouse recommends method and device |
| CN106204165A (en) * | 2016-08-11 | 2016-12-07 | 广州出益信息科技有限公司 | A kind of advertisement placement method and device |
| CN107908671A (en) * | 2017-10-25 | 2018-04-13 | 南京擎盾信息科技有限公司 | Knowledge mapping construction method and system based on law data |
| CN108038096A (en) * | 2017-11-10 | 2018-05-15 | 平安科技（深圳）有限公司 | Knowledge database documents method for quickly retrieving, application server computer readable storage medium storing program for executing |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109961091A (en) * | 2019-03-01 | 2019-07-02 | 杭州叙简科技股份有限公司 | A kind of accident word tag of self study and summarization generation system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Rich visual knowledge-based augmentation network for visual question answering | |
| CN112087656B (en) | Online note generation method and device and electronic equipment | |
| Zheng et al. | Recent advances of deep learning for sign language recognition | |
| CN102855317B (en) | A kind of multi-mode indexing means and system based on demonstration video | |
| TWI312945B (en) | Method and apparatus for multimedia data management | |
| CN107679070B (en) | Intelligent reading recommendation method and device and electronic equipment | |
| CN111524045A (en) | Dictation method and device | |
| CN110297933A (en) | A kind of theme label recommended method and tool based on deep learning | |
| CN109343768A (en) | Display methods and system, computer system and computer readable storage medium | |
| CN111985184A (en) | Auxiliary writing font copying method, system and device based on AI vision | |
| CN113535915A (en) | Method for expanding a data set | |
| CN109492112A (en) | A kind of method of the computer aided writing scientific popular article of knowledge based map | |
| TW200425063A (en) | Recognition method to integrate speech input and handwritten input, and system thereof | |
| CN109101495A (en) | A kind of fashion world document creation method based on image recognition and knowledge mapping | |
| CN110458751B (en) | Face replacement method, device and medium based on Guangdong play pictures | |
| Smith et al. | A lexical frequency analysis of Irish Sign Language | |
| CN111552830A (en) | User-selected music score exercise method and system | |
| Nagai | On the improvement of recognizing single-line strings of Japanese historical cursive | |
| Goel et al. | Real-time sign language to text and speech translation and hand gesture recognition using the LSTM model | |
| JP2018066800A (en) | Japanese speech recognition model learning device and program | |
| CN110059179A (en) | A kind of song text name entity recognition method based on deep learning | |
| Hu et al. | Multimedia analysis and access of ancient maya epigraphy: Tools to support scholars on Maya hieroglyphics | |
| CN112989811B (en) | History book reading auxiliary system based on BiLSTM-CRF and control method thereof | |
| Shi et al. | Searching for fingerspelled content in American Sign Language | |
| CN114090777A (en) | Text data processing method and device |
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: 20181228 |
|
| RJ01 | Rejection of invention patent application after publication |