CN110070512A - The method and device of picture modification - Google Patents
The method and device of picture modification Download PDFInfo
- Publication number
- CN110070512A CN110070512A CN201910362660.7A CN201910362660A CN110070512A CN 110070512 A CN110070512 A CN 110070512A CN 201910362660 A CN201910362660 A CN 201910362660A CN 110070512 A CN110070512 A CN 110070512A
- Authority
- CN
- China
- Prior art keywords
- modification
- label
- picture
- scene
- modification data
- 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.)
- Granted
Links
- 230000004048 modification Effects 0.000 title claims abstract description 316
- 238000012986 modification Methods 0.000 title claims abstract description 316
- 238000000034 method Methods 0.000 title claims abstract description 68
- 238000012545 processing Methods 0.000 claims abstract description 12
- 230000015654 memory Effects 0.000 claims description 21
- 230000008569 process Effects 0.000 claims description 21
- 238000012549 training Methods 0.000 claims description 15
- 230000011218 segmentation Effects 0.000 claims description 8
- 238000004891 communication Methods 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 5
- 230000000694 effects Effects 0.000 abstract description 8
- 239000000463 material Substances 0.000 abstract description 5
- 241000251468 Actinopterygii Species 0.000 description 11
- 238000010586 diagram Methods 0.000 description 6
- 241001481833 Coryphaena hippurus Species 0.000 description 5
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 5
- 238000013461 design Methods 0.000 description 4
- 241000283153 Cetacea Species 0.000 description 3
- 241000251730 Chondrichthyes Species 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000029058 respiratory gaseous exchange Effects 0.000 description 2
- 239000013535 sea water Substances 0.000 description 2
- 230000009182 swimming Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 244000283207 Indigofera tinctoria Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 241000566150 Pandion haliaetus Species 0.000 description 1
- 238000005034 decoration Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/77—Retouching; Inpainting; Scratch removal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
Abstract
This application provides a kind of method and devices of picture modification, it is related to the technical field of picture processing, wherein, a kind of method of picture modification provided by the present application, the scene of picture and the prediction probability of scene are determined by trained Bayes classifier, and at least one modification data are obtained from the modification database of foundation according to the label of picture, and the score by calculating every obtained modification data, selection modification data modify picture to be matched from least one modification data of acquisition, so that the material that server can be used for modifying is more, the effect of picture after improving modification.
Description
Technical field
This application involves the technical fields of picture processing, more particularly, to a kind of method and device of picture modification.
Background technique
In application scenes, server needs to modify picture, the picture after generating a modification, and should
Picture after modification is used for specific occasion, for example, when needing to modify picture, server can in the scene of chat
To add corresponding sentence to the picture, and the picture after addition sentence is sent to user.
Currently, usually being modified the color of picture when server modifies picture, or using ornament to figure
The content of piece is modified, or the content based on picture directly generates corresponding description language, for example, description language be sky very
Indigo plant, seawater are very limpid etc..But in specific occasion, when the above method being selected to modify picture, the material of server modification
It is limited, so that the effect of the picture after modification is bad.
Summary of the invention
In view of this, the application's is designed to provide a kind of picture method of modifying and device, to improve picture after modification
Effect.
In a first aspect, the embodiment of the present application provides a kind of method of picture modification, which comprises
Picture to be matched is obtained, and identifies at least one label of the picture to be matched；
At least one label of the picture is input in trained Bayes classifier, the picture is obtained and is divided
It Biao Ji be not the prediction probability of each scene in scene；
According at least one label of the picture, obtained from the modification database of foundation at least one modification data,
And in labeled at least one scene of every modification data each scene prediction probability, and obtain at least one label
The rare degree of each label；
For every modification data at least one modification data of acquisition, according to the corresponding label of the modification data
Rare degree, in labeled at least one scene of the modification data each scene prediction probability and the picture quilt
The prediction probability of each scene at least one scene of label calculates the score of the modification data；
The score that data are modified according to every selects the modification of preset quantity from least one modification data of acquisition
Data；
The modification data of the preset quantity of selection are combined with picture to be matched respectively, the figure after generating multiple modifications
Piece, and the picture after modification is sent to user terminal.
In some embodiments of the present application, the training process of the Bayes classifier, comprising:
Multiple scenes are constructed, for each scene, are executed:
The corresponding a plurality of samples pictures of the scene are obtained, at least one corresponding mark is labeled in every samples pictures
Label；
The weight of the label is determined according to the quantity of the label in a plurality of samples pictures for any one label；
At least one label with weight is input in Bayes classifier, the Bayes classifier is instructed
Practice.
In some embodiments of the present application, the process of the modification database is established, comprising:
A plurality of modification data are acquired, for every modification data, are executed:
The modification data are subjected to word segmentation processing, obtain at least one corresponding word of the modification data；
According at least one described word, the corresponding label of modification data is determined；
The corresponding label of modification data is input in trained Bayes classifier, the modification data is obtained and is marked
At least one scene of note and the prediction probability of each scene；
At least one labeled field of every modification data in a plurality of modification data and a plurality of modification data
The prediction probability of scape, each scene constitutes the modification database.
In some embodiments of the present application, described at least one word according to determines the corresponding mark of modification data
Label, comprising:
For each word, execute:
The word is input in preset tag library and is identified, wherein includes multiple labels in the tag library
Collection, wherein include the corresponding synonym of the label in each tally set；
According to tag library to the recognition result of the word, the corresponding label of the word is determined；
By the corresponding label of the word, it is determined as the corresponding label of modification data.
In some embodiments of the present application, the process of the modification database is established, further includes:
Count the number of each label at least one corresponding label of a plurality of modification data；
According to the corresponding number of the label, the rare degree of the label is determined；
The rare degree of each label at least one corresponding label of a plurality of modification data is input to the modification data
In library.
Second aspect, the embodiment of the present application also provide a kind of device of picture modification, and described device includes:
Picture recognition module for obtaining picture to be matched, and identifies at least one mark of the picture to be matched
Label；
Scene prediction module, at least one label of the picture to be input to trained Bayes classifier
In, obtain the prediction probability that the picture is respectively labeled as each scene at least one scene；
Data acquisition module is modified, at least one label according to the picture, from the modification database of foundation
The prediction for obtaining each scene at least one scene that at least one modification data and every modification data are labeled is general
Rate, and obtain the rare degree of each label at least one label；
Points calculating module, for modifying every modification data in data at least one obtained, according to described
Rare degree, the prediction for modifying each scene at least one labeled scene of data for modifying the corresponding label of data are general
The prediction probability for each scene at least one scene that rate and the picture are labeled calculates the modification data
Score；
Data selecting module is modified, for modifying the score of data according to every, modifies data from at least one of acquisition
In, select the modification data of preset quantity；
Picture modifies module, for the modification data of the preset quantity of selection to be combined with picture to be matched respectively,
Picture after generating multiple modifications, and the picture after modification is sent to user terminal.
In some embodiments of the present application, described device further include: Bayes classifier training module；
The Bayes classifier training module, is specifically used for:
Multiple scenes are constructed, for each scene, are executed:
The corresponding a plurality of samples pictures of the scene are obtained, at least one corresponding mark is labeled in every samples pictures
Label；
The weight of the label is determined according to the quantity of the label in a plurality of samples pictures for any one label；
At least one label with weight is input in Bayes classifier, the Bayes classifier is instructed
Practice.
In some embodiments of the present application, described device further include: modification Database module；
The modification Database module, is specifically used for:
A plurality of modification data are acquired, for every modification data, are executed:
The modification data are subjected to word segmentation processing, obtain at least one corresponding word of the modification data；
According at least one described word, the corresponding label of modification data is determined；
The corresponding label of modification data is input in trained Bayes classifier, the modification data is obtained and is marked
At least one scene of note and the prediction probability of each scene；
At least one labeled field of every modification data in a plurality of modification data and a plurality of modification data
The prediction probability of scape, each scene constitutes the modification database.
The third aspect, the embodiment of the present application also provide a kind of electronic equipment, comprising: processor, memory and bus, it is described
Memory is stored with the executable machine readable instructions of the processor, when electronic equipment operation, the processor with it is described
By bus communication between memory, the machine readable instructions executed when being executed by the processor it is above-mentioned in a first aspect, or
The step of method of the modification of picture described in any possible embodiment of first aspect.
Fourth aspect, the embodiment of the present application also provide a kind of computer readable storage medium, the computer-readable storage medium
Computer program is stored in matter, which executes above-mentioned in a first aspect, or first aspect when being run by processor
The step of method of the modification of picture described in any possible embodiment.
The method and device of picture modification provided by the embodiments of the present application, passes through at least one of identification picture to be matched
At least one label of picture is input in trained Bayes classifier by label, obtain picture be respectively labeled as to
The prediction probability of each scene in a few scene；And according at least one label of picture, from the modification database of foundation
Obtain at least one modification data, and at least one be labeled according to the rare degree of the corresponding label of modification data, modification data
The prediction of each scene in a scene at least one labeled scene of the prediction probability and picture of each scene is general
Rate calculates the score of modification data, and the score of data is modified according to every, from least one modification data of acquisition, selection
The modification data of preset quantity；The modification data of the preset quantity of selection are combined with picture to be matched respectively, are generated more
Picture after a modification, and the picture after modification is sent to user terminal.
The method modified by picture provided by the present application, can be by trained Bayes classifier to the field of picture
Scape and the prediction probability of scene are determined, and are obtained at least one from the modification database of foundation according to the label of picture and repaired
Data, and the score by calculating every obtained modification data are adornd, selects modification from least one modification data of acquisition
Data modify picture to be matched, so that the material that server can be used for modifying is more, the picture after improving modification
Effect.
To enable the above objects, features, and advantages of the application to be clearer and more comprehensible, preferred embodiment is cited below particularly, and cooperate
Appended attached drawing, is described in detail below.
Detailed description of the invention
Technical solution in ord to more clearly illustrate embodiments of the present application, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only some embodiments of the application, therefore is not construed as pair
The restriction of range for those of ordinary skill in the art without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 shows a kind of flow chart of the method for picture modification provided by the embodiment of the present application；
Fig. 2 shows a kind of flow charts of the training process of Bayes classifier provided by the embodiment of the present application；
Fig. 3 shows a kind of flow chart for the process for establishing modification database provided by the embodiment of the present application；
Fig. 4 shows a kind of structural block diagram of the device of picture modification provided by the embodiment of the present application；
Fig. 5 shows the structural schematic diagram of a kind of electronic equipment 600 provided by the embodiment of the present application.
Icon: 41- picture recognition module；42- scene prediction module；43- modifies data acquisition module；44- score calculates
Module；45- modifies data selecting module；46- picture modifies module；500- electronic equipment；501- processor；502- memory；
503- bus；5021- memory；5022- external memory.
Specific embodiment
To keep the purposes, technical schemes and advantages of the embodiment of the present application clearer, below in conjunction with the embodiment of the present application
Middle attached drawing, the technical scheme in the embodiment of the application is clearly and completely described, it is clear that described embodiment is only
It is some embodiments of the present application, instead of all the embodiments.The application being usually described and illustrated herein in the accompanying drawings is real
The component for applying example can be arranged and be designed with a variety of different configurations.Therefore, below to the application's provided in the accompanying drawings
The detailed description of embodiment is not intended to limit claimed scope of the present application, but is merely representative of the selected reality of the application
Apply example.Based on embodiments herein, those skilled in the art institute obtained without making creative work
There are other embodiments, shall fall in the protection scope of this application.
When modifying in view of server picture, usually the color of picture is modified, or uses ornament pair
The content of picture is modified, or the content based on picture directly generates corresponding description language, for example, description language is sky
Very blue, seawater is very limpid etc..But in specific occasion, when the above method being selected to modify picture, the element of server modification
Material is limited, so that the effect of the picture after modification is bad.Based on this, the embodiment of the present application provides a kind of modification of picture
Method and device is described below by embodiment.
For the method convenient for understanding the present embodiment, first to a kind of modification of picture disclosed in the embodiment of the present application
It describes in detail.
Embodiment one
The embodiment of the present application provides a kind of method of picture modification, as shown in Fig. 1 a kind of method of picture modification
Flow chart, this method comprises:
S101 obtains picture to be matched, and identifies at least one label of picture to be matched.
As an alternative embodiment, picture to be matched can be input in picture recognition model, treat matched picture
Content identified, obtain at least one label of picture to be matched.For example, a picture to be matched is chosen,
It treats matched picture to be identified, the corresponding label of the picture: water, wing, marine organisms, fish, shark, dolphin, life can be obtained
Object etc..
At least one label of picture is input in trained Bayes classifier by S102, is obtained picture and is distinguished
Labeled as the prediction probability of each scene at least one scene.
Specifically, as shown in Fig. 2, a kind of flow chart of the training process of Bayes classifier is shown in the figure, this Shen
In a kind of method for picture modification that please be provide, the training process of Bayes classifier, comprising:
S201 constructs multiple scenes, wherein the type of scene can be configured according to practical situation.The application is real
It is illustrated for applying in example in the scene that the method to modify the picture is applied to public notice, constructs multiple scenes, specifically,
As shown in table 1, multiple scenes of building include:
Table 1, the schematic diagram of multiple scenes
For each scene in multiple scenes of above-mentioned building, execute:
S202 obtains the corresponding a plurality of samples pictures of the scene, be labeled in every samples pictures it is corresponding at least one
Label.
For the scene, a plurality of samples pictures under the scene are obtained, according to picture recognition model to every samples pictures
It is identified, obtains at least one label of every samples pictures, specifically, the quantity of samples pictures can be according to actual feelings
Condition is configured.For example, it includes blue sky, white clouds, building, river, the then picture in a picture that picture recognition model, which recognizes,
Label can be blue sky, white clouds, building, river.
S203 determines the power of the label according to the quantity of the label in a plurality of samples pictures for any one label
Weight.
In the embodiment of the present application, under the scene, it can be determined every according to the quantity of label each in a plurality of samples pictures
The corresponding weight of a label.For example, illustrating by campus of scene, under the scene of campus, obtain more under the scene of campus
Samples pictures, identification when carrying out to a plurality of samples pictures under the scene of campus, obtaining corresponding label includes: building, teaching
Building, playground, blue sky, meadow, dining room etc., wherein counting the quantity of each label under the scene of campus, the quantity of building is 8, then
The building, weight can be 8, and the quantity of teaching building is 15, then the weight of the teaching building is 15, and so on, it can be obtained
The weight of each label.Specifically, above-mentioned example is exemplary description, the application is not especially limited this.
At least one label with weight is input in Bayes classifier by S204, is carried out to Bayes classifier
Training.
By under the scene, at least one label with weight is input in Bayes classifier, to Bayes classifier
It is trained, includes the pre- of at least one corresponding label of each scene in multiple scenes in the Bayes classifier after training
Survey probability.For example, as shown in table 2 below, specific 5 scenes and the corresponding prediction of each scene in table to choose
Higher preceding 5 labels of probability；Wherein, the higher label of the corresponding prediction probability of marsh scene be wetland, nature reserve area,
Water, marsh, river mouth；The corresponding prediction probability of wetland is 0.042475076257328485.
Table 2, the schematic diagram of Bayes classifier middle part branch scape after training
Specifically, at least one corresponding label of picture is input in trained Bayes classifier, can be somebody's turn to do
Picture is respectively labeled as the prediction probability of each scene at least one scene.For example, by taking the picture illustrated in S101 as an example
It is illustrated, the corresponding label of the picture includes water, wing, marine organisms, fish, shark, dolphin, biology etc., which is obtained
Above-mentioned label be input in Bayes classifier, the picture can be obtained and be respectively labeled as each scene at least one scene
Prediction probability: sea floor world, 0.933；Aquarium, 0.067；Ocean, 0.0；Swimming pool, 0.0；Ship travelling, 0.0；Deng.
Further, above-mentioned S102 is accepted, the method for picture modification further includes S103.
S103, according at least one label of picture, obtained from the modification database of foundation at least one modification data,
And in labeled at least one scene of every modification data each scene prediction probability, and obtain at least one label
The rare degree of each label.
In the embodiment of the present application, the modification data modified in database can be text data, image data etc., wherein
The particular content of modification data can be configured according to the actual situation, and the application is to this without specifically limiting.
In the embodiment of the present application, the establishment process of modification database is illustrated by taking text data as an example, specifically, ginseng
The flow chart of the process of a kind of foundation modification database as shown in Figure 3, wherein establish the process of modification database, comprising:
S301 acquires a plurality of modification data, for every modification data, executes S302-S305.
In the embodiment of the present application, a plurality of modification data can be acquired, wherein can be in social platform, search platform, book
In nationality, image data etc., a plurality of modification data are acquired.
The modification data are carried out word segmentation processing, obtain at least one corresponding word of the modification data by S302；
Word segmentation processing is carried out to every modification data in a plurality of modification data collected in S301, obtains every modification
At least one corresponding word of data, for example, modification data: not every fish can all live in a piece of nautical mile.At participle
The word obtained after reason includes: fish, life, sea.
S303 determines the corresponding label of modification data according at least one word；
Specifically, determining the corresponding label of modification data according at least one word, process includes:
For each word, following processes are executed:
It is identified firstly, word is input in preset tag library, wherein it include multiple tally sets in tag library,
It wherein, include the corresponding synonym of the label in each tally set；Secondly, being determined according to tag library to the recognition result of word
The corresponding label of word；Then, by the corresponding label of word, it is determined as the corresponding label of modification data.
In the embodiment of the present application, the example accepted in S302 is illustrated, at least one word includes: fish, life, sea,
And include the tally set in sea in multiple tally sets in tag library, it include that the sea is corresponding synonymous in the tally set in sea
Word, such as sea, ocean etc..It is then handled for word sea, sea is input in preset tag library and is identified, sea is obtained
Corresponding label is sea, then label: sea is the corresponding label of the modification data, and so on, then fish, life can be obtained
Corresponding label living, can further obtain modification data: not every fish can all live in a piece of nautical mile, corresponding
Label: sea, life, fish.
S304, the corresponding label of modification data is defeated into trained Bayes classifier, obtain the modification data
The prediction probability of labeled at least one scene and each scene；
In the embodiment of the present application, the example accepted in S303 is illustrated, by the corresponding label of modification data: sea,
Life, fish, are input in trained Bayes classifier, can obtain at least one labeled scene of the modification data,
And the prediction probability of each scene, for example, aquarium, 0.855；Sea floor world, 0.768；Ship travelling, 0.213；Deng.Specifically
, foregoing description is merely illustrative citing, is not especially limited.
At least one labeled scene of every modification data in S305, a plurality of modification data and a plurality of modification data,
The prediction probability of each scene constitutes modification database.
Further, as an alternative embodiment, the process of modification database is established, further includes:
Firstly, a plurality of number for modifying each label at least one corresponding label of data of statistics；
Secondly, determining the rare degree of the label according to the corresponding number of label；
Finally, the rare degree of each label at least one corresponding label of a plurality of modification data is input to modification data
In library.
In the embodiment of the present application, for example: the calculation formula of rare degree is a=n/c, and a is the rare degree of label, and n is
Positive integer, c is the corresponding number of label, and the value of n can be configured according to the actual situation, such as n=3.Specifically, rare
The calculation formula of degree, which can according to need, to be determined, and the application is merely illustrative description, is not especially limited.
In the embodiment of the present application, the number of each label in all labels for including in a plurality of modification data, example are counted
Such as, if a plurality of modify label in data: the number that sea occurs is 50, can determine the label: the corresponding rare degree in sea, and
By the label: the corresponding rare degree in sea is input in modification database, further, the institute that will include in a plurality of modification data
There is the rare degree of each label in label to be input in modification database.
Further, above-mentioned S103 is accepted, the method for picture modification further includes S104.
S104, for every modification data at least one modification data of acquisition, according to the corresponding mark of modification data
The prediction probability and picture of each scene are labeled at least one labeled scene of the rare degree of label, modification data
The prediction probability of each scene at least one scene calculates the score of modification data；
In the embodiment of the present application, the score of modification data can be calculated according to matching score formula, for example, the matching
Dividing formula is b=m*a1+m*a2+…+m*ai+g*x1*y1+…+g*xj*yj, wherein b is the score for modifying data, and m is rare
Parameter is spent, g is prediction probability parameter, aiFor the corresponding rare degree of label i, xjThe prediction for corresponding to scene j for the picture to be matched is general
Rate, yjThe prediction probability of scene j is corresponded to for the modification data, the value of m, g can be just configured according to the actual situation.Specifically
, matching score formula can be configured according to actual needs, and the application is not especially limited.
Further, above-mentioned S104 is accepted, the method for picture modification further includes S105, S106.
S105 modifies the score of data according to every, from least one modification data of acquisition, selects preset quantity
Modify data；
In the embodiment of the present application, for example, the score of data can be modified according to every, score higher preceding 3 is selected
Data alternately modifies data.
The modification data of the preset quantity of selection are combined with picture to be matched respectively, generate multiple modifications by S106
Picture afterwards, and the picture after modification is sent to user terminal.
In the embodiment of the present application, by the modification data of the preset data of selection, i.e., alternatively modification data respectively with it is to be matched
Picture is combined, the picture after generating multiple modifications, and the picture after modification is sent to user terminal, and user can be in user
The picture that end selects oneself desired from the picture after multiple modifications.Wherein, modification data are combined with picture to be matched
When, a template can be selected from preset template library, and modification data and the template of picture to be matched and selection are combined,
Picture after generating modification.
Exemplary illustration obtains picture to be matched, identifies at least one label of the picture, at least one label packet
It includes: water, wing, marine organisms, fish, shark, biology, dolphin, marine mammal, cartilaginous fish, hotel, osprey reef etc., it will be upper
The label for stating acquisition is input in trained Bayes classifier, is obtained the picture and is respectively labeled as at least one scene
The prediction probability of each scene, specifically, the prediction probability of corresponding scene and scene be sea floor world, 0.067；The Shui nationality
Shop, 0.933；Ocean, 0.0；Swimming pool, 0.0；Ship travelling, 0.0.Meanwhile according to the label of above-mentioned acquisition, from the modification of foundation
In database, each scene at least one scene that at least one modification data and every modification data are labeled is obtained
Prediction probability, and obtain the rare degree of each label at least one label, for example, the rare degree of water is 1.12, wing it is dilute
Degree of having is 3.0, and the rare degree of marine organisms is 2.0, etc.；Modify data are as follows: and dolphin has a talk about words, changes a good mood, then should
Modifying the corresponding label of data is dolphin, and prediction probability is 0.764 when which is aquarium, the modification
Prediction probability is 0.023 when the corresponding scene of data is sea floor world；Modify data are as follows: I is still delithted with you, as whale is sunken to
The breathing of seabed tenderness, extremely angry pole of being crazy about, then the corresponding label of modification data is marine organisms, and corresponding scene is pre- when being aquarium
Surveying probability is 0.644, and prediction probability is 0.262 when which is sea floor world, and the modification data are corresponding
Scene when being ocean prediction probability be 0.006.Further, for every modification at least one modification data of acquisition
Data, according to each scene at least one labeled scene of the rare degree of the corresponding label of modification data, modification data
Prediction probability and each scene in labeled at least one scene of picture prediction probability, calculate the modification data
Score.For example, modification data: I is still delithted with you, as whale is sunken to the breathing of seabed tenderness, silly extremely angry pole.According in S104
Matching score formula b=m*a1+m*a2+…+m*ai+g*x1*y1+…+g*xj*yj, wherein the value of m=1, g=5, m and g can
To be selected according to the actual situation, the score of the modification data is calculated are as follows: b=1*2+5*0.644*0.933+5*0.262*
0.067+5*0.006*0.0=5.09203 is approximately 5.1.And so on can be obtained at least one modification data in
Every modification data score, select the higher 3 modifications data of score, respectively by 3 modification data of selection with it is to be matched
Picture is combined by the template of selection, and the picture after modification is sent to user terminal by the picture after generating modification.
The method of a kind of picture modification provided by the embodiments of the present application, by trained Bayes classifier to picture
The prediction probability of scene and scene is determined, and obtains at least one from the modification database of foundation according to the label of picture
Data, and the score by calculating every obtained modification data are modified, selects to repair from least one modification data of acquisition
Decorations data modify picture to be matched, so that the material that server can be used for modifying is more, the picture after improving modification
Effect.
Based on the same inventive concept, picture corresponding with the method for picture modification is additionally provided in the embodiment of the present application to modify
Device, the method for the above-mentioned picture modification of the principle and the embodiment of the present application solved the problems, such as due to the device in the embodiment of the present application
It is similar, therefore the implementation of device may refer to the implementation of method, overlaps will not be repeated.
Embodiment two
The embodiment of the present application provides a kind of device of picture modification, as shown in Fig. 4 a kind of device of picture modification
Structural block diagram, which includes:
Picture recognition module 41 for obtaining picture to be matched, and identifies at least one of the picture to be matched
Label；
Scene prediction module 42, at least one label of picture to be input in trained Bayes classifier,
Obtain the prediction probability that picture is respectively labeled as each scene at least one scene；
Modification data acquisition module 43 is obtained from the modification database of foundation at least one label according to picture
The prediction probability of each scene at least one scene for taking at least one modification data and every modification data to be labeled,
And obtain the rare degree of each label at least one label；
Points calculating module 44, for for every modification data in at least one modification data obtained, according to repairing
Adorn the prediction probability of each scene at least one labeled scene of the rare degree of the corresponding label of data, modification data, with
And the prediction probability of each scene in labeled at least one scene of picture, calculate the score of modification data；
Data selecting module 45 is modified, for modifying the score of data according to every, modifies number from at least one of acquisition
In, the modification data of preset quantity are selected；
Picture modifies module 46, for tying the modification data of the preset quantity of selection with picture to be matched respectively
It closes, the picture after generating multiple modifications, and the picture after modification is sent to user terminal.
As an alternative embodiment, the device of picture modification further include: Bayes classifier training module；
Bayes classifier training module, is specifically used for:
Multiple scenes are constructed, for each scene, are executed:
The corresponding a plurality of samples pictures of the scene are obtained, at least one corresponding mark is labeled in every samples pictures
Label；
The weight of the label is determined according to the quantity of the label in a plurality of samples pictures for any one label；
At least one label with weight is input in Bayes classifier, Bayes classifier is trained.
As an alternative embodiment, the device of picture modification further include: modification Database module；
Database module is modified, is specifically used for:
A plurality of modification data are acquired, for every modification data, are executed:
The modification data are subjected to word segmentation processing, obtain at least one corresponding word of the modification data；
According at least one word, the corresponding label of modification data is determined；
The corresponding label of modification data is input in trained Bayes classifier, the modification data is obtained and is marked
At least one scene of note and the prediction probability of each scene；
It is at least one labeled scene of every modification data in a plurality of modification data and a plurality of modification data, each
The prediction probability of scene constitutes the modification database.
As an alternative embodiment, Database module is modified, also particularly useful for:
For each word, execute:
The word is input in preset tag library and is identified, wherein it include multiple tally sets in tag library,
In, it include the corresponding synonym of the label in each tally set；
According to tag library to the recognition result of word, the corresponding label of word is determined；
By the corresponding label of word, it is determined as the corresponding label of modification data.
As an alternative embodiment, Database module is modified, also particularly useful for:
Count the number of each label at least one corresponding label of a plurality of modification data；
According to the corresponding number of label, the rare degree of label is determined；
The rare degree of each label at least one corresponding label of a plurality of modification data is input in modification database.
The device of picture modification provided by the embodiments of the present application, the method for the picture modification provided with above-described embodiment one have
There is identical technical characteristic, so also can solve identical technical problem, reaches identical technical effect.
Embodiment three
Based on same technical concept, the embodiment of the present application also provides a kind of electronic equipment.It referring to Figure 5, is this Shen
Please the structural schematic diagram of electronic equipment 500 that provides of embodiment, including processor 501, memory 502 and bus 503.Wherein,
Memory 502 is executed instruction for storing, including memory 5021 and external memory 5022；Here memory 5021 is also referred to as memory
Reservoir, for temporarily storing the operational data in processor 501, and the data exchanged with external memories 5022 such as hard disks,
Processor 501 carries out data exchange by memory 5021 and external memory 5022, when electronic equipment 500 is run, processor
It is communicated between 501 and memory 502 by bus 503, so that processor 501 is being executed to give an order:
Picture to be matched is obtained, and identifies at least one label of picture to be matched；
At least one label of picture is input in trained Bayes classifier, picture is obtained and is respectively labeled as
The prediction probability of each scene at least one scene；
According at least one label of picture, obtained from the modification database of foundation at least one modification data and
The prediction probability of each scene at least one labeled scene of every modification data, and obtain each at least one label
The rare degree of label；
For every modification data at least one modification data of acquisition, according to the dilute of the corresponding label of modification data
The prediction probability and picture of each scene are labeled at least one labeled scene of degree of having, modification data at least one
The prediction probability of each scene in a scene calculates the score of modification data；
The score that data are modified according to every selects the modification of preset quantity from least one modification data of acquisition
Data；
The modification data of the preset quantity of selection are combined with picture to be matched respectively, the figure after generating multiple modifications
Piece, and the picture after modification is sent to user terminal.
In a kind of possible design, in the instruction that processor 501 may execute, further includes:
The training process of Bayes classifier, comprising:
Multiple scenes are constructed, for each scene, are executed:
The corresponding a plurality of samples pictures of scene are obtained, are labeled at least one corresponding label in every samples pictures；
The weight of the label is determined according to the quantity of the label in a plurality of samples pictures for any one label；
At least one label with weight is input in Bayes classifier, Bayes classifier is trained.
In a kind of possible design, in the instruction that processor 501 may execute, further includes:
Establish the process of modification database, comprising:
A plurality of modification data are acquired, for every modification data, are executed:
The modification data are subjected to word segmentation processing, obtain at least one corresponding word of the modification data；
According at least one word, the corresponding label of modification data is determined；
The corresponding label of modification data is input in trained Bayes classifier, the modification data is obtained and is marked
At least one scene of note and the prediction probability of each scene；
It is at least one labeled scene of every modification data in a plurality of modification data and a plurality of modification data, each
The prediction probability of scene constitutes modification database.
In a kind of possible design, in the instruction that processor 501 may execute, further includes:
According at least one described word, the corresponding label of modification data is determined, comprising:
For each word, execute:
Word is input in preset tag library and is identified, wherein includes multiple tally sets in tag library, wherein
It include the corresponding synonym of the label in each tally set；
According to tag library to the recognition result of word, the corresponding label of word is determined；
By the corresponding label of word, it is determined as the corresponding label of modification data.
In a kind of possible design, in the instruction that processor 501 may execute, further includes:
Establish the process of modification database, further includes:
Count the number of each label at least one corresponding label of a plurality of modification data；
According to the corresponding number of label, the rare degree of label is determined；
The rare degree of each label at least one corresponding label of a plurality of modification data is input in modification database.
Example IV
The embodiment of the present application also provides a kind of computer readable storage medium, is stored on the computer readable storage medium
Computer program, which executes any of the above-described picture modification as described in the examples method when being run by processor
The step of.
Specifically, which can be general storage medium, such as mobile disk, hard disk, on the storage medium
Computer program when being run, the step of being able to carry out the method for above-mentioned picture modification, to improve the picture after modification
Effect.
The computer program product of the method for picture modification is carried out provided by the embodiment of the present application, including stores processing
The computer readable storage medium of the executable non-volatile program code of device, the instruction that said program code includes can be used for holding
Row previous methods method as described in the examples, specific implementation can be found in embodiment of the method, and details are not described herein.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description,
The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In several embodiments provided herein, it should be understood that disclosed systems, devices and methods, it can be with
It realizes by another way.The apparatus embodiments described above are merely exemplary, for example, the division of the unit,
Only a kind of logical function partition, there may be another division manner in actual implementation, in another example, multiple units or components can
To combine or be desirably integrated into another system, or some features can be ignored or not executed.Another point, it is shown or beg for
The mutual coupling, direct-coupling or communication connection of opinion can be through some communication interfaces, device or unit it is indirect
Coupling or communication connection can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.
It, can be with if the function is realized in the form of SFU software functional unit and when sold or used as an independent product
It is stored in the executable non-volatile computer-readable storage medium of a processor.Based on this understanding, the application
Technical solution substantially the part of the part that contributes to existing technology or the technical solution can be with software in other words
The form of product embodies, which is stored in a storage medium, including some instructions use so that
One computer equipment (can be personal computer, server or the network equipment etc.) executes each embodiment institute of the application
State all or part of the steps of method.And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (Read-Only
Memory, ROM), random access memory (Random Access Memory, RAM), magnetic or disk etc. is various to deposit
Store up the medium of program code.
Finally, it should be noted that embodiment described above, the only specific embodiment of the application, to illustrate the application
Technical solution, rather than its limitations, the protection scope of the application is not limited thereto, although with reference to the foregoing embodiments to this Shen
It please be described in detail, those skilled in the art should understand that: anyone skilled in the art
Within the technical scope of the present application, it can still modify to technical solution documented by previous embodiment or can be light
It is readily conceivable that variation or equivalent replacement of some of the technical features；And these modifications, variation or replacement, do not make
The essence of corresponding technical solution is detached from the spirit and scope of the embodiment of the present application technical solution, should all cover the protection in the application
Within the scope of.Therefore, the protection scope of the application shall be subject to the protection scope of the claim.
Claims (10)
1. a kind of method of picture modification, which is characterized in that the described method includes:
Picture to be matched is obtained, and identifies at least one label of the picture to be matched；
At least one label of the picture is input in trained Bayes classifier, the picture is obtained and is marked respectively
It is denoted as the prediction probability of each scene at least one scene；
According at least one label of the picture, obtained from the modification database of foundation at least one modification data and
The prediction probability of each scene at least one labeled scene of every modification data, and obtain each at least one label
The rare degree of label；
For every modification data at least one modification data of acquisition, according to the dilute of the corresponding label of the modification data
The prediction probability and the picture of each scene are labeled at least one labeled scene of degree of having, the modification data
At least one scene in each scene prediction probability, calculate it is described modification data score；
The score that data are modified according to every selects the modification data of preset quantity from least one modification data of acquisition；
The modification data of the preset quantity of selection are combined with picture to be matched respectively, the picture after generating multiple modifications,
And the picture after modification is sent to user terminal.
2. the method according to claim 1, wherein the training process of the Bayes classifier, comprising:
Multiple scenes are constructed, for each scene, are executed:
The corresponding a plurality of samples pictures of the scene are obtained, are labeled at least one corresponding label in every samples pictures；
The weight of the label is determined according to the quantity of the label in a plurality of samples pictures for any one label；
At least one label with weight is input in Bayes classifier, the Bayes classifier is trained.
3. according to the method described in claim 2, it is characterized in that, establishing the process of the modification database, comprising:
A plurality of modification data are acquired, for every modification data, are executed:
The modification data are subjected to word segmentation processing, obtain at least one corresponding word of the modification data；
According at least one described word, the corresponding label of modification data is determined；
The corresponding label of modification data is input in trained Bayes classifier, obtains what the modification data were labeled
The prediction probability of at least one scene and each scene；
At least one labeled scene of every modification data in a plurality of modification data and a plurality of modification data,
The prediction probability of each scene constitutes the modification database.
4. according to the method described in claim 3, it is characterized in that, described at least one word according to, determines the modification
The corresponding label of data, comprising:
For each word, execute:
The word is input in preset tag library and is identified, wherein it include multiple tally sets in the tag library,
In, it include the corresponding synonym of the label in each tally set；
According to tag library to the recognition result of the word, the corresponding label of the word is determined；
By the corresponding label of the word, it is determined as the corresponding label of modification data.
5. according to the method described in claim 4, it is characterized in that, establishing the process of the modification database, further includes:
Count the number of each label at least one corresponding label of a plurality of modification data；
According to the corresponding number of the label, the rare degree of the label is determined；
The rare degree of each label at least one corresponding label of a plurality of modification data is input in the modification database.
6. a kind of device of picture modification, which is characterized in that described device includes:
Picture recognition module for obtaining picture to be matched, and identifies at least one label of the picture to be matched；
Scene prediction module is obtained at least one label of the picture to be input in trained Bayes classifier
The prediction probability of each scene at least one scene is respectively labeled as to the picture；
Modification data acquisition module is obtained from the modification database of foundation at least one label according to the picture
The prediction probability of each scene at least one scene that at least one modification data and every modification data are labeled, and
Obtain the rare degree of each label at least one label；
Points calculating module, for modifying every modification data in data at least one obtained, according to the modification
The prediction probability of each scene at least one labeled scene of the rare degree of the corresponding label of data, the modification data,
And the prediction probability of each scene in labeled at least one scene of the picture, calculate the modification data
Point；
Data selecting module is modified, for modifying the score of data according to every, from least one modification data of acquisition, choosing
Select the modification data of preset quantity；
Picture modifies module, for being combined the modification data of the preset quantity of selection with picture to be matched respectively, generates
Picture after multiple modifications, and the picture after modification is sent to user terminal.
7. device according to claim 6, which is characterized in that described device further include: Bayes classifier training module；
The Bayes classifier training module, is specifically used for:
Multiple scenes are constructed, for each scene, are executed:
The corresponding a plurality of samples pictures of the scene are obtained, are labeled at least one corresponding label in every samples pictures；
The weight of the label is determined according to the quantity of the label in a plurality of samples pictures for any one label；
At least one label with weight is input in Bayes classifier, the Bayes classifier is trained.
8. device according to claim 7, which is characterized in that described device further include: modification Database module；
The modification Database module, is specifically used for:
A plurality of modification data are acquired, for every modification data, are executed:
The modification data are subjected to word segmentation processing, obtain at least one corresponding word of the modification data；
According at least one described word, the corresponding label of modification data is determined；
The corresponding label of modification data is input in trained Bayes classifier, obtains what the modification data were labeled
The prediction probability of at least one scene and each scene；
At least one labeled scene of every modification data in a plurality of modification data and a plurality of modification data,
The prediction probability of each scene constitutes the modification database.
9. a kind of electronic equipment characterized by comprising processor, memory and bus, the memory are stored with the place
The executable machine readable instructions of device are managed, when electronic equipment operation, pass through bus between the processor and the memory
Communication, the machine readable instructions execute picture modification as claimed in claim 1 to 5 when being executed by the processor
The step of method.
10. a kind of computer readable storage medium, which is characterized in that be stored with computer journey on the computer readable storage medium
Sequence, the computer program execute the step of the method for picture modification as claimed in claim 1 to 5 when being run by processor
Suddenly.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910362660.7A CN110070512B (en) | 2019-04-30 | 2019-04-30 | Picture modification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910362660.7A CN110070512B (en) | 2019-04-30 | 2019-04-30 | Picture modification method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110070512A true CN110070512A (en) | 2019-07-30 |
| CN110070512B CN110070512B (en) | 2021-06-01 |
Family
ID=67370056
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910362660.7A Active CN110070512B (en) | 2019-04-30 | 2019-04-30 | Picture modification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110070512B (en) |
Citations (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040070598A1 (en) * | 2002-09-24 | 2004-04-15 | Fuji Photo Film Co., Ltd. | Image retouching method, apparatus, and program storage medium, image correcting method, apparatus, and program storage medium, and eye detecting and correcting method apparatus, and program storage medium |
| DE10251419A1 (en) * | 2002-11-05 | 2004-05-19 | Proeck-Zvlcil, Christian, von | Method for releasing/retouching image themes/motives e.g. for automotive work, involves positioning duplicate-plane over locations where chrominance of background is continued in foreground |
| CN101571856A (en) * | 2008-04-30 | 2009-11-04 | 福特安（苏州）图像管理有限公司 | Method for deducing theme and scene in image |
| KR20100055941A (en) * | 2008-11-18 | 2010-05-27 | 삼성전자주식회사 | Apparatus and method for previously becoming known to a level of face compensation in digital image processing device |
| CN101848303A (en) * | 2009-03-27 | 2010-09-29 | 夏普株式会社 | Image processing apparatus, image processing system and image processing method |
| CN102027505A (en) * | 2008-07-30 | 2011-04-20 | 泰塞拉技术爱尔兰公司 | Automatic face and skin beautification using face detection |
| CN103336961A (en) * | 2013-07-22 | 2013-10-02 | 中国科学院自动化研究所 | Interactive natural scene text detection method |
| CN103473327A (en) * | 2013-09-13 | 2013-12-25 | 广东图图搜网络科技有限公司 | Image retrieval method and image retrieval system |
| CN103559220A (en) * | 2013-10-18 | 2014-02-05 | 北京奇虎科技有限公司 | Image searching device, method and system |
| JP2015041261A (en) * | 2013-08-22 | 2015-03-02 | 株式会社リコー | Image processing system, image processing method, and program |
| CN104951554A (en) * | 2015-06-29 | 2015-09-30 | 浙江大学 | Method for matching landscape with verses according with artistic conception of landscape |
| US20160055237A1 (en) * | 2014-08-20 | 2016-02-25 | Mitsubishi Electric Research Laboratories, Inc. | Method for Semantically Labeling an Image of a Scene using Recursive Context Propagation |
| CN106296621A (en) * | 2015-05-22 | 2017-01-04 | 腾讯科技（深圳）有限公司 | Image processing method and device |
| CN106446782A (en) * | 2016-08-29 | 2017-02-22 | 北京小米移动软件有限公司 | Image identification method and device |
| CN106844659A (en) * | 2017-01-23 | 2017-06-13 | 宇龙计算机通信科技(深圳)有限公司 | A kind of multimedia data processing method and device |
| CN107247755A (en) * | 2017-05-27 | 2017-10-13 | 深圳市唯特视科技有限公司 | A kind of personalized image method for generating captions based on context serial memorization network |
| CN107862239A (en) * | 2017-09-15 | 2018-03-30 | 广州唯品会研究院有限公司 | A kind of combination text carries out the method and its device of picture recognition with picture |
| CN108665414A (en) * | 2018-05-10 | 2018-10-16 | 上海交通大学 | Natural scene picture generation method |
| CN108764370A (en) * | 2018-06-08 | 2018-11-06 | Oppo广东移动通信有限公司 | Image processing method, device, computer readable storage medium and computer equipment |
-
2019
- 2019-04-30 CN CN201910362660.7A patent/CN110070512B/en active Active
Patent Citations (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20040070598A1 (en) * | 2002-09-24 | 2004-04-15 | Fuji Photo Film Co., Ltd. | Image retouching method, apparatus, and program storage medium, image correcting method, apparatus, and program storage medium, and eye detecting and correcting method apparatus, and program storage medium |
| DE10251419A1 (en) * | 2002-11-05 | 2004-05-19 | Proeck-Zvlcil, Christian, von | Method for releasing/retouching image themes/motives e.g. for automotive work, involves positioning duplicate-plane over locations where chrominance of background is continued in foreground |
| CN101571856A (en) * | 2008-04-30 | 2009-11-04 | 福特安（苏州）图像管理有限公司 | Method for deducing theme and scene in image |
| CN102027505A (en) * | 2008-07-30 | 2011-04-20 | 泰塞拉技术爱尔兰公司 | Automatic face and skin beautification using face detection |
| KR20100055941A (en) * | 2008-11-18 | 2010-05-27 | 삼성전자주식회사 | Apparatus and method for previously becoming known to a level of face compensation in digital image processing device |
| CN101848303A (en) * | 2009-03-27 | 2010-09-29 | 夏普株式会社 | Image processing apparatus, image processing system and image processing method |
| CN103336961A (en) * | 2013-07-22 | 2013-10-02 | 中国科学院自动化研究所 | Interactive natural scene text detection method |
| JP2015041261A (en) * | 2013-08-22 | 2015-03-02 | 株式会社リコー | Image processing system, image processing method, and program |
| CN103473327A (en) * | 2013-09-13 | 2013-12-25 | 广东图图搜网络科技有限公司 | Image retrieval method and image retrieval system |
| CN103559220A (en) * | 2013-10-18 | 2014-02-05 | 北京奇虎科技有限公司 | Image searching device, method and system |
| US20160055237A1 (en) * | 2014-08-20 | 2016-02-25 | Mitsubishi Electric Research Laboratories, Inc. | Method for Semantically Labeling an Image of a Scene using Recursive Context Propagation |
| CN106296621A (en) * | 2015-05-22 | 2017-01-04 | 腾讯科技（深圳）有限公司 | Image processing method and device |
| CN104951554A (en) * | 2015-06-29 | 2015-09-30 | 浙江大学 | Method for matching landscape with verses according with artistic conception of landscape |
| CN106446782A (en) * | 2016-08-29 | 2017-02-22 | 北京小米移动软件有限公司 | Image identification method and device |
| CN106844659A (en) * | 2017-01-23 | 2017-06-13 | 宇龙计算机通信科技(深圳)有限公司 | A kind of multimedia data processing method and device |
| CN107247755A (en) * | 2017-05-27 | 2017-10-13 | 深圳市唯特视科技有限公司 | A kind of personalized image method for generating captions based on context serial memorization network |
| CN107862239A (en) * | 2017-09-15 | 2018-03-30 | 广州唯品会研究院有限公司 | A kind of combination text carries out the method and its device of picture recognition with picture |
| CN108665414A (en) * | 2018-05-10 | 2018-10-16 | 上海交通大学 | Natural scene picture generation method |
| CN108764370A (en) * | 2018-06-08 | 2018-11-06 | Oppo广东移动通信有限公司 | Image processing method, device, computer readable storage medium and computer equipment |
Non-Patent Citations (2)
| Title |
|---|
| XIAOHUI WANG ET AL.: "Affective image adjustment with a single word", 《THE VISUAL COMPUTER》 * |
| 朱钰恒等: "采用叠加图像的浅浮雕生成算法及实现", 《华侨大学学报（自然科学版）》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110070512B (en) | 2021-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US12014259B2 (en) | Generating natural language descriptions of images | |
| Baldassarre et al. | Deep koalarization: Image colorization using cnns and inception-resnet-v2 | |
| EP3186753B1 (en) | Processing images using deep neural networks | |
| CN108846375B (en) | A kind of multi-modal Cooperative Study method and device neural network based | |
| CN105205453B (en) | Human eye detection and localization method based on depth self-encoding encoder | |
| CN111091151B (en) | Construction method of generation countermeasure network for target detection data enhancement | |
| CN109034152A (en) | License plate locating method and device based on LSTM-CNN built-up pattern | |
| CN109961442A (en) | Training method, device and the electronic equipment of neural network model | |
| CN109919110A (en) | Video area-of-interest-detection method, device and equipment | |
| CN108876935A (en) | A kind of method and device quickly carrying out house threedimensional model splicing in mobile terminal | |
| Jue | Proteus and the digital: Scalar transformations of seawater’s materiality in ocean animations | |
| CN109523558A (en) | A kind of portrait dividing method and system | |
| CN113838158B (en) | Image and video reconstruction method and device, terminal equipment and storage medium | |
| Zhang et al. | Research on fish identification in tropical waters under unconstrained environment based on transfer learning | |
| CN110070512A (en) | The method and device of picture modification | |
| CN117611938A (en) | Multi-mode model training method, device, equipment and storage medium | |
| Cheung et al. | Conquering the Electron: The Geniuses, Visionaries, Egomaniacs, and Scoundrels who Built Our Electronic Age | |
| CN116433468A (en) | Data processing method and device for image generation | |
| CN114973424A (en) | Feature extraction model training method, hand action recognition method, device and electronic equipment | |
| Sra et al. | Deepspace: Mood-based image texture generation for virtual reality from music | |
| CN110428051A (en) | A kind of method and system being multiplexed deep neural network model training pattern | |
| Połap | Designing mazes for 2D games by artificial ant colony algorithm | |
| CN110826726A (en) | Object processing method, object processing apparatus, object processing device, and medium | |
| Gong et al. | E $^ 2$ GAN: Efficient Training of Efficient GANs for Image-to-Image Translation | |
| Pei et al. | The Influence of Modern Information Technology on the Future of Museums |
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 |