CN108762503A - A kind of man-machine interactive system based on multi-modal data acquisition - Google Patents
A kind of man-machine interactive system based on multi-modal data acquisition Download PDFInfo
- Publication number
- CN108762503A CN108762503A CN201810509996.7A CN201810509996A CN108762503A CN 108762503 A CN108762503 A CN 108762503A CN 201810509996 A CN201810509996 A CN 201810509996A CN 108762503 A CN108762503 A CN 108762503A
- Authority
- CN
- China
- Prior art keywords
- data
- sample
- classification
- training
- acquisition
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
Abstract
The present invention provides a kind of man-machine interactive system acquired based on multi-modal data, including:Acquisition module, for acquiring multi-modal data, the multi-modal data includes collected environmental data, physiological data and behavioral data;Condition judgment module, the multi-modal data for being acquired according to the acquisition module, determines user's current state;Processing module, user's current state for being determined according to the condition judgment module execute corresponding processing.The present invention can consider the multi-modal data of acquisition and react, and intelligent level is high, be suitable for different occasions.
Description
Technical field
The present invention relates to human-computer interaction technique field, especially a kind of human-computer interaction system based on multi-modal data acquisition
System.
Background technology
In the prior art, man-machine interactive system is all single data to be obtained from user, and be subject to the data mostly
Corresponding response is made after analysis and judgement, the type of gathered data is single, cannot analyze the case where user will express, intelligence comprehensively
Energyization is horizontal low, cannot meet user's use.
Invention content
In view of the above-mentioned problems, the present invention is intended to provide a kind of man-machine interactive system based on multi-modal data acquisition.
The purpose of the present invention is realized using following technical scheme:
A kind of man-machine interactive system based on multi-modal data acquisition, including:
Acquisition module, for acquiring multi-modal data, the multi-modal data includes collected environmental data, physiology number
According to and behavioral data;
Condition judgment module, the multi-modal data for being acquired according to the acquisition module, determines user's current state;
Processing module, user's current state for being determined according to the condition judgment module execute corresponding processing.
Preferably, the processing module, be specifically used for according in user's current state beyond the clouds database into line number
According to matching, suitable processing scheme is obtained, and corresponding processing is executed according to the processing scheme.
Preferably, the acquisition module specifically includes environment collecting device, physiological acquisition equipment and behavior collecting device,
Wherein, the environment collecting device includes temperature data acquisition equipment, one or more in humidity data collecting device;Physiology
Collecting device includes brain electric data collecting equipment, blood pressure data collecting device, temperature data collecting device, and breath data acquisition is set
It is one or more in standby;The behavior collecting device includes image capture device, one or more in voice capture device.
The present invention provides a kind of man-machine interactive systems based on multi-modal data acquisition, and user is acquired by acquisition module
And the multi-modal data of environment, and analyzed according to the data of acquisition, determine the current state of user, based on this
It being handled, the multi-modal data of acquisition can be considered and is reacted with corresponding processing scheme, intelligent level is high,
Suitable for different occasions.
Description of the drawings
Using attached drawing, the invention will be further described, but the embodiment in attached drawing does not constitute any limit to the present invention
System, for those of ordinary skill in the art, without creative efforts, can also obtain according to the following drawings
Other attached drawings.
Fig. 1 is the frame construction drawing of the present invention;
Fig. 2 is the frame construction drawing of condition judgment module of the present invention;
Fig. 3 is the frame construction drawing of data fusion unit of the present invention.
Specific implementation mode
In conjunction with following application scenarios, the invention will be further described.
Referring to Fig. 1, it illustrates a kind of man-machine interactive systems based on multi-modal data acquisition, including:
Acquisition module 1, for acquiring multi-modal data, multi-modal data includes collected environmental data, physiological data
And behavioral data;
Condition judgment module 2, the multi-modal data for being acquired according to acquisition module 1, determines user's current state;
Processing module 3, user's current state for being determined according to condition judgment module 2 execute corresponding processing.
The above embodiment of the present invention, acquires the multi-modal data of user and environment by acquisition module 1, and according to adopting
The data of collection are analyzed, and determine the current state of user, are matched corresponding processing scheme based on this and are handled, can
Consider the multi-modal data of acquisition and react, intelligent level is high, is suitable for different occasions.
Preferably, environmental data includes the temperature of system local environment, humidity etc.;
Physiological data includes the eeg data of user, blood pressure data, temperature data, breath data etc.;
Behavioral data includes the countenance data of user, eye movement data and voice data etc..
Preferably, processing module 3 is connected with the various electric appliances in intelligent domestic system, is controlled according to user's current state
Electric appliance is reconciled in control, to adapt to user's current state;Such as:When judging that the current temperature data of user and breath data be higher than
When normality threshold, processing module 3 controls air-conditioning and automatically opens and reconcile suitable temperature.
Preferably, processing module 3, specifically for carrying out Data Matching according in user's current state beyond the clouds database,
Suitable processing scheme is obtained, and corresponding processing is executed according to processing scheme.
Preferably, acquisition module 1 specifically includes environment collecting device, physiological acquisition equipment and behavior collecting device,
In, environment collecting device includes temperature data acquisition equipment, one or more in humidity data collecting device;Physiological acquisition is set
Standby includes brain electric data collecting equipment, blood pressure data collecting device, temperature data collecting device, in breath data collecting device
It is one or more;Behavior collecting device includes image capture device, one or more in voice capture device.
Preferably, referring to Fig. 2, condition judgment module 2 further includes data fusion unit 21 and analyzes and determines unit 22,
In, data fusion unit 21 is specifically used for 1 collected multi-modal data of acquisition module carrying out Data Fusion;Analysis is sentenced
Disconnected unit 22 is specifically used for carrying out analyzing processing to the data after progress data fusion in data integrated unit 21, determines that user works as
Preceding state.
The above embodiment of the present invention due to the data type and complexity that are acquired from different collecting devices and differs,
In order to reduce the redundancy of data and improve the processing speed of interactive system, the data acquired from acquisition module 1 are divided
The data of same type are merged and carry out analysis and judgement processing again, can effectively mitigation state be sentenced by class and fusion treatment
The burden of disconnected module 2, improves the process performance of system.
Preferably, referring to Fig. 3, data fusion unit 21 further includes pretreatment subelement 211, data fusion subelement 212
With fusion revise subelemen 213, wherein pretreatment subelement 211 is used for the different acquisition equipment acquisition from acquisition module 1
Data carry out classification processing;Data fusion subelement 212, for the grouped data conduct obtained in subelement 211 will to be pre-processed
Data source carries out fusion to the data of different data sources and secondary classification is handled;Revise subelemen 213 is merged, for fusion
Data be modified, determine final data fusion result.
The above embodiment of the present invention carries out fusion treatment using above-mentioned layered shaping mode to the data of acquisition, can
The harmful effect that noise and uncertain data are brought effectively effectively is reduced, the performance of data fusion unit is improved.
Preferably, subelement 211 is pre-processed, is specifically included:Different acquisition is set using non-directed graph model training grader
The standby data obtained carry out classification processing;
Wherein, the training of grader specifically includes:
(1) mark training sample set ψ of the initialization for training grader, the data sequence obtained from collecting device are made
Not mark sample, sample set y is charged to(n), test sample collection c, using training sample set training preliminary classification device H(n), wherein changing
Generation number n=0, and use grader H(n)To sample set y(n)In do not mark sample carry out probabilistic forecasting;
(2) according to undirected graph model in sample set y(n)Upper construction non-directed graph, and reject the isolated point in image graph, that is, it makes an uproar
Sound sample point, and by isolated point from sample set y(n)Middle rejecting;
(3) grader H is utilized in each connected region in non-directed graph(n)Forecast sample belongs to the probability of each classification,
And the current value for not marking sample each is obtained, it is candidate that the maximum sample composition of current value is selected out of each connected region
Sample set Φ, and the sample optimization value of candidate samples collection Φ is obtained,
Wherein, the acquisition function of sample current value is:
In formula, α expressions do not mark sample, J (α, H(n)) indicate not marking sample α to grader H(n)Value,WithIndicate that not marking sample α utilizes grader H respectively(n)The optimal and suboptimum classification of prediction is general
Rate, β1And β2It is the optimal and suboptimum class label of the sample respectively;
I.e. candidate samples collection Φ is represented by:
In formula, α expressions do not mark sample, and y expressions do not mark sample set;
Wherein, the sample optimization value function used for:
In formula, Y (α) indicates the optimal value of sample α in sample set Φ,WithIt indicates respectively not
It marks sample α and utilizes grader H(n)The optimal classification β of prediction1With suboptimum classification β1Probability,WithIndicate that not marking sample α utilizes provisional classifications device H respectively(n+1)(β1) prediction optimal classification β '1With it is secondary
Excellent classification β '2Probability, wherein provisional classifications device H(n+1)(β1) it is the sample that candidate samples collection Φ is added using current training sample set ψ
This α and its optimal classification label β1Training gained,WithIt indicates not mark sample respectively
This α utilizes provisional classifications device H(n+1)(β2) prediction optimal classification β "1With suboptimum classification β "2Probability, wherein provisional classifications device H(n +1)(β2) it is the sample α and next excellent tag along sort β that candidate samples collection Φ is added using current training sample set ψ2Training gained;
The sample that do not mark that optimal value in sample set Φ is more than to the threshold value W of setting is labeled, and is added to trained sample
In this collection ψ;
(4) training sample set ψ update graders H is utilized(n), carried out on test sample collection c using updated grader
Test, calculates the classification accuracy rate of grader, if accuracy is more than the threshold value T of setting or number of training reaches setting
Threshold value or it is front and back twice test in training sample set ψ sizes no longer increase, then terminate to train;Otherwise (3) continuation is jumped to
Suitable sample is selected to be trained, iterations n=n+1.
The above embodiment of the present invention adopts the data that grader trained in manner just described acquires different acquisition equipment
Classification processing is carried out, the data classification processing method based on undirected graph model is used, effectively according to the dependence between data
Relationship proposes noise data, the harmful effect for avoiding noise data from bringing;Undirected graph model is chosen according to sample optimization value simultaneously
In different types of optimization data grader is trained so that grader to the adaptability of different classifications type significantly
It improves, improves the accuracy of data classification indirectly.
Preferably, data fusion subelement 212, for the grouped data obtained in subelement 211 will to be pre-processed as number
According to source, fusion is carried out to the data of different data sources and secondary classification is handled, is specifically included:
Using the grouped data obtained using different undirected graph models as different aforementioned sources, by using based on D-S evidences
Theoretical union rule carries out fusion treatment to information source, to form a new pooling information source, specifically includes:
(1) information source from different acquisition equipment is obtained, remembers that the quantity of information source is L, establishes identification framework D=[d1,
d2,…,dV], wherein d1,d2,…,dVIndicate that datum target type, V indicate the sum of the data type of setting, v class data class
The feature vector of type is represented by:Gv=[gv1,gv2,…,gvw]T, wherein information source and identification framework is all w dimensional feature vectors;
(2) for each information source Y, eigenvectors matrix R, R=that information source Y is constituted with identification framework D are calculated
(R1,R2,…,RV+1)=(Y, G1,G2,…,GV), wherein wherein Y=[u1,u2,…,uw]T, the data characteristics of u expression information sources
Component;
(3) the similarity relation λ in calculating matrix R between each componentv,j, constitute Y and GvRelational matrix Z:
In formula,Indicate v1Kth dimensional feature vector in a data type;
(4) relational matrix Z is converted to its transitive closure matrixWherein transitive closure matrixIn row vector ElementBy target d in target to be identified and identification framework when being fusionv-1It is divided into one kind
Certainty value, settingThat is bvIndicate that data to be identified are identified as target type dvCertainty value;
(5) information source Y is obtained to target type dvAppropriateness value B (v),
The acquisition function being moderately worth wherein used for:
In formula, bvIndicate that data to be identified are identified as target type dvCertainty value, σ representative function regulatory factors;
If information source Y is to all target type dvAppropriateness value B (v) be respectively less than set threshold value, then information source Y is marked
It is denoted as uncertain type;Otherwise the target type d corresponding to appropriateness value B (v) maximums is chosenvData type as information source Y;
(6) for all information source, same target classification d will be belonged to using Dempster rules of combinationvInformation source into
Row fusion, obtains data fusion result.
The above embodiment of the present invention adopts the preprocessed subelement in manner just described to being obtained from different collecting devices
Grouped data that treated carries out fusion treatment, at the D-S evidence theory union rule information source different to data
Reason obtains the eigenmatrix according to information source and identification framework composition respectively, obtains appropriateness value of the information source to different classifications, then
The information source for belonging to same category is merged using the method based on Dempster rules of combination, data can be effectively improved
The performance of fusion.
Preferably, fusion revise subelemen 213 is stated, is specifically included:It is true to label to use quadratic classifier Ε (γ)
The information source for determining type carries out subseries again, determines its final classification result, wherein the foundation of quadratic classifier specifically includes:
(1) it uses in data fusion subelement 212 and classifies successful information source as training sample (γ1,y1),(γ2,
y2),…,(γN,yN), wherein γn∈ Γ, Γ indicate that training sample space, N indicate training sample sum, ynIndicate information source
γnClass vector Indicate ynKth tie up subcomponent, indicate γnIt is divided into classificationIt is general
Rate vector, K indicate the sum of the data type of setting;
(2) remember iterations m=1, initialize training sample weight,Preliminary classification device Εm(γ);
(3) data sample distribution p is calculatedm, and by data sample distribution pmPass to grader Εm(γ) is simultaneously fitted training
Sample calculates ΕmThe error q of (γ)m,
(4) Dynamic gene is setCalculating judge factor ξ, wherein use judge saturation as:
In formula,Indicate training sample γnKth dimension classification subvector;
(5) to judging that the factor judges:
If it is determined that factor ξ is more than the threshold value of setting, then weight distribution is adjustedAnd the m times iteration is re-started,
Otherwise, weight is redistributedAnd carry out next round iteration, m=m+1, wherein
In formula, [| Εm(γn)=yn|] indicate the m times iteration in grader Εm(γ) is by sample γnClassifying, it is corresponded to
Target classification y 'nProbability;
When the maximum number of iterations is reached, the condition that Integration obtaining grader Ε (γ), wherein grader Ε (γ) meet
For:
Wherein, the Ε for meeting above-mentioned requirements is chosenm(γ) is as finally determining quadratic classifier Ε (γ).
The above embodiment of the present invention adopts the information source with the aforedescribed process to not determining classification in data fusant unit
Carry out secondary classification, choose the information source of successful classification in data fusion subelement as training sample, to quadratic classifier into
Row training, by adjusting the weight of different samples in training sample so that the quadratic classifier that trained sample training goes out has
Better adaptability and higher accuracy, it is accurate to determine its classification type to not determining that the information source of classification carries out classification
Exactness is high.
The above embodiment of the present invention, using layer-stepping Data Fusion, first by pre-processing subelement to never
Classify with the data acquired in collecting device, the data for belonging to same data type are categorized into together, number is then passed through
Data Fusion is carried out to same type of data (information source) according to fusion subelement, confirms its target classification, finally by
It merges revise subelemen and the data (information source) of uncertain type is subjected to secondary classification processing, confirm its final classification class
Type effectively increases accuracy and the efficiency of data fusion, to be for further processing to the data of acquisition in man-machine interactive system
It lays a good foundation.
Finally it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than the present invention is protected
The limitation of range is protected, although being explained in detail to the present invention with reference to preferred embodiment, those skilled in the art answer
Work as analysis, technical scheme of the present invention can be modified or replaced equivalently, without departing from the reality of technical solution of the present invention
Matter and range.
Claims (6)
1. a kind of man-machine interactive system based on multi-modal data acquisition, which is characterized in that including:
Acquisition module, for acquiring multi-modal data, the multi-modal data includes collected environmental data, physiological data and
Behavioral data;
Condition judgment module, the multi-modal data for being acquired according to the acquisition module, determines user's current state;
Processing module, user's current state for being determined according to the condition judgment module execute corresponding processing.
2. a kind of man-machine interactive system based on multi-modal data acquisition according to claim 1, which is characterized in that described
Processing module, specifically for according to Data Matching is carried out in user's current state beyond the clouds database, obtaining suitable place
Reason scheme, and corresponding processing is executed according to the processing scheme.
3. a kind of man-machine interactive system based on multi-modal data acquisition according to claim 1, which is characterized in that described
Acquisition module specifically includes environment collecting device, physiological acquisition equipment and behavior collecting device, wherein the environment acquisition is set
Standby includes temperature data acquisition equipment, one or more in humidity data collecting device;Physiological acquisition equipment includes brain electricity number
According to collecting device, blood pressure data collecting device, temperature data collecting device is one or more in breath data collecting device;
The behavior collecting device includes image capture device, one or more in voice capture device.
4. a kind of man-machine interactive system based on multi-modal data acquisition according to claim 3, which is characterized in that described
Condition judgment module further includes data fusion unit and analytical judgment unit, wherein the data fusion unit is specifically used for will
The collected multi-modal data of acquisition module carries out Data Fusion;The analytical judgment unit is specifically used for described
The data after data fusion are carried out in data fusion unit and carry out analyzing processing, determine user's current state.
5. a kind of man-machine interactive system based on multi-modal data acquisition according to claim 4, which is characterized in that described
Data fusion unit further includes pretreatment subelement, data fusion subelement and fusion revise subelemen, wherein the pre- place
Reason subelement is used to carry out classification processing to the data that different acquisition equipment obtains from the acquisition module;The data fusion
Subelement, for will in the pretreatment subelement grouped data that obtains as data source, to the data of different data sources into
Row fusion and secondary classification processing;The fusion revise subelemen, is modified for the data to fusion, determines final number
According to fusion results.
6. a kind of man-machine interactive system based on multi-modal data acquisition according to claim 5, which is characterized in that described
Subelement is pre-processed, is specifically included:The data obtained to different acquisition equipment using non-directed graph model training grader are divided
Class processing;
Wherein, the training of the grader specifically includes:
(1) mark training sample set ψ of the initialization for training grader, the data sequence obtained from the collecting device are made
Not mark sample, sample set y is charged to(n), test sample collection c, using training sample set training preliminary classification device H(n), wherein changing
Generation number n=0, and use grader H(n)To sample set y(n)In do not mark sample carry out probabilistic forecasting;
(2) according to undirected graph model in sample set y(n)Upper construction non-directed graph, and reject the isolated point in image graph, i.e. noise sample
This point, and by isolated point from sample set y(n)Middle rejecting;
(3) grader H is utilized in each connected region in non-directed graph(n)Forecast sample belongs to the probability of each classification, and obtains
Each current value for not marking sample is taken, the maximum sample composition candidate samples of current value are selected out of each connected region
Collect Φ, and obtain the sample optimization value of candidate samples collection Φ,
Wherein, the acquisition function of the sample current value is:
In formula, α expressions do not mark sample, J (α, H(n)) indicate not marking sample α to grader H(n)Value,WithIndicate that not marking sample α utilizes grader H respectively(n)The optimal and suboptimum class probability of prediction, β1And β2Respectively
It is the optimal and suboptimum class label of the sample;
I.e. candidate samples collection Φ is represented by:
In formula, α expressions do not mark sample, and y expressions do not mark sample set;
Wherein, the sample optimization value function used for:
In formula, Y (α) indicates the optimal value of sample α in sample set Φ,WithIt indicates not mark respectively
Sample α utilizes grader H(n)The optimal classification β of prediction1With suboptimum classification β1Probability,WithIndicate that not marking sample α utilizes provisional classifications device H respectively(n+1)(β1) prediction optimal classification β '1With it is secondary
Excellent classification β '2Probability, wherein the provisional classifications device H(n+1)(β1) it is that candidate samples collection Φ is added using current training sample set ψ
Sample α and its optimal classification label β1Training gained,WithIt indicates respectively not
It marks sample α and utilizes provisional classifications device H(n+1)(β2) prediction optimal classification β "1With suboptimum classification β "2Probability, wherein described interim
Grader H(n+1)(β2) it is the sample α and next excellent tag along sort β that candidate samples collection Φ is added using current training sample set ψ2
Training gained;
The sample that do not mark that optimal value in sample set Φ is more than to the threshold value W of setting is labeled, and is added to training sample set ψ
In;
(4) training sample set ψ update graders H is utilized(n), surveyed on test sample collection c using updated grader
Examination, calculates the classification accuracy rate of grader, if accuracy is more than the threshold value T of setting or number of training reaches setting
Training sample set ψ sizes no longer increase in threshold value or front and back test twice, then terminate to train;Otherwise (3) are jumped to continue to select
It selects suitable sample to be trained, iterations n=n+1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810509996.7A CN108762503A (en) | 2018-05-24 | 2018-05-24 | A kind of man-machine interactive system based on multi-modal data acquisition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810509996.7A CN108762503A (en) | 2018-05-24 | 2018-05-24 | A kind of man-machine interactive system based on multi-modal data acquisition |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108762503A true CN108762503A (en) | 2018-11-06 |
Family
ID=64005904
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810509996.7A Pending CN108762503A (en) | 2018-05-24 | 2018-05-24 | A kind of man-machine interactive system based on multi-modal data acquisition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108762503A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109920544A (en) * | 2019-01-28 | 2019-06-21 | 浙江工业大学 | Real-time adaptive intelligent Building System based on body-sensing information |
CN111966724A (en) * | 2020-06-29 | 2020-11-20 | 北京津发科技股份有限公司 | Interactive behavior data acquisition and analysis method and device based on human-computer interaction interface area automatic identification technology |
CN112613534A (en) * | 2020-12-07 | 2021-04-06 | 北京理工大学 | Multi-mode information processing and interaction system |
CN112687022A (en) * | 2020-12-18 | 2021-04-20 | 山东盛帆蓝海电气有限公司 | Intelligent building inspection method and system based on video |
CN113780426A (en) * | 2021-09-14 | 2021-12-10 | 中国联合网络通信集团有限公司 | Multi-mode information fusion method, MEC, modal information acquisition unit and system |
CN114760201A (en) * | 2022-03-30 | 2022-07-15 | 烽台科技(北京)有限公司 | Data acquisition method, device and equipment of industrial control equipment and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8406926B1 (en) * | 2011-05-06 | 2013-03-26 | Google Inc. | Methods and systems for robotic analysis of environmental conditions and response thereto |
CN104298722A (en) * | 2014-09-24 | 2015-01-21 | 张鸿勋 | Multimedia interaction system and method |
CN105807913A (en) * | 2015-01-19 | 2016-07-27 | 三星电子株式会社 | Wearable device based on bioinformation, system, and operating method |
CN107272900A (en) * | 2017-06-21 | 2017-10-20 | 叶富阳 | A kind of wearable music player of autonomous type |
-
2018
- 2018-05-24 CN CN201810509996.7A patent/CN108762503A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8406926B1 (en) * | 2011-05-06 | 2013-03-26 | Google Inc. | Methods and systems for robotic analysis of environmental conditions and response thereto |
CN104298722A (en) * | 2014-09-24 | 2015-01-21 | 张鸿勋 | Multimedia interaction system and method |
CN105807913A (en) * | 2015-01-19 | 2016-07-27 | 三星电子株式会社 | Wearable device based on bioinformation, system, and operating method |
CN107272900A (en) * | 2017-06-21 | 2017-10-20 | 叶富阳 | A kind of wearable music player of autonomous type |
Non-Patent Citations (1)
Title |
---|
时玉杰: "基于重采样思想的图像主动分类方法研究", 《中国优秀硕士学位论文全文数据库(电子期刊)•信息科技辑》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109920544A (en) * | 2019-01-28 | 2019-06-21 | 浙江工业大学 | Real-time adaptive intelligent Building System based on body-sensing information |
CN109920544B (en) * | 2019-01-28 | 2021-06-08 | 浙江工业大学 | Real-time self-adaptive intelligent building system based on somatosensory information |
CN111966724A (en) * | 2020-06-29 | 2020-11-20 | 北京津发科技股份有限公司 | Interactive behavior data acquisition and analysis method and device based on human-computer interaction interface area automatic identification technology |
CN112613534A (en) * | 2020-12-07 | 2021-04-06 | 北京理工大学 | Multi-mode information processing and interaction system |
CN112687022A (en) * | 2020-12-18 | 2021-04-20 | 山东盛帆蓝海电气有限公司 | Intelligent building inspection method and system based on video |
CN113780426A (en) * | 2021-09-14 | 2021-12-10 | 中国联合网络通信集团有限公司 | Multi-mode information fusion method, MEC, modal information acquisition unit and system |
CN113780426B (en) * | 2021-09-14 | 2023-06-30 | 中国联合网络通信集团有限公司 | Multi-mode information fusion method, MEC, mode information acquisition unit and system |
CN114760201A (en) * | 2022-03-30 | 2022-07-15 | 烽台科技(北京)有限公司 | Data acquisition method, device and equipment of industrial control equipment and storage medium |
CN114760201B (en) * | 2022-03-30 | 2023-05-23 | 烽台科技(北京)有限公司 | Data acquisition method, device and equipment of industrial control equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108762503A (en) | A kind of man-machine interactive system based on multi-modal data acquisition | |
CN111027487B (en) | Behavior recognition system, method, medium and equipment based on multi-convolution kernel residual error network | |
CN103268495B (en) | Human body behavior modeling recognition methods based on priori knowledge cluster in computer system | |
CN105488456B (en) | Method for detecting human face based on adaptive threshold adjustment rejection sub-space learning | |
CN109166094A (en) | A kind of insulator breakdown positioning identifying method based on deep learning | |
CN110738247B (en) | Fine-grained image classification method based on selective sparse sampling | |
CN109614488B (en) | Text classification and image recognition-based distribution network live working condition judgment method | |
CN106529499A (en) | Fourier descriptor and gait energy image fusion feature-based gait identification method | |
CN105354565A (en) | Full convolution network based facial feature positioning and distinguishing method and system | |
CN107256017B (en) | Route planning method and system | |
CN106971197A (en) | The Subspace clustering method of multi-view data based on otherness and consistency constraint | |
CN111832608A (en) | Multi-abrasive-particle identification method for ferrographic image based on single-stage detection model yolov3 | |
CN110555417A (en) | Video image recognition system and method based on deep learning | |
CN109033944A (en) | A kind of all-sky aurora image classification and crucial partial structurtes localization method and system | |
CN105956570A (en) | Lip characteristic and deep learning based smiling face recognition method | |
CN110458022A (en) | It is a kind of based on domain adapt to can autonomous learning object detection method | |
CN111931562A (en) | Unsupervised feature selection method and system based on soft label regression | |
CN106529470A (en) | Gesture recognition method based on multistage depth convolution neural network | |
CN112364708A (en) | Multi-mode human body action recognition method based on knowledge distillation and antagonistic learning | |
CN105868776A (en) | Transformer equipment recognition method and device based on image processing technology | |
CN106056627A (en) | Robustness object tracking method based on local identification sparse representation | |
CN103942572A (en) | Method and device for extracting facial expression features based on bidirectional compressed data space dimension reduction | |
CN111325282B (en) | Mammary gland X-ray image identification method and device adapting to multiple models | |
CN106951924B (en) | Seismic coherence body image fault automatic identification method and system based on AdaBoost algorithm | |
CN110378405A (en) | The Hyperspectral Remote Sensing Imagery Classification method of Adaboost algorithm based on transfer learning |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181106 |