CN107291232A - A kind of somatic sensation television game exchange method and system based on deep learning and big data - Google Patents
A kind of somatic sensation television game exchange method and system based on deep learning and big data Download PDFInfo
- Publication number
- CN107291232A CN107291232A CN201710471290.1A CN201710471290A CN107291232A CN 107291232 A CN107291232 A CN 107291232A CN 201710471290 A CN201710471290 A CN 201710471290A CN 107291232 A CN107291232 A CN 107291232A
- Authority
- CN
- China
- Prior art keywords
- neural networks
- convolutional neural
- video
- player
- game
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
Abstract
The invention discloses a kind of somatic sensation television game exchange method and system based on deep learning and big data, belong to human-computer interaction technique field.Method includes：Step one collects action video sample data set；Step 2 is set up and off-line training depth convolutional neural networks model；Step 3 is used depth convolutional neural networks model.System includes：Depth convolutional neural networks off-line training module, real time human-machine interaction module, the depth network model on-line optimization module based on big data.The present invention gathers game player's game operation video in real time using common camera, high-level semantics features are acted by the extraction of depth convolutional neural networks, limb action is differentiated, and actual control data to Mission Objective is converted to, action corresponding with player's motion is made by player's limb control Mission Objective so as to realize.
Description
Technical field
The invention discloses a kind of somatic sensation television game exchange method and system based on deep learning and big data, belong to man-machine
Interaction technique field.
Background technology
Somatic sensation television game is that the novel electron that a kind of limb action by player changes to operate and experience is played.Compared to
For electronic game system of the equipment such as traditional dependence mouse-keyboard, game paddle as interaction medium, somatic sensation television game passes through
Identification player " limb action " is used as interactive mode.Conditional electronic game requires that player is seated at before game station for a long time, is unfavorable for
Physical and mental health, and in somatic sensation television game, player sends all instructions by body to game, body is waved together with game not to be had
Any constraint, this is greatly enriched the game feeling of immersion of player, gives player splendid body-sensing experience, and in game joy
It can also be taken exercises while happy.
At present, somatic sensation television game has implementations below：1st, the XBOX 360 of Microsoft is adopted by using kinect 3D video cameras
Collect the limbs bone information of player to recognize player's limb action, its recognition accuracy is high but equipment is expensive；2nd, by wearing or
Hand-held sensor gathers the limbs information of player, so as to recognize player's limb action, the realization requires that player wears or hand
Sensor is held, this is likely to result in the discomfort of player, influence player experience.Sum it up, these somatic sensation television games are to hardware device
It is required that strict, configuration surroundings are complicated, and games cost is high, is unfavorable for the popularization and popularization of somatic sensation television game.
The content of the invention
In view of the defect of existing somatic sensation television game, it is proposed that a kind of somatic sensation television game side of interaction based on deep learning with big data
Method and system.Gather game player's game operation video in real time using common camera, extracted by depth convolutional neural networks
High-level semantics features are acted, limb action is differentiated, and are converted to the actual control data to Mission Objective, so as to realize
Action corresponding with player's motion is made by player's limb control Mission Objective.
The invention provides following scheme：
A kind of somatic sensation television game man-machine interaction method based on deep learning and big data, methods described includes：
Step one collects action video sample data set
Limb action video segment during different players is collected in advance, obtains action video sample data set, and be
The corresponding action label of sample data addition, action label is corresponded with Mission Objective control instruction；
Step 2 is set up and off-line training depth convolutional neural networks model
Deep learning training is carried out to the video sample data set that step one is obtained, depth convolutional neural networks model is set up
And off-line training is carried out, by error back propagation during off-line training, and network weight parameter is updated using stochastic gradient descent method,
The final loss function for making network reaches a minimum value, when the depth convolutional neural networks model is used to input player
Limb action video, output player actions predict the outcome；The player actions predict the outcome including the classification of motion and action point
The probability distribution data of class；
Step 3 is used depth convolutional neural networks model
Depth convolutional neural networks Model Fusion is entered in somatic sensation television game interactive system, passes through depth convolutional neural networks point
Real-time limb action during player is analysed, and is converted to the actual control data to Mission Objective.
Further, the action video sample database includes tranining database, validation database, test database,
Size is 70%：10%：20%.
Further, to the off-line training of depth convolutional neural networks, specifically include：
It is determined that for network model train different players when limb action video, to the size normalizing of frame of video
Change, facilitate follow-up analysis and extraction of features；
Action video is done into repetitive exercise in convolutional neural networks model, and using boarding steps during repetitive exercise
Descent method adjustment network weight parameter is spent, until the accuracy rate continuous one for reaching maximum iteration or being detected on checking collection
Fix time and not have dropped deconditioning again, finally give the convolution that real-time limb action during for player is analyzed
Neural network model.
Further, depth convolutional neural networks model include convolutional layer, down-sampling layer, full articulamentum, softmax layers,
The convolutional layer includes five layers of convolutional layer, first and third, five layers of convolutional layer be respectively provided with a down-sampling layer (max-pooling layers),
Down-sampling layer is used to carry out down-sampling to the characteristic pattern obtained after convolution operation processing, realizes Feature Dimension Reduction；Convolutional layer and complete
The activation primitive of articulamentum is using linear unit R eLU is corrected, and this can improve the convergence rate of model.
Further, methods described also includes：
Depth network model on-line optimization of the step 4 based on big data
It is invested in by the somatic game system after market, the actual game operation of player is collected using big data platform
Limb action video；
After being pre-processed to video, Cloud Server is sent to, the sample of extensive tape label is built in Cloud Server
Database；
The depth convolutional neural networks mould periodically obtained using the action video sample database to the off-line training
Type carries out on-line fine, further improves the recognition accuracy of network；
Periodically the network model after the on-line fine is updated in somatic sensation television game interactive system, allows game player to obtain
More preferable body-sensing experience.
Further, video is pre-processed, including rejects unrelated with game player's limbs operation content in video
Frame, and video size is normalized.
Further, video is sent to before Cloud Server, the redundancy of video data is reduced using video compression technology
Afterwards, then by the video after compression store into Cloud Server, so as to build the sample number of extensive tape label in Cloud Server
According to storehouse.
The structure of one high accuracy depth convolutional neural networks model needs huge training dataset, training dataset
Too small to be easily caused network over-fitting, the generalization ability of network is poor.Therefore, using the sample data of above-mentioned extensive tape label
Storehouse is finely adjusted to the depth network model of above-mentioned pre-training can further lift the precision of network.
A kind of somatic sensation television game man-machine interactive system based on deep learning and big data, the system includes：
Depth convolutional neural networks off-line training module, the training that limb action video during for being played by player is constituted
Sample data set carries out deep learning training, sets up depth convolutional neural networks.By error back propagation during training, and using with
Machine gradient descent method updates network weight parameter, the loss function of network is reached a minimum value.The depth convolution
Neutral net is used to input limb action video during player, and output gamer action predicts the outcome；The game is played
Family's action prediction result includes the classification of motion and its probability distribution data；
Real time human-machine interaction module, for when game player starts game, starting the common shooting that game host is carried
Head, the limb action that collection game player interacts with Mission Objective personage in real time, and it is single to be sent to the center processing of game host
Member, as the data input stream of above-mentioned depth convolutional neural networks model, after the analysis by depth network, obtains game and plays
Family's action classification, then the movement locus for instructing and controlling Mission Objective personage corresponding with the category is sent, and then more new game is drawn
Face, so as to realize the interactive process of somatic sensation television game.
Depth network model on-line optimization module based on big data, during for collecting player using big data platform
Limb action video, be sent to Cloud Server after being pre-processed to video, extensive tape label built in Cloud Server
Sample database；The depth convolutional neural networks model periodically obtained using the sample database to the off-line training is entered
Row on-line fine, and periodically the network model after the on-line fine is updated in somatic sensation television game interactive system, allow game to play
Family obtains more preferable body-sensing experience.
The beneficial effects of the present invention are：A kind of somatic sensation television game side of interaction based on deep learning with big data of the present invention
Method and system, compared with existing somatic sensation television game implementation, the present invention has advantages below：
First, not only easy to operate due to the present invention only with common camera as data sampling sensor, Er Qieji
The earth reduces cost, is conducive to promotion and popularization；
Second, because the present invention uses depth convolutional neural networks model, thus it need not manually go to choose description object for appreciation
The feature of family's action, but allow network oneself to remove learning characteristic by training so that the self-adapting estimation ability of depth network is non-
Chang Qiang；
3rd, because the present invention makes full use of the advantage of " big data ", game player's practical operation video is collected, and reach
Cloud Server, so as to build the sample database of extensive tape label in Cloud Server, solves depth convolutional Neural
Network because tranining database it is small caused by over-fitting problem, improve the generalization ability of depth network.Using game player certainly
The practical operation video of body carries out adaptive training to depth network model, and this not only further increasing the accurate of network
Rate, and the participation of game player is drastically increased, and bring more preferable body-sensing experience to game player.
Brief description of the drawings
Fig. 1 is a kind of somatic sensation television game exchange method schematic diagram based on deep learning provided in an embodiment of the present invention；
Fig. 2 is that a kind of deep learning network model method for on-line optimization based on big data provided in an embodiment of the present invention is shown
It is intended to；
Fig. 3 is the somatic sensation television game interactive system schematic diagram provided in an embodiment of the present invention based on deep learning and big data；
Fig. 4 is the principle of the somatic sensation television game interactive system provided in an embodiment of the present invention based on deep learning and big data
Figure.
Embodiment
Below, with reference to accompanying drawing and embodiment, the present invention is described further：
Embodiment 1
As shown in figure 1, the embodiments of the invention provide a kind of somatic sensation television game exchange method based on deep learning, including：
Step S101, collects limb action video during different game, obtains action video sample database, and be in advance
The corresponding action label of sample data addition, action label is corresponded with Mission Objective control instruction；
Step S102, carries out deep learning training to above-mentioned video sample data set, sets up depth convolutional neural networks, instructs
By error back propagation when practicing, and network weight parameter is updated using stochastic gradient descent method, finally make the loss function of network
Reach a minimum value.The depth convolutional neural networks are used to input limb action video during player, output game
Player actions predict the outcome；The gamer action predicts the outcome including the classification of motion and its probability distribution data；
Step S103, the depth convolutional neural networks Model Fusion is entered in somatic sensation television game interactive system, passes through depth
Convolutional neural networks analyze real-time limb action during player, and are converted to the actual control data to Mission Objective.
The embodiment of the present invention is not limited the mode of setting up of depth convolutional neural networks, on the basis of above-described embodiment
On, it is preferred that the step S102 can be specially：Limb action video during to above-mentioned player is trained, and is set up
Depth is M convolutional neural networks, and M is positive integer；The depth convolutional neural networks are also big to convolution kernel type, convolution kernel
The parameters such as small, pond mode, learning rate, iterations, loss function are configured；Affiliated depth is M convolutional Neural net
Network, including M layers of convolutional layer；1st layer is used to input the actual game operation video of game player, the spy that output first layer is proposed
Levy, i-th layer of convolutional layer is used for the feature for inputting the i-th -1 layer extraction, export the feature of i-th layer of extraction, i is the positive integer more than 1.
Final depth convolutional neural networks are constituted using multilayer convolutional layer, full articulamentum, diagnostic horizon.On this basis, it is of the invention
Embodiment is not limited the iterations of depth convolutional neural networks, it is preferred that network can be 5000 times with iteration.
The convolution kernel form that the embodiment of the present invention is used to convolutional layer is not limited, further, and convolution operation is actual
On be a related operation, formula can be defined as：
The pond mode that the embodiment of the present invention is used to down-sampling layer is not limited, it is preferred that take average pond, and it is counted
Calculating formula is：
The activation primitive that the embodiment of the present invention is used to network is not limited, it is preferred that using amendment linear unit
ReLU functions are defined as activation primitive：F (x)=max (0, x), the benefit so done be no saturation region and convergence is fast.
It is described by depth convolutional neural networks, in the step of carrying out analysis prediction to the actual game operation video of player, can adopt
Made decisions with Softmax function pair player's limb actions.Give a training set T={ (x(1),y(1)),...,(x(m),y(m)), Softmax graders will export the vector that a k is tieed up, and calculates function representation and is：
Wherein
It is model parameter.
The embodiment of the present invention is not limited game limb action classification, and game control command is not limited, it is preferred that
It can be designed according to specific scene of game.
The embodiment of the present invention is not limited the number of the game player selected by off-line training step, it is preferred that for certain
A kind of specific somatic sensation television game, can choose N number of people (N is positive integer), and everyone makes various possible for this game
Limb action, and the common camera carried using game host is acquired, and then player's limb action is labelled manually,
So as to form a number of training sample database.
Embodiment 2
As shown in Fig. 2 the embodiments of the invention provide a kind of deep learning network model on-line optimization based on big data
Method, including：
Step S201, limb action video during player is collected using big data platform；
Step S202, is pre-processed to video, is sent to Cloud Server, and action video sample is built in Cloud Server
Database；
Step S203, the depth convolutional Neural net periodically obtained using the video sample database to the off-line training
Network model carries out on-line fine, further improves the recognition accuracy of network；
Network model after the on-line fine, is periodically updated in somatic sensation television game interactive system by step S204, allows trip
The player that plays obtains more preferable body-sensing experience.
The embodiment of the present invention is pre-processed to video, including rejects unrelated with game player's operation content in video
Frame, and the yardstick of frame of video is normalized, it is preferred that video frame size is normalized to：256X340, obtains standard
The video flowing of change.Advantage of this is that reducing transmission information content, memory space is saved, and simplify follow-up evaluation work
Amount.
The embodiment of the present invention is not limited the mode that the network model obtained by off-line training is finely adjusted, it is preferred that examined
Consider for a somatic sensation television game, the limb action video of different game player's practical operations of later stage online acquisition with
Action video collection used in off-line training is similar in terms of content, therefore can use offline depth network model as initial
Change weight to be finely adjusted whole network, so as to further optimize depth network.
Embodiment 3
With reference to shown in Fig. 3 and Fig. 4, the embodiments of the invention provide a kind of somatic sensation television game based on deep learning and big data
Interactive system, including：
Depth convolutional neural networks off-line training module 301, is used for：
To by player when the training sample data collection that constitutes of limb action video carry out deep learning training, set up
Depth convolutional neural networks.By error back propagation during training, and network weight parameter is updated using stochastic gradient descent method, most
The loss function of network is set to reach a minimum value eventually.The depth convolutional neural networks are regarded for inputting gamer action
Frequently, output gamer action predicts the outcome；The gamer action predicts the outcome including the classification of motion and its probability distribution
Data；
Real time human-machine interaction module 302, is used for：
When game player starts game, start the common camera that carries of game host, in real time collection game player with
The limb action of Mission Objective personage interaction, and the CPU of game host is sent to, it is used as above-mentioned depth convolution god
Data input stream through network model, after the analysis by depth network, obtains gamer action classification, then send with being somebody's turn to do
Classification instructs the movement locus of control Mission Objective personage accordingly, and then updates game picture, so as to realize somatic sensation television game
Interactive process.
Depth network model on-line optimization module 303 based on big data, is used for：
Limb action video during player is collected using big data platform, cloud is sent to after being pre-processed to video
Server, builds the sample database of extensive tape label in Cloud Server；Periodically using the sample database to described
The depth convolutional neural networks model that off-line training is obtained carries out on-line fine；Periodically by the network model after the on-line fine
It is updated in somatic sensation television game interactive system.
It will be apparent to those skilled in the art that technical scheme that can be as described above and design, make other various
It is corresponding to change and deformation, and all these change and deformation should all belong to the protection domain of the claims in the present invention
Within.
Claims (8)
1. a kind of somatic sensation television game exchange method based on deep learning and big data, it is characterised in that methods described includes：
Step one collects action video sample data set
Limb action video segment during different players is collected in advance, obtains action video sample data set, and be sample
The corresponding action label of data addition, action label is corresponded with Mission Objective control instruction；
Step 2 is set up and off-line training depth convolutional neural networks model
Deep learning training is carried out to the video sample data set that step one is obtained, depth convolutional neural networks model is set up and goes forward side by side
Row off-line training, by error back propagation during off-line training, and updates network weight parameter, finally using stochastic gradient descent method
The loss function of network is set to reach a minimum value, the depth convolutional neural networks model is used to input limb during player
Body action video, output player actions predict the outcome；The player actions predict the outcome including the classification of motion and the classification of motion
Probability distribution data；
Step 3 is used depth convolutional neural networks model
Depth convolutional neural networks Model Fusion is entered in somatic sensation television game interactive system, is analyzed and played by depth convolutional neural networks
Real-time limb action during family's game, and be converted to the actual control data to Mission Objective.
2. the somatic sensation television game exchange method as claimed in claim 1 based on deep learning and big data, it is characterised in that described
Action video sample database includes tranining database, validation database, test database, and size is 70%：10%：
20%.
3. the somatic sensation television game exchange method as claimed in claim 1 based on deep learning and big data, it is characterised in that to depth
The off-line training of degree convolutional neural networks is specifically included：
It is determined that train for network model different players when limb action video, the size of frame of video is normalized；
Action video is done into repetitive exercise in convolutional neural networks model, and used during repetitive exercise under stochastic gradient
Drop method adjusts network weight parameter, until reaching maximum iteration or the accuracy rate continuous certain time detected on checking collection
Deconditioning when not declining, obtains the convolutional neural networks mould that real-time limb action during for player is analyzed
Type.
4. the somatic sensation television game exchange method as claimed in claim 1 based on deep learning and big data, it is characterised in that depth
Convolutional neural networks model include convolutional layer, down-sampling layer, full articulamentum, softmax layers, the convolutional layer includes five layers of convolution
Layer, first and third, five layers of convolutional layer be respectively provided with a down-sampling layer, down-sampling layer is used for the spy to being obtained after convolution operation processing
Levy figure and carry out down-sampling.
5. the somatic sensation television game exchange method as claimed in claim 1 based on deep learning and big data, it is characterised in that described
Method also includes：
Depth network model on-line optimization of the step 4 based on big data
The actual game operation limb action video of player is collected using big data platform；
After being pre-processed to video, Cloud Server is sent to, the sample database of tape label is built in Cloud Server；
The depth convolutional neural networks model periodically obtained using the action video sample database to the off-line training is entered
Row on-line fine；
Periodically the network model after the on-line fine is updated in somatic sensation television game interactive system.
6. the somatic sensation television game exchange method based on deep learning and big data as claimed in claim 5, it is characterised in that to regarding
Frequency is pre-processed, including rejects frame unrelated with game player's limbs operation content in video, and video size is normalized.
7. the somatic sensation television game exchange method as claimed in claim 5 based on deep learning and big data, it is characterised in that will regard
Keep pouring in before delivering to Cloud Server, after the redundancy that video data is reduced using video compression technology, then the video after compression deposited
Storage is into Cloud Server.
8. a kind of somatic sensation television game man-machine interactive system based on deep learning and big data, it is characterised in that the system includes：
Depth convolutional neural networks off-line training module, the training sample that limb action video during for being played by player is constituted
Data set carries out deep learning training, sets up depth convolutional neural networks, by error back propagation during training, and uses boarding steps
Spend descent method and update network weight parameter, the loss function of network is reached a minimum value, depth convolutional neural networks
For inputting limb action video during player, output gamer action predicts the outcome；Gamer action prediction knot
Fruit includes the classification of motion and its probability distribution data；
Real time human-machine interaction module, it is real for when game player starts game, starting the common camera that game host is carried
When the limb action that is interacted with Mission Objective personage of collection game player, and be sent to the CPU of game host, work
For the data input stream of depth convolutional neural networks model, after the analysis by depth network, gamer action class is obtained
Not, then send it is corresponding with the category instruct control Mission Objective personage movement locus；
Depth network model on-line optimization module based on big data, for collecting limb during player using big data platform
Body action video, Cloud Server is sent to after being pre-processed to video, and the sample of extensive tape label is built in Cloud Server
Database；The depth convolutional neural networks model periodically obtained using sample database to off-line training carries out on-line fine,
And periodically the network model after on-line fine is updated in somatic sensation television game interactive system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710471290.1A CN107291232A (en) | 2017-06-20 | 2017-06-20 | A kind of somatic sensation television game exchange method and system based on deep learning and big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710471290.1A CN107291232A (en) | 2017-06-20 | 2017-06-20 | A kind of somatic sensation television game exchange method and system based on deep learning and big data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107291232A true CN107291232A (en) | 2017-10-24 |
Family
ID=60096917
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710471290.1A Pending CN107291232A (en) | 2017-06-20 | 2017-06-20 | A kind of somatic sensation television game exchange method and system based on deep learning and big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107291232A (en) |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107818571A (en) * | 2017-12-11 | 2018-03-20 | 珠海大横琴科技发展有限公司 | Ship automatic tracking method and system based on deep learning network and average drifting |
| CN107909060A (en) * | 2017-12-05 | 2018-04-13 | 前海健匠智能科技（深圳）有限公司 | Gymnasium body-building action identification method and device based on deep learning |
| CN108057246A (en) * | 2017-11-08 | 2018-05-22 | 江苏名通信息科技有限公司 | Hand based on deep neural network study swims augmented reality method |
| CN108096833A (en) * | 2017-12-20 | 2018-06-01 | 北京奇虎科技有限公司 | Somatic sensation television game control method and device based on cascade neural network, computing device |
| CN108447487A (en) * | 2018-03-27 | 2018-08-24 | 中国科学院长春光学精密机械与物理研究所 | Method and system based on text input with output training simulation human brain thinking |
| CN108446649A (en) * | 2018-03-27 | 2018-08-24 | 百度在线网络技术（北京）有限公司 | Method and device for alarm |
| CN108491767A (en) * | 2018-03-06 | 2018-09-04 | 北京因时机器人科技有限公司 | Autonomous roll response method, system and manipulator based on Online Video perception |
| CN108537178A (en) * | 2018-04-12 | 2018-09-14 | 佘堃 | A kind of intelligent recognition and interactive method and system |
| CN108815845A (en) * | 2018-05-15 | 2018-11-16 | 百度在线网络技术（北京）有限公司 | The information processing method and device of human-computer interaction, computer equipment and readable medium |
| CN108830380A (en) * | 2018-04-11 | 2018-11-16 | 开放智能机器（上海）有限公司 | A kind of training pattern generation method and system based on cloud service |
| CN109558892A (en) * | 2018-10-30 | 2019-04-02 | 银河水滴科技（北京）有限公司 | A kind of target identification method neural network based and system |
| CN109718558A (en) * | 2017-10-31 | 2019-05-07 | 腾讯科技(成都)有限公司 | The determination method and apparatus of game information, storage medium, electronic device |
| CN109726808A (en) * | 2017-10-27 | 2019-05-07 | 腾讯科技（深圳）有限公司 | Neural network training method and device, storage medium and electronic device |
| CN109766795A (en) * | 2018-12-25 | 2019-05-17 | 同济大学 | Man-machine interaction method and system under virtual reality scenario based on Recognition with Recurrent Neural Network |
| CN109784133A (en) * | 2017-11-15 | 2019-05-21 | 财团法人资讯工业策进会 | Act evaluation model generating means and its movement evaluation model generating method |
| CN109821239A (en) * | 2019-02-20 | 2019-05-31 | 网易（杭州）网络有限公司 | Implementation method, device, equipment and the storage medium of somatic sensation television game |
| CN109922354A (en) * | 2019-03-29 | 2019-06-21 | 广州虎牙信息科技有限公司 | Living broadcast interactive method, apparatus, live broadcast system and electronic equipment |
| WO2019144346A1 (en) * | 2018-01-25 | 2019-08-01 | 腾讯科技（深圳）有限公司 | Object processing method in virtual scene, device and storage medium |
| CN110119815A (en) * | 2019-05-21 | 2019-08-13 | 深圳市腾讯网域计算机网络有限公司 | Model training method, device, storage medium and equipment |
| CN110119547A (en) * | 2019-04-28 | 2019-08-13 | 腾讯科技（深圳）有限公司 | A kind of prediction group defeats negative method, apparatus and control equipment |
| CN110334642A (en) * | 2019-07-01 | 2019-10-15 | 河南牧业经济学院 | The machine vision recognition method and system of one boar behavior |
| CN110427484A (en) * | 2018-04-26 | 2019-11-08 | 上海意仕腾教育科技有限公司 | A kind of Chinese natural language processing method based on deep learning |
| WO2020063758A1 (en) * | 2018-09-26 | 2020-04-02 | Guangdong Oppo Mobile Telecommunications Corp., Ltd. | Game Picture Display Method and Apparatus, Storage Medium and Electronic Device |
| CN111318009A (en) * | 2020-01-19 | 2020-06-23 | 张衡 | Somatosensory health entertainment system based on wearable inertial sensing and working method thereof |
| CN111325116A (en) * | 2020-02-05 | 2020-06-23 | 武汉大学 | Remote sensing image target detection method capable of evolving based on offline training-online learning depth |
| CN111506514A (en) * | 2020-07-01 | 2020-08-07 | 深圳云步互娱网络科技有限公司 | Intelligent testing method and system applied to elimination game |
| CN111553930A (en) * | 2020-05-08 | 2020-08-18 | 吴修文 | Online somatosensory self-adaptive interaction method combined with video intelligent analysis |
| CN113058258A (en) * | 2021-04-22 | 2021-07-02 | 杭州当贝网络科技有限公司 | Method, system and storage medium for recognizing expected somatosensory action based on player game |
| CN113952731A (en) * | 2021-12-21 | 2022-01-21 | 广州优刻谷科技有限公司 | Motion sensing game action recognition method and system based on multi-stage joint training |
| CN113058258B (en) * | 2021-04-22 | 2024-04-19 | 杭州当贝网络科技有限公司 | Method, system and storage medium for identifying motion based on expected motion of player game |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6334121B1 (en) * | 1998-05-04 | 2001-12-25 | Virginia Commonwealth University | Usage pattern based user authenticator |
| CN105718879A (en) * | 2016-01-19 | 2016-06-29 | 华南理工大学 | Free-scene egocentric-vision finger key point detection method based on depth convolution nerve network |
| CN105787439A (en) * | 2016-02-04 | 2016-07-20 | 广州新节奏智能科技有限公司 | Depth image human body joint positioning method based on convolution nerve network |
| CN106227341A (en) * | 2016-07-20 | 2016-12-14 | 南京邮电大学 | Unmanned plane gesture interaction method based on degree of depth study and system |
| CN106570474A (en) * | 2016-10-27 | 2017-04-19 | 南京邮电大学 | Micro expression recognition method based on 3D convolution neural network |
-
2017
- 2017-06-20 CN CN201710471290.1A patent/CN107291232A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6334121B1 (en) * | 1998-05-04 | 2001-12-25 | Virginia Commonwealth University | Usage pattern based user authenticator |
| CN105718879A (en) * | 2016-01-19 | 2016-06-29 | 华南理工大学 | Free-scene egocentric-vision finger key point detection method based on depth convolution nerve network |
| CN105787439A (en) * | 2016-02-04 | 2016-07-20 | 广州新节奏智能科技有限公司 | Depth image human body joint positioning method based on convolution nerve network |
| CN106227341A (en) * | 2016-07-20 | 2016-12-14 | 南京邮电大学 | Unmanned plane gesture interaction method based on degree of depth study and system |
| CN106570474A (en) * | 2016-10-27 | 2017-04-19 | 南京邮电大学 | Micro expression recognition method based on 3D convolution neural network |
Cited By (43)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109726808A (en) * | 2017-10-27 | 2019-05-07 | 腾讯科技（深圳）有限公司 | Neural network training method and device, storage medium and electronic device |
| CN109726808B (en) * | 2017-10-27 | 2022-12-09 | 腾讯科技（深圳）有限公司 | Neural network training method and device, storage medium and electronic device |
| CN109718558B (en) * | 2017-10-31 | 2021-05-11 | 腾讯科技(成都)有限公司 | Game information determination method and device, storage medium and electronic device |
| CN109718558A (en) * | 2017-10-31 | 2019-05-07 | 腾讯科技(成都)有限公司 | The determination method and apparatus of game information, storage medium, electronic device |
| CN108057246A (en) * | 2017-11-08 | 2018-05-22 | 江苏名通信息科技有限公司 | Hand based on deep neural network study swims augmented reality method |
| CN109784133A (en) * | 2017-11-15 | 2019-05-21 | 财团法人资讯工业策进会 | Act evaluation model generating means and its movement evaluation model generating method |
| CN109784133B (en) * | 2017-11-15 | 2021-11-02 | 财团法人资讯工业策进会 | Action evaluation model generation device and action evaluation model generation method thereof |
| CN107909060A (en) * | 2017-12-05 | 2018-04-13 | 前海健匠智能科技（深圳）有限公司 | Gymnasium body-building action identification method and device based on deep learning |
| CN107818571A (en) * | 2017-12-11 | 2018-03-20 | 珠海大横琴科技发展有限公司 | Ship automatic tracking method and system based on deep learning network and average drifting |
| CN107818571B (en) * | 2017-12-11 | 2018-07-20 | 珠海大横琴科技发展有限公司 | Ship automatic tracking method and system based on deep learning network and average drifting |
| US10706285B2 (en) | 2017-12-11 | 2020-07-07 | Zhuhai Da Hengqin Technology Development Co., Ltd. | Automatic ship tracking method and system based on deep learning network and mean shift |
| CN108096833A (en) * | 2017-12-20 | 2018-06-01 | 北京奇虎科技有限公司 | Somatic sensation television game control method and device based on cascade neural network, computing device |
| CN108096833B (en) * | 2017-12-20 | 2021-10-01 | 北京奇虎科技有限公司 | Motion sensing game control method and device based on cascade neural network and computing equipment |
| CN110325965A (en) * | 2018-01-25 | 2019-10-11 | 腾讯科技（深圳）有限公司 | Object processing method, equipment and storage medium in virtual scene |
| WO2019144346A1 (en) * | 2018-01-25 | 2019-08-01 | 腾讯科技（深圳）有限公司 | Object processing method in virtual scene, device and storage medium |
| CN108491767A (en) * | 2018-03-06 | 2018-09-04 | 北京因时机器人科技有限公司 | Autonomous roll response method, system and manipulator based on Online Video perception |
| CN108491767B (en) * | 2018-03-06 | 2022-08-09 | 北京因时机器人科技有限公司 | Autonomous rolling response method and system based on online video perception and manipulator |
| CN108447487A (en) * | 2018-03-27 | 2018-08-24 | 中国科学院长春光学精密机械与物理研究所 | Method and system based on text input with output training simulation human brain thinking |
| CN108446649A (en) * | 2018-03-27 | 2018-08-24 | 百度在线网络技术（北京）有限公司 | Method and device for alarm |
| CN108447487B (en) * | 2018-03-27 | 2020-09-25 | 长春市长光芯忆科技有限公司 | Method and system for training and simulating human brain thinking based on text input and output |
| CN108830380A (en) * | 2018-04-11 | 2018-11-16 | 开放智能机器（上海）有限公司 | A kind of training pattern generation method and system based on cloud service |
| CN108537178A (en) * | 2018-04-12 | 2018-09-14 | 佘堃 | A kind of intelligent recognition and interactive method and system |
| CN110427484A (en) * | 2018-04-26 | 2019-11-08 | 上海意仕腾教育科技有限公司 | A kind of Chinese natural language processing method based on deep learning |
| CN108815845A (en) * | 2018-05-15 | 2018-11-16 | 百度在线网络技术（北京）有限公司 | The information processing method and device of human-computer interaction, computer equipment and readable medium |
| CN108815845B (en) * | 2018-05-15 | 2019-11-26 | 百度在线网络技术（北京）有限公司 | The information processing method and device of human-computer interaction, computer equipment and readable medium |
| WO2020063758A1 (en) * | 2018-09-26 | 2020-04-02 | Guangdong Oppo Mobile Telecommunications Corp., Ltd. | Game Picture Display Method and Apparatus, Storage Medium and Electronic Device |
| CN109558892A (en) * | 2018-10-30 | 2019-04-02 | 银河水滴科技（北京）有限公司 | A kind of target identification method neural network based and system |
| CN109766795B (en) * | 2018-12-25 | 2023-06-30 | 吉林动画学院 | Man-machine interaction method and system based on cyclic neural network in virtual reality scene |
| CN109766795A (en) * | 2018-12-25 | 2019-05-17 | 同济大学 | Man-machine interaction method and system under virtual reality scenario based on Recognition with Recurrent Neural Network |
| CN109821239A (en) * | 2019-02-20 | 2019-05-31 | 网易（杭州）网络有限公司 | Implementation method, device, equipment and the storage medium of somatic sensation television game |
| CN109922354A (en) * | 2019-03-29 | 2019-06-21 | 广州虎牙信息科技有限公司 | Living broadcast interactive method, apparatus, live broadcast system and electronic equipment |
| CN110119547A (en) * | 2019-04-28 | 2019-08-13 | 腾讯科技（深圳）有限公司 | A kind of prediction group defeats negative method, apparatus and control equipment |
| CN110119547B (en) * | 2019-04-28 | 2021-07-30 | 腾讯科技（深圳）有限公司 | Method, device and control equipment for predicting group war victory or defeat |
| CN110119815A (en) * | 2019-05-21 | 2019-08-13 | 深圳市腾讯网域计算机网络有限公司 | Model training method, device, storage medium and equipment |
| CN110334642A (en) * | 2019-07-01 | 2019-10-15 | 河南牧业经济学院 | The machine vision recognition method and system of one boar behavior |
| CN111318009A (en) * | 2020-01-19 | 2020-06-23 | 张衡 | Somatosensory health entertainment system based on wearable inertial sensing and working method thereof |
| CN111325116A (en) * | 2020-02-05 | 2020-06-23 | 武汉大学 | Remote sensing image target detection method capable of evolving based on offline training-online learning depth |
| CN111553930A (en) * | 2020-05-08 | 2020-08-18 | 吴修文 | Online somatosensory self-adaptive interaction method combined with video intelligent analysis |
| CN111506514B (en) * | 2020-07-01 | 2020-10-27 | 深圳云步互娱网络科技有限公司 | Intelligent testing method and system applied to elimination game |
| CN111506514A (en) * | 2020-07-01 | 2020-08-07 | 深圳云步互娱网络科技有限公司 | Intelligent testing method and system applied to elimination game |
| CN113058258A (en) * | 2021-04-22 | 2021-07-02 | 杭州当贝网络科技有限公司 | Method, system and storage medium for recognizing expected somatosensory action based on player game |
| CN113058258B (en) * | 2021-04-22 | 2024-04-19 | 杭州当贝网络科技有限公司 | Method, system and storage medium for identifying motion based on expected motion of player game |
| CN113952731A (en) * | 2021-12-21 | 2022-01-21 | 广州优刻谷科技有限公司 | Motion sensing game action recognition method and system based on multi-stage joint training |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107291232A (en) | A kind of somatic sensation television game exchange method and system based on deep learning and big data | |
| CN108198601A (en) | Motion scores method, apparatus, equipment and storage medium | |
| WO2020156245A1 (en) | Action recognition method, apparatus and device and storage medium | |
| CN107423398A (en) | Exchange method, device, storage medium and computer equipment | |
| CN111414839B (en) | Emotion recognition method and device based on gesture | |
| CN110781765A (en) | Human body posture recognition method, device, equipment and storage medium | |
| CN109479110A (en) | The system and method that dynamic creation individualizes exercise videos | |
| CN107633317A (en) | Establish stroke planning model and cook up the method, apparatus of parade journey | |
| CN110503074A (en) | Information labeling method, apparatus, equipment and the storage medium of video frame | |
| US11551479B2 (en) | Motion behavior pattern classification method, system and device | |
| CN108905209A (en) | Plug-in detection method and system, electronic equipment and storage medium in game | |
| CN108921284A (en) | Interpersonal interactive body language automatic generation method and system based on deep learning | |
| CN111222486B (en) | Training method, device and equipment for hand gesture recognition model and storage medium | |
| CN106454108B (en) | Track up method, apparatus and electronic equipment based on artificial intelligence | |
| CN112422946B (en) | Intelligent yoga action guidance system based on 3D reconstruction | |
| CN109045664B (en) | Diving scoring method, server and system based on deep learning | |
| CN110175619A (en) | Method, equipment and storage medium are determined based on the board group of playing a card of machine learning model | |
| KR102498577B1 (en) | Method and apparatus for recommending customized golf clubs according to golf ball pitch, club head speed, and user information using artificial neural networks | |
| Chang et al. | Redirection controller using reinforcement learning | |
| CN109847366A (en) | Data for games treating method and apparatus | |
| CN111401219A (en) | Palm key point detection method and device | |
| CN111282272B (en) | Information processing method, computer readable medium and electronic device | |
| CN113947022B (en) | Near-end strategy optimization method based on model | |
| CA3087629C (en) | System for managing user experience and method therefor | |
| CN116630736A (en) | Training method and system for user expression capturing model |
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: 20171024 |
|
| RJ01 | Rejection of invention patent application after publication |