CN109157842A - A kind of cultural method and device of the develop class game role based on machine learning - Google Patents
A kind of cultural method and device of the develop class game role based on machine learning Download PDFInfo
- Publication number
- CN109157842A CN109157842A CN201811003307.1A CN201811003307A CN109157842A CN 109157842 A CN109157842 A CN 109157842A CN 201811003307 A CN201811003307 A CN 201811003307A CN 109157842 A CN109157842 A CN 109157842A
- Authority
- CN
- China
- Prior art keywords
- things
- game role
- target game
- role
- training
- 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
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
- A63F13/00—Video games, i.e. games using an electronically generated display having two or more dimensions
- A63F13/80—Special adaptations for executing a specific game genre or game mode
- A63F13/825—Fostering virtual characters
-
- A—HUMAN NECESSITIES
- A63—SPORTS; GAMES; AMUSEMENTS
- A63F—CARD, BOARD, OR ROULETTE GAMES; INDOOR GAMES USING SMALL MOVING PLAYING BODIES; VIDEO GAMES; GAMES NOT OTHERWISE PROVIDED FOR
- A63F2300/00—Features of games using an electronically generated display having two or more dimensions, e.g. on a television screen, showing representations related to the game
- A63F2300/80—Features of games using an electronically generated display having two or more dimensions, e.g. on a television screen, showing representations related to the game specially adapted for executing a specific type of game
- A63F2300/8058—Virtual breeding, e.g. tamagotchi
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Electrically Operated Instructional Devices (AREA)
- Image Analysis (AREA)
Abstract
The cultural method and device of the invention discloses a kind of develop class game role based on machine learning, belong to machine learning techniques and game design technical field.The described method includes: receiving the request to create for carrying the target game role of character types information；The corresponding things of the character types information is identified into initial model, is determined as the things identification model of the target game role；The recognition training material for the target game role that user provides is obtained, the recognition training material includes at least video, picture or the audio for being marked with things mark；Based on the recognition training material of the target game role, model training is carried out to the things identification model of the target game role in a manner of deep learning.Using the present invention, the playability and interest of develop class game can be improved.
Description
Technical field
The present invention relates to machine learning techniques and game design technical field, in particular to a kind of supporting based on machine learning
At the cultural method and device of class game role.
Background technique
With the continuous development of computer technology and network technology, electronic game has become extensive common entertainment item instantly
Mesh, wherein develop class game is a kind of game more popular with players, and player can simulate the specific game of culture in gaming
Role (people or animal), makes it constantly grow up, to reach or trigger the specified conditions in game, to meet the achievement of player
Sense.
When playing develop class game, player initially can create or select an initial game role in game, so
Afterwards according to the setting in game, different game selection and operation are carried out, selects different training methods to initial game angle
Color is cultivated, and so as to push game process according to different game scenarios, triggers the different trips in different game process
Play event, the state finally realized different game outcomes or want the initial game role culture of game to player.
In the implementation of the present invention, the inventor finds that the existing technology has at least the following problems:
Game scenario and training method are pre-set in develop class game, if player meet it is same in game
Premise can obtain it is identical as a result, namely triggering game scenario it is identical, the state of the game role after culture
It is almost the same, so, after the playing of certain time, game easily losing novelty, playability and interest are greatly reduced.
Summary of the invention
In order to solve problems in the prior art, the develop class game based on machine learning that the embodiment of the invention provides a kind of
The cultural method and device of role.The technical solution is as follows:
In a first aspect, providing a kind of cultural method of develop class game role based on machine learning, the method packet
It includes:
Receive the request to create for carrying the target game role of character types information；
The corresponding things of the character types information is identified into initial model, is determined as the things of the target game role
Identification model；
The recognition training material for the target game role that user provides is obtained, the recognition training material includes at least
It is marked with video, picture or the audio of things mark；
Based on the recognition training material of the target game role, to the target game role in a manner of deep learning
Things identification model carry out model training.
Optionally, the recognition training material based on the target game role, to described in a manner of deep learning
The things identification model of target game role carries out model training, comprising:
The things identification feature in the recognition training material is determined in a manner of deep learning；
According to the things mark and the things identification feature training target trip marked in the recognition training material
The things identification model of play role, so that the things identification model determines the things mark of any things.
Optionally, the method also includes:
When receiving when the starting request of things identification competition for the target game role, random selection includes
Video to be identified, picture or the audio of things identification feature；
By the things identification model of target game role described in the video to be identified, picture or audio input, known
Not Chu things mark；
The things mark identified according to described and the video to be identified, picture or the corresponding things of audio identify, really
Determine things identification scoring.
Optionally, the recognition training material based on the target game role, to described in a manner of deep learning
The things identification model of target game role carries out model training, comprising:
Things movement and/or the things posture in the recognition training material, Yi Jisuo are determined in a manner of deep learning
It states things and acts corresponding limbs displacement and/or the corresponding position of the posture；
The things identification model of the target game role is trained according to limbs displacement and/or the position,
So that the things identification model imitates the things movement and/or the things posture.
Optionally, the method also includes:
When starting request of competition is imitated when receiving the limbs for the target game role, random selection includes
Video to be imitated, picture or the audio of things movement and/or things posture；
By the things identification model of target game role described in the video to be imitated, picture or audio input, mould is obtained
Imitative movement and/or imitation posture；-
According to the echomotism and/imitate posture, acted with the things in the video to be imitated, picture or audio
And/or the similarity of things posture, it generates limbs and imitates scoring.
Optionally, the method also includes:
Material is tested in the identification for obtaining the target game role that user provides, and calls the thing of the target game role
Object identification model identifies identification test material, and to the user feedback recognition result.
Optionally, the method also includes:
The corresponding role characteristics training material of the character types information is periodically obtained from preset special material database,
The role characteristics training material is exclusively used in training different role characteristics；
Based on role characteristics training material, the things of the target game role is identified in a manner of deep learning
Model carries out model training.
Optionally, the method also includes:
When receiving acquisition request of the other users for the recognition training material of the target game role, institute is judged
State the shared condition whether other users meet the recognition training material；
If it is satisfied, then being based on the recognition training material, the other users are specified in a manner of deep learning
The things identification model of game role carries out model training.
Second aspect provides a kind of culture apparatus of develop class game role based on machine learning, described device packet
It includes:
Receiving module, for receiving the request to create for carrying the target game role of character types information；
Determining module is determined as the target for the corresponding things of the character types information to be identified initial model
The things identification model of game role；
Module is obtained, the recognition training material of the target game role for obtaining user's offer, the identification instruction
Practice material and includes at least video, picture or the audio for being marked with things mark；
Training module, for the recognition training material based on the target game role, to institute in a manner of deep learning
The things identification model for stating target game role carries out model training.
Optionally, the training module, is specifically used for:
The things identification feature in the recognition training material is determined in a manner of deep learning；
According to the things mark and the things identification feature training target trip marked in the recognition training material
The things identification model of play role, so that the things identification model determines the things mark of any things.
Optionally, described device further include:
Selecting module, for when receive for the target game role things identification compare with start request when,
Random selection includes video to be identified, picture or the audio of things identification feature；
Identification module, for knowing the things of target game role described in the video to be identified, picture or audio input
Other model, the things mark identified；
Grading module, the things mark for identifying according to are corresponding with the video to be identified, picture or audio
Things mark, determine things identification scoring.
Optionally, the training module, is specifically used for:
Things movement and/or the things posture in the recognition training material, Yi Jisuo are determined in a manner of deep learning
It states things and acts corresponding limbs displacement and/or the corresponding position of the posture；
The things identification model of the target game role is trained according to limbs displacement and/or the position,
So that the things identification model imitates the things movement and/or the things posture.
Optionally, described device further include:
Selecting module, for when receive for the target game role limbs imitate compare with start request when,
Random selection includes video to be imitated, picture or the audio of things movement and/or things posture；
Identification module, for knowing the things of target game role described in the video to be imitated, picture or audio input
Other model obtains echomotism and/or imitates posture；
Grading module is used for according to the echomotism and/imitation posture, with video, picture or the audio to be imitated
In things movement and/or things posture similarity, generate limbs imitate scoring.
Optionally, described device further include:
Test module tests material for obtaining the identification of the target game role of user's offer, calls the mesh
The things identification model for marking game role is tested material to the identification and is identified, and to the user feedback recognition result.
Optionally, the training module, is also used to:
The corresponding role characteristics training material of the character types information is periodically obtained from preset special material database,
The role characteristics training material is exclusively used in training different role characteristics；
Based on role characteristics training material, the things of the target game role is identified in a manner of deep learning
Model carries out model training.
Optionally, the training module, is also used to:
When receiving acquisition request of the other users for the recognition training material of the target game role, institute is judged
State the shared condition whether other users meet the recognition training material；
If it is satisfied, then being based on the recognition training material, the other users are specified in a manner of deep learning
The things identification model of game role carries out model training.
The third aspect, provides a kind of background server, and the background server includes processor and memory, described to deposit
Be stored at least one instruction, at least one section of program, code set or instruction set in reservoir, at least one instruction, it is described extremely
Few one section of program, the code set or instruction set loaded by the processor and executed with realize as described in relation to the first aspect based on
The cultural method of the develop class game role of machine learning.
Fourth aspect provides a kind of computer readable storage medium, at least one finger is stored in the storage medium
Enable, at least one section of program, code set or instruction set, at least one instruction, at least one section of program, the code set or
Instruction set is loaded by processor and is executed the develop class game role based on machine learning with realization as described in relation to the first aspect
Cultural method.
Technical solution provided in an embodiment of the present invention has the benefit that
In the embodiment of the present invention, the request to create for carrying the target game role of character types information is received, by role
The corresponding things of type information identifies initial model, is determined as the things identification model of target game role, obtains user and provides
Target game role recognition training material, recognition training material include at least be marked with things mark video, picture or
Audio, the recognition training material based on target game role identify the things of target game role in a manner of deep learning
Model carries out model training.In this way, by deep learning in conjunction with develop class game, it is excellent by the continuous personalized culture of user
Change the things identification model of game role in develop class game, so that game role can have different degrees of, different-style thing
Object recognition capability, so as to improve the playability and interest of develop class game.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings other
Attached drawing.
Fig. 1 is a kind of cultural method process of develop class game role based on machine learning provided in an embodiment of the present invention
Figure；
Fig. 2 is a kind of game main interface schematic diagram of develop class game provided in an embodiment of the present invention；
Fig. 3 is a kind of culture apparatus structure of develop class game role based on machine learning provided in an embodiment of the present invention
Schematic diagram；
Fig. 4 is a kind of structural schematic diagram of background server provided in an embodiment of the present invention.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached drawing to embodiment party of the present invention
Formula is described in further detail.
The cultural method of the embodiment of the invention provides a kind of develop class game role based on machine learning, this method
It can be used in the background server of develop class game, and assist realizing by game terminal.Wherein, background server and game
Terminal realizes the game function of develop class game using C/S model jointly, the two can by the game process of each self-operating come
Different game processings is executed, specifically, game terminal can be used for acquiring the game operation of user, completion partial game is patrolled
Game picture etc. after collecting processing, operating with background server interactive game data, presentation, background server can be used for storing
Game data, completion partial game processing logic and game terminal interactive game data of different user etc..Above-mentioned background service
Device may include processor, memory, transceiver, and processor can be used for carrying out the develop class game role in following processes
Culture processing, memory can be used for storing the data of the data and generation that need in following treatment processes, and transceiver can be with
For sending and receiving the related data in following treatment processes.
Below in conjunction with specific embodiment, process flow shown in FIG. 1 is described in detail, content can be as
Under:
Step 101, the request to create for carrying the target game role of character types information is received.
In an implementation, user can open the client for the develop class game installed on game terminal, and game terminal can be with
Show corresponding game main interface, as shown in Fig. 2, in game main interface can include at least " character list ", " hall ",
Four options such as " shop " and " individual center ", each option correspond to a game sub-interface.Wherein, " character list " sub- boundary
Can show all game roles that user is cultivating in face, user can in the sub-interface with each game role into
Row game interaction, such as check state, culture attribute, power of test；In " hall " sub-interface, user can be with other users
Game interaction is carried out, interaction mode includes making friends, chatting, compare with etc.；It may include game payment project in " shop " sub-interface,
Such as open special play role, purchase proprietary game is dressed up；Game account pipe can be provided in " individual center " sub-interface
The functions such as reason and suggestion feedback.In addition, background server can be to game end after game terminal and background server establish connection
End push frontier science and technology information and relevant knowledge question and answer, user are reading frontier science and technology information and are carrying out to knowledge question correct
After answering, available random game reward, so as to allow user to learn more about some forward position sections while playing game
Skill information.
In this way, the angle in above-mentioned " character list " sub-interface can be clicked when user wants the new game role of culture
Color creates key, then selects the character types of game role, wherein character types, which can be people, (can be subdivided into boy, girl
Deng), animal (cat, dog etc. can be subdivided into), virtual legend biological (dragon, spirit etc. can be subdivided into) etc., user's selection completion simultaneously
After being confirmed, game terminal can send the request to create of target game role to background server, correspondingly, backstage takes
Business device then can receive the request to create of the above-mentioned target game role for carrying character types information.
Step 102, the corresponding things of character types information is identified into initial model, is determined as the things of target game role
Identification model.
In an implementation, after background server receives the request to create of target game role, available request to create is taken
Then the corresponding things of character types information is identified initial model, is determined as target game role by the character types information of band
Things identification model.In turn, background server can carry out the things identification model of target game role and user account
Binding, and be stored in the corresponding memory space of user account.It is noted that the corresponding things of different role type information
Identify that initial model can be identical, i.e., all character types information correspond to the same things identification initial model, can also
With by the developer of develop class game, according to imaginary different role type to the Different Cognitive of identical things and different angles
Color type creates the familiarities of different things.Such as things " ball of yarn ", people may be identified as " textile product ",
And cat may be identified as " toy ", then can be known in advance using things of the picture for the ball of yarn for being marked with " textile product " to people
Other initial model is trained, and is carried out using things identification initial model of the picture for the ball of yarn for being marked with " toy " to cat
Training；In another example boy is generally familiar to things such as toy, game, then " boy " corresponding things identification initial model can
Initial training has been carried out to be in advance based on the materials such as toy, game by developer.In this way, different role type information is corresponding
Things identification initial model it is different, ensure that the diversity and interest of game, same character types information are corresponding
Things identifies that initial model is identical, ensure that the fairness of game.
Step 103, the recognition training material for the target game role that user provides is obtained.
Wherein, recognition training material includes at least video, picture or the audio for being marked with things mark；
It in an implementation, can be by providing the video for being marked with things mark, figure after user creates target game role
The recognition trainings material such as piece or audio, carrys out training objective game role.Specifically, user, which can upload, to be referred to for video material
Video in measured length, and at least one things label is marked for video content；For picture materials, user is in upper blit
When piece, the region for needing to identify in picture can be irised out, and inputs corresponding things mark；For audio material, user can be with
The audio in designated length is uploaded, and marks at least one things to identify for audio content.Correspondingly, background server can
To get the recognition training material that above-mentioned user provides, and recognition training material is arranged, then by the knowledge after arrangement
Not Xun Lian material be stored in the corresponding memory space of user account.The above-mentioned processing to recognition training material can specifically such as
Under: for video material, background server first can intercept video according to designated length, then with the frequency abstraction of 2 frame per second
Video pictures, then carry out PCA dimension-reduction treatment；For picture materials, the region that background server can iris out user is plucked out, and
It is zoomed in and out by specified wide height；For audio material, background server first can intercept audio according to designated length, then utilize
The audio that RNN processing is truncated to.In addition, background server can also be to the knowledge in the presence of the unhealthy element such as pornographic, violence, bloody
It Xun Lian not material progress shielding processing.It is noted that the mode that user provides recognition training material can be by user
It passes, can also be selected in the material database provided by background server as user, and the knowledge for being selected in material database
Not Xun Lian material, background server is without arranging recognition training material.
Step 104, based on the recognition training material of target game role, to target game role in a manner of deep learning
Things identification model carry out model training.
In an implementation, after background server gets the recognition training material of the target game role of user's offer every time,
It can be based on recognition training material, model instruction is carried out to the things identification model of target game role in a manner of deep learning
Practice, i.e. the recognition trainings material such as the every upload primary video of user, picture or audio, background server all can be to things identification model
Re -training is primary, to obtain new things identification model.It is appreciated that if there are multiple game angles in user account
Color, user is when providing the recognition training material of target game role, it is possible to specify the recognition training whether provided using this
The other game roles of material culture.
Optionally, the things recognition capability trained with things identification model, which can include at least, determines things mark and mould
Things movement or two kinds of posture are imitated, correspondingly, the processing of step 104 can be as follows respectively:
First, when things recognition capability is to determine things mark: determining recognition training material in a manner of deep learning
In things recognize feature；Feature training objective game angle is recognized according to the things mark and things that mark in recognition training material
The things identification model of color, so that things identification model determines the things mark of any things.
In an implementation, background server can determine knowledge after getting recognition training material in a manner of deep learning
Not Xun Lian the things in material recognize feature, may then pass through the things mark marked in recognition training material and determine
Things identification feature carrys out the things identification model of training objective game role, so that things identification model has the thing of any things
Object identifies capability really.For example, recognition training material be auto graph, it is determined that things identification feature can be
Then things mark " automobile " and " tire ", " steering wheel ", " license plate " etc. can be used in " tire ", " steering wheel ", " license plate " etc.
Things recognizes feature training things identification model.
Second, when things recognition capability is to imitate things movement or posture: determining identification instruction in a manner of deep learning
The things movement and/or things posture and things practiced in material act corresponding limbs displacement and/or the corresponding limbs of posture
Position；According to limbs displacement and/or the things identification model of position training objective game role, so that things identification model
Imitate things movement and/or things posture.
In an implementation, background server can determine knowledge after getting recognition training material in a manner of deep learning
Not Xun Lian things movement and/or things posture in material, and record that things acts corresponding limbs displacement and/or posture is corresponding
Position.Later, limbs displacement can be used in the server day after tomorrow and position carrys out the thing of training objective game role
Object identification model, so that things identification model has the ability to model of things movement and things posture.For example, recognition training material
For handstand picture, it is determined that things posture be " handstands ", corresponding position can be " both hands contact to earth ", " double-legged
Towards day " etc., the positions training things such as things posture " handstand " corresponding " both hands contact to earth ", " double-legged towards day " then can be used
Object identification model.
It optionally, can be with the game role of other users after having carried out a degree of culture to game role
The competition of things recognition capability is carried out, corresponding 104 different things recognition capability of above-mentioned steps, background server can to the processing of competition
With as follows:
First, when things recognition capability is to determine things mark: knowing when receiving the things for target game role
Do not compare with start request when, random selection include things identification feature video to be identified, picture or audio；It will be to be identified
The things identification model of video, picture or audio input target game role, the things mark identified；According to identifying
Things identify corresponding with video to be identified, picture or audio things mark, determine that things identification is scored.
In an implementation, it after user reaches given extent to the culture of target game role, can open with other players'
Game role compares with the function of things identification, specifically, user can open " hall " sub-interface, clicks what things identification compared with
Option tab initiates matching request so as to trigger game terminal to background server.In turn, background server can be according to
Preset matching rule is the game role of other players of target game role match specified quantity.After successful match,
The user being fitted on can click competition and start key, and background server then can receive for each trip for participating in this competition
What the things identification of play role compared with starts to request.By taking target game role as an example, background server, which receives, above-mentioned to be started to ask
After asking, video to be identified, picture or the sound that random selection in material database includes things identification feature can be compared with from preset
Frequently.Next, the things of video to be identified, picture or audio input target game role can be identified mould by background server
Type, so as to obtain the things mark that things identification model identifies.In turn, background server may determine that the thing identified
Object identifies things corresponding with video to be identified, picture or audio and identifies whether consistent degree, and determines therefrom that things identification is commented
Point.In this way, background server scores for the available things identification of each game role, it then can be according to things
The height of identification scoring determines to compare with as a result, can simultaneously give the game reward of corresponding different stage.
Second, imitating when starting request of competition when receiving the limbs for target game role, random selection includes
There are video to be imitated, picture or the audio of things movement and/or things posture；It will video, picture or audio input mesh be imitated
The things identification model for marking game role obtains echomotism and/or imitates posture；According to echomotism and/imitate posture, with
The similarity of things movement and/or things posture in video to be imitated, picture or audio, generates limbs and imitates scoring.
In an implementation, limbs imitate compare with start triggering processing with things identification compare with start triggering processing it is similar,
It does not exist together and is only that user clicks the option tab that the limbs imitation of " hall " sub-interface compares with.Equally with target game role
For, background server receive limbs imitate compare with start request after, can be selected at random from preset competition material database
Select include things movement and/or things posture video to be imitated, picture or audio.Next, background server can incite somebody to action
Video to be imitated, picture or the things identification model of audio input target game role, so as to obtain things identification model
The echomotism and/or imitation posture of generation.In turn, background server can determine echomotism and/imitate posture, and to mould
The similarity of things movement and/or things posture in imitative video, picture or audio, and determine therefrom that limbs imitate scoring.This
Sample, background server imitate scoring for the available limbs of each game role, can then imitate according to limbs
The height of scoring determines to compare with as a result, can simultaneously give the game reward of corresponding different stage.
Optionally, user can be by providing the things recognition capability for identifying test material to detect game role, accordingly
Processing can be such that obtain user provide target game role identification test material, the thing of invocation target game role
Object identification model identifies identification test material, and to user feedback recognition result.
Wherein, identification test material can be video, picture or the audio for being not marked with things mark.
In an implementation, user can choose identification in creation target game role and after having carried out certain culture to it
Material is tested, and provides it to background server, to be used to test the things recognition capability of target game role.From
And background server can call newest instruction after getting the identification test material of the target game role of user's offer
The things identification model of the target game role got identifies identification test material, and recognition result is fed back to use
Family.It should be noted that user uploads preceding background server after uploading to the processing and user that identify test material to identification
The content that above-mentioned user uploads recognition training material can be referred to by testing arranging for material, and difference be only that user is not necessarily to for knowledge
It Ce Shi not material label things mark.Similarly, the mode that user provides recognition training material can be and be uploaded by user, can also be with
It is selected in the material database provided by background server as user, and the test element of the identification for being selected in material database
Material, background server are not necessarily to arrange identification test material.
Optionally, during user cultivates game role, background server can be automatically to different role type
Game role carries out difference culture, and corresponding processing, which can be such that, periodically obtains role class from preset special material database
The corresponding role characteristics training material of type information, role characteristics training material are exclusively used in training different role characteristics；Based on angle
Color specifit training material carries out model training to the things identification model of target game role in a manner of deep learning.
In an implementation, the developer of develop class game can according to imaginary different role type to identical things not
Special material database is created to the familiarities of different things with cognition and different role type, it can be in the special material database
Multiclass role characteristics training material, every corresponding character types of class role characteristics training material, angle are stored with by character types
Color specifit training material can be exclusively used in training different role characteristics.Such as the role characteristics instruction under character types " mouse "
Practice material in, there may be marked with " natural enemy " " " picture of cat ".In this way, user is after creating target game role, after
Platform server can periodically obtain the corresponding role characteristics training material of character types information from special material database, be based on
Role characteristics train material, carry out model training to the things identification model of target game role in a manner of deep learning, with
The target game role role characteristics for meeting character types are assigned, the diversity and interest of game are improved.All trips herein
The period that play role trains automatically is all the same, on the one hand can simplify the execution logic of background server, saves background server
Process resource, on the other hand can to cultivate the higher game role of frequency is influenced smaller by role characteristics, is had more
It for accurately things recognition capability, is more cultivated diligently to reach, game role is more clever, and culture is fewer, and game role is easier
The game effect influenced by instinct ensure that the interest and fairness of game to a certain extent.
Optionally, the recognition training material that user uploads can share to other users, correspondingly, the place of background server
Reason can be such that when receiving acquisition request of the other users for the recognition training material of target game role, judge it
Whether its user meets the shared condition of recognition training material；If it is satisfied, then recognition training material is based on, with deep learning
The things identification model for the game role that mode specifies other users carries out model training.
In an implementation, user can be set after uploading recognition training material and whether share the recognition training material, such as
Fruit be shared condition accordingly then can be further set, such as pay certain money, game role reaches certain grade.This
Sample, background server can summarize all sharable recognition training materials, and creation is " shared in game main interface
Material " sub-interface.User can open " shared material " sub-interface, consult the recognition training material of other user sharings and corresponding
Shared condition.By taking the recognition training material of target game role as an example, if other users are wanted to obtain recognition training element
Material can then click the corresponding acquisition key of the recognition training material, thus background server in " shared material " sub-interface
Receive corresponding acquisition request.Later, background server may determine that whether the user meets the shared of recognition training material
Condition, if it is satisfied, then recognition training material, the thing for the game role specified in a manner of deep learning to user can be based on
Object identification model carries out model training, and certainly, user needs to have the administration authority of its specified game role.
In the embodiment of the present invention, the request to create for carrying the target game role of character types information is received, by role
The corresponding things of type information identifies initial model, is determined as the things identification model of target game role, obtains user and provides
Target game role recognition training material, recognition training material include at least be marked with things mark video, picture or
Audio, the recognition training material based on target game role identify the things of target game role in a manner of deep learning
Model carries out model training.In this way, by deep learning in conjunction with develop class game, it is excellent by the continuous personalized culture of user
Change the things identification model of game role in develop class game, so that game role can have different degrees of, different-style thing
Object recognition capability, so as to improve the playability and interest of develop class game.
Based on the same technical idea, the embodiment of the invention also provides a kind of develop class game angle based on machine learning
The culture apparatus of color, as shown in figure 3, described device includes
Receiving module 301, for receiving the request to create for carrying the target game role of character types information；
Determining module 302 is determined as the mesh for the corresponding things of the character types information to be identified initial model
Mark the things identification model of game role；
Module 303 is obtained, the recognition training material of the target game role for obtaining user's offer, the identification
Training material includes at least video, picture or the audio for being marked with things mark；
Training module 304, for the recognition training material based on the target game role, in a manner of deep learning pair
The things identification model of the target game role carries out model training.
Optionally, the training module 304, is specifically used for:
The things identification feature in the recognition training material is determined in a manner of deep learning；
According to the things mark and the things identification feature training target trip marked in the recognition training material
The things identification model of play role, so that the things identification model determines the things mark of any things.
Optionally, described device further include:
Selecting module, for when receive for the target game role things identification compare with start request when,
Random selection includes video to be identified, picture or the audio of things identification feature；
Identification module, for knowing the things of target game role described in the video to be identified, picture or audio input
Other model, the things mark identified；
Grading module, the things mark for identifying according to are corresponding with the video to be identified, picture or audio
Things mark, determine things identification scoring.
Optionally, the training module 304, is specifically used for:
Things movement and/or the things posture in the recognition training material, Yi Jisuo are determined in a manner of deep learning
It states things and acts corresponding limbs displacement and/or the corresponding position of the posture；
The things identification model of the target game role is trained according to limbs displacement and/or the position,
So that the things identification model imitates the things movement and/or the things posture.
Optionally, described device further include:
Selecting module, for when receive for the target game role limbs imitate compare with start request when,
Random selection includes video to be imitated, picture or the audio of things movement and/or things posture；
Identification module, for knowing the things of target game role described in the video to be imitated, picture or audio input
Other model obtains echomotism and/or imitates posture；
Grading module is used for according to the echomotism and/imitation posture, with video, picture or the audio to be imitated
In things movement and/or things posture similarity, generate limbs imitate scoring.
Optionally, described device further include:
Test module tests material for obtaining the identification of the target game role of user's offer, calls the mesh
The things identification model for marking game role is tested material to the identification and is identified, and to the user feedback recognition result.
Optionally, the training module 304, is also used to:
The corresponding role characteristics training material of the character types information is periodically obtained from preset special material database,
The role characteristics training material is exclusively used in training different role characteristics；
Based on role characteristics training material, the things of the target game role is identified in a manner of deep learning
Model carries out model training.
Optionally, the training module 304, is also used to:
When receiving acquisition request of the other users for the recognition training material of the target game role, institute is judged
State the shared condition whether other users meet the recognition training material；
If it is satisfied, then being based on the recognition training material, the other users are specified in a manner of deep learning
The things identification model of game role carries out model training.
In the embodiment of the present invention, the request to create for carrying the target game role of character types information is received, by role
The corresponding things of type information identifies initial model, is determined as the things identification model of target game role, obtains user and provides
Target game role recognition training material, recognition training material include at least be marked with things mark video, picture or
Audio, the recognition training material based on target game role identify the things of target game role in a manner of deep learning
Model carries out model training.In this way, by deep learning in conjunction with develop class game, it is excellent by the continuous personalized culture of user
Change the things identification model of game role in develop class game, so that game role can have different degrees of, different-style thing
Object recognition capability, so as to improve the playability and interest of develop class game.
It should be understood that the culture apparatus of the develop class game role provided by the above embodiment based on machine learning exists
When based on machine learning culture develop class game role, only the example of the division of the above functional modules, actually answer
In, it can according to need and be completed by different functional modules above-mentioned function distribution, i.e., divide the internal structure of device
At different functional modules, to complete all or part of the functions described above.In addition, provided by the above embodiment be based on machine
The culture apparatus of the develop class game role of device study and the cultural method of the develop class game role based on machine learning are implemented
Example belongs to same design, and specific implementation process is detailed in embodiment of the method, and which is not described herein again.
Fig. 4 is the structural schematic diagram of background server provided in an embodiment of the present invention.The background server 400 can be because of configuration
Or performance is different and generate bigger difference, may include one or more central processing units 422 (for example, one or
More than one processor) and memory 432, the storage medium of one or more storage application programs 442 or data 444
430 (such as one or more mass memory units).Wherein, memory 432 and storage medium 430 can be of short duration storage
Or persistent storage.The program for being stored in storage medium 430 may include one or more modules (diagram does not mark), each
Module may include to the series of instructions operation in background server.Further, central processing unit 422 can be set to
It is communicated with storage medium 430, the series of instructions operation in storage medium 430 is executed on background server 400.
Background server 400 can also include one or more power supplys 426, one or more are wired or wireless
Network interface 450, one or more input/output interfaces 458, one or more keyboards 456, and/or, one or
More than one operating system 441, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD etc..
Background server 400 may include have memory and one perhaps more than one program one of them or
More than one program is stored in memory, and be configured to be executed by one or more than one processor it is one or
More than one program includes to instruct for carrying out the culture of the above-mentioned develop class game role based on machine learning.
Those of ordinary skill in the art will appreciate that realizing that all or part of the steps of above-described embodiment can pass through hardware
It completes, relevant hardware can also be instructed to complete by program, the program can store in a kind of computer-readable
In storage medium, storage medium mentioned above can be read-only memory, disk or CD etc..
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in spirit of the invention and
Within principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (18)
1. a kind of cultural method of the develop class game role based on machine learning, which is characterized in that the described method includes:
Receive the request to create for carrying the target game role of character types information；
The corresponding things of the character types information is identified into initial model, is determined as the things identification of the target game role
Model；
The recognition training material for the target game role that user provides is obtained, the recognition training material includes at least label
Video, picture or the audio for thering is things to identify；
Based on the recognition training material of the target game role, to the thing of the target game role in a manner of deep learning
Object identification model carries out model training.
2. the method according to claim 1, wherein the recognition training element based on the target game role
Material carries out model training to the things identification model of the target game role in a manner of deep learning, comprising:
The things identification feature in the recognition training material is determined in a manner of deep learning；
According to the things mark and the things identification feature training target game angle marked in the recognition training material
The things identification model of color, so that the things identification model determines the things mark of any things.
3. according to the method described in claim 2, it is characterized in that, the method also includes:
When receiving when the starting request of things identification competition for the target game role, random selection includes things
Recognize video to be identified, picture or the audio of feature；
By the things identification model of target game role described in the video to be identified, picture or audio input, identified
Things mark；
The things mark identified according to described and the video to be identified, picture or the corresponding things of audio identify, and determine thing
Object identification scoring.
4. the method according to claim 1, wherein the recognition training element based on the target game role
Material carries out model training to the things identification model of the target game role in a manner of deep learning, comprising:
Things movement and/or things posture and the thing in the recognition training material are determined in a manner of deep learning
Object acts corresponding limbs displacement and/or the corresponding position of the posture；
According to limbs displacement and/or the things identification model of the position training target game role, so that
The things identification model imitates the things movement and/or the things posture.
5. according to the method described in claim 4, it is characterized in that, the method also includes:
When starting request of competition is imitated when receiving the limbs for the target game role, random selection includes things
Video to be imitated, picture or the audio of movement and/or things posture；
By the things identification model of target game role described in the video to be imitated, picture or audio input, obtain imitating dynamic
Make and/or imitate posture；
According to the echomotism and/imitate posture, in the video to be imitated, picture or audio things movement and/or
The similarity of things posture generates limbs and imitates scoring.
6. the method according to claim 1, wherein the method also includes:
Material is tested in the identification for obtaining the target game role that user provides, and the things of the target game role is called to know
Other model identifies identification test material, and to the user feedback recognition result.
7. the method according to claim 1, wherein the method also includes:
The corresponding role characteristics training material of the character types information is periodically obtained from preset special material database, it is described
Role characteristics training material is exclusively used in training different role characteristics；
Based on role characteristics training material, to the things identification model of the target game role in a manner of deep learning
Carry out model training.
8. the method according to claim 1, wherein the method also includes:
When receiving acquisition request of the other users for the recognition training material of the target game role, judgement it is described its
Whether its user meets the shared condition of the recognition training material；
If it is satisfied, then it is based on the recognition training material, the game that the other users are specified in a manner of deep learning
The things identification model of role carries out model training.
9. a kind of culture apparatus of the develop class game role based on machine learning, which is characterized in that described device includes:
Receiving module, for receiving the request to create for carrying the target game role of character types information；
Determining module is determined as the target game for the corresponding things of the character types information to be identified initial model
The things identification model of role；
Module is obtained, the recognition training material of the target game role for obtaining user's offer, the recognition training element
Material includes at least video, picture or the audio for being marked with things mark；
Training module, for the recognition training material based on the target game role, to the mesh in a manner of deep learning
The things identification model for marking game role carries out model training.
10. device according to claim 9, which is characterized in that the training module is specifically used for:
The things identification feature in the recognition training material is determined in a manner of deep learning；
According to the things mark and the things identification feature training target game angle marked in the recognition training material
The things identification model of color, so that the things identification model determines the things mark of any things.
11. device according to claim 10, which is characterized in that described device further include:
Selecting module, for when receive for the target game role things identification compare with start request when, at random
Selection includes video to be identified, picture or the audio of things identification feature；
Identification module, for the things of target game role described in the video to be identified, picture or audio input to be identified mould
Type, the things mark identified；
Grading module, the things mark thing corresponding with the video to be identified, picture or audio for being identified according to
Object mark determines things identification scoring.
12. device according to claim 9, which is characterized in that the training module is specifically used for:
Things movement and/or things posture and the thing in the recognition training material are determined in a manner of deep learning
Object acts corresponding limbs displacement and/or the corresponding position of the posture；
According to limbs displacement and/or the things identification model of the position training target game role, so that
The things identification model imitates the things movement and/or the things posture.
13. device according to claim 12, which is characterized in that described device further include:
Selecting module, for when receive for the target game role limbs imitate compare with start request when, at random
Selection includes video to be imitated, picture or the audio of things movement and/or things posture；
Identification module, for the things of target game role described in the video to be imitated, picture or audio input to be identified mould
Type obtains echomotism and/or imitates posture；
Grading module, for according to the echomotism and/imitate posture, in the video to be imitated, picture or audio
The similarity of things movement and/or things posture generates limbs and imitates scoring.
14. device according to claim 9, which is characterized in that described device further include:
Test module tests material for obtaining the identification of the target game role of user's offer, the target is called to swim
The things identification model of role of playing is tested material to the identification and is identified, and to the user feedback recognition result.
15. device according to claim 9, which is characterized in that the training module is also used to:
The corresponding role characteristics training material of the character types information is periodically obtained from preset special material database, it is described
Role characteristics training material is exclusively used in training different role characteristics；
Based on role characteristics training material, to the things identification model of the target game role in a manner of deep learning
Carry out model training.
16. device according to claim 9, which is characterized in that the training module is also used to:
When receiving acquisition request of the other users for the recognition training material of the target game role, judgement it is described its
Whether its user meets the shared condition of the recognition training material；
If it is satisfied, then it is based on the recognition training material, the game that the other users are specified in a manner of deep learning
The things identification model of role carries out model training.
17. a kind of background server, which is characterized in that the background server includes processor and memory, the memory
In be stored at least one instruction, at least one section of program, code set or instruction set, at least one instruction, described at least one
Duan Chengxu, the code set or instruction set are loaded by the processor and are executed as described in any of the claims 1 to 8 to realize
The cultural method of develop class game role based on machine learning.
18. a kind of computer readable storage medium, which is characterized in that be stored at least one instruction, extremely in the storage medium
Few one section of program, code set or instruction set, at least one instruction, at least one section of program, the code set or the instruction
Collection is loaded by processor and is executed to realize the develop class game angle as described in any of the claims 1 to 8 based on machine learning
The cultural method of color.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811003307.1A CN109157842A (en) | 2018-08-30 | 2018-08-30 | A kind of cultural method and device of the develop class game role based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811003307.1A CN109157842A (en) | 2018-08-30 | 2018-08-30 | A kind of cultural method and device of the develop class game role based on machine learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109157842A true CN109157842A (en) | 2019-01-08 |
Family
ID=64893371
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811003307.1A Pending CN109157842A (en) | 2018-08-30 | 2018-08-30 | A kind of cultural method and device of the develop class game role based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109157842A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109794065A (en) * | 2019-01-28 | 2019-05-24 | 网易（杭州）网络有限公司 | Game role creation method, device, electronic equipment and storage medium |
| CN109908578A (en) * | 2019-01-28 | 2019-06-21 | 努比亚技术有限公司 | A kind of game seismaesthesia control method, terminal and computer readable storage medium |
| CN111860852A (en) * | 2019-04-30 | 2020-10-30 | 百度时代网络技术（北京）有限公司 | Method, device and system for processing data |
| CN113707274A (en) * | 2021-02-10 | 2021-11-26 | 上海市同济医院 | Cognitive ability detection training system and device based on customized link memory game |
| CN115317896A (en) * | 2022-06-20 | 2022-11-11 | 赤子城网络技术（北京）有限公司 | Game picture making method and related equipment |
| CN117034385A (en) * | 2023-08-30 | 2023-11-10 | 四开花园网络科技(广州)有限公司 | AI system supporting creative design of humanoid roles |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107030691A (en) * | 2017-03-24 | 2017-08-11 | 华为技术有限公司 | A kind of data processing method and device for nursing robot |
-
2018
- 2018-08-30 CN CN201811003307.1A patent/CN109157842A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107030691A (en) * | 2017-03-24 | 2017-08-11 | 华为技术有限公司 | A kind of data processing method and device for nursing robot |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109794065A (en) * | 2019-01-28 | 2019-05-24 | 网易（杭州）网络有限公司 | Game role creation method, device, electronic equipment and storage medium |
| CN109908578A (en) * | 2019-01-28 | 2019-06-21 | 努比亚技术有限公司 | A kind of game seismaesthesia control method, terminal and computer readable storage medium |
| CN109794065B (en) * | 2019-01-28 | 2022-02-11 | 网易（杭州）网络有限公司 | Game role creating method and device, electronic equipment and storage medium |
| CN109908578B (en) * | 2019-01-28 | 2022-07-05 | 努比亚技术有限公司 | Game vibration sense control method, terminal and computer readable storage medium |
| CN111860852A (en) * | 2019-04-30 | 2020-10-30 | 百度时代网络技术（北京）有限公司 | Method, device and system for processing data |
| CN113707274A (en) * | 2021-02-10 | 2021-11-26 | 上海市同济医院 | Cognitive ability detection training system and device based on customized link memory game |
| CN113707274B (en) * | 2021-02-10 | 2024-04-16 | 上海市同济医院 | Cognitive ability detection training system and device based on customized linked memory game |
| CN115317896A (en) * | 2022-06-20 | 2022-11-11 | 赤子城网络技术（北京）有限公司 | Game picture making method and related equipment |
| CN117034385A (en) * | 2023-08-30 | 2023-11-10 | 四开花园网络科技(广州)有限公司 | AI system supporting creative design of humanoid roles |
| CN117034385B (en) * | 2023-08-30 | 2024-04-02 | 四开花园网络科技(广州)有限公司 | AI system supporting creative design of humanoid roles |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109157842A (en) | A kind of cultural method and device of the develop class game role based on machine learning | |
| JP6366128B2 (en) | program | |
| US20190344162A1 (en) | Systems and methods for gamification of a problem | |
| CN109144610B (en) | Audio playing method and device, electronic device and computer readable storage medium | |
| Lacasa | Learning in real and virtual worlds: Commercial video games as educational tools | |
| CN113660987A (en) | Generating gaming robots adapted for gaming applications | |
| CN106489259A (en) | Target is reached online quick question-and-answer game and provides method and system | |
| CN112001990B (en) | Scene-based data processing method and device, storage medium and electronic device | |
| Battistella et al. | Design and large-scale evaluation of educational games for teaching sorting algorithms | |
| Guss et al. | The MineRL 2019 competition on sample efficient reinforcement learning using human priors | |
| Stefanidis et al. | Learning prosocial skills through multiadaptive games: A case study | |
| Sezgin | Digital player typologies in gamification and game-based learning: A meta-synthesis | |
| Canning et al. | The culture of deep learning in esports: An insider's perspective | |
| KR102342778B1 (en) | Golf simulation device providing personalized avatar for user and operating method thereof | |
| Savitri et al. | Augmented reality application for science education on animal classification | |
| CN115944921B (en) | Game data processing method, device, equipment and medium | |
| Maxwell | Put Your Dream to the Test: 10 Questions that Will Help You See it and Seize it | |
| KR20230129779A (en) | User targeting method for game promotion and user targeting system using the same | |
| Kazai et al. | Ecir 2015 workshop on gamification for information retrieval (gamifir'15) | |
| Machado | Aplpication developmente over IoT platform Thingworx | |
| Ikeda et al. | Machine-learning of shape names for the game of Go | |
| US20230233944A1 (en) | Method of creating interaction-based cooperative agent, method of providing cooperative agent, and agent management server for performing methods | |
| Dingli et al. | Gamifying history using beacons and mobile technologies | |
| Cain | Tingle-Topic-independent Gamification Learning Environment | |
| Φραγκιάς | Video game character learning with artificial intelligent algorithms |
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: 20190108 |
|
| RJ01 | Rejection of invention patent application after publication |