CN110472576A - A kind of method and device for realizing mobile human body Activity recognition - Google Patents
A kind of method and device for realizing mobile human body Activity recognition Download PDFInfo
- Publication number
- CN110472576A CN110472576A CN201910752716.XA CN201910752716A CN110472576A CN 110472576 A CN110472576 A CN 110472576A CN 201910752716 A CN201910752716 A CN 201910752716A CN 110472576 A CN110472576 A CN 110472576A
- Authority
- CN
- China
- Prior art keywords
- data
- label
- compressed sensing
- new data
- new
- 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
- 238000000034 method Methods 0.000 title claims abstract description 35
- 230000037081 physical activity Effects 0.000 title claims abstract description 19
- 230000006399 behavior Effects 0.000 claims abstract description 42
- 239000011159 matrix material Substances 0.000 claims abstract description 29
- 230000001133 acceleration Effects 0.000 claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 16
- 230000004044 response Effects 0.000 claims abstract description 15
- 238000005516 engineering process Methods 0.000 claims abstract description 10
- 238000013526 transfer learning Methods 0.000 claims abstract description 10
- 241001269238 Data Species 0.000 claims description 5
- 241000894007 species Species 0.000 claims description 3
- 230000001360 synchronised effect Effects 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 2
- 238000010276 construction Methods 0.000 abstract 1
- 239000013598 vector Substances 0.000 description 9
- 230000000694 effects Effects 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 210000004709 eyebrow Anatomy 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 208000012802 recumbency Diseases 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention relates to mobile application field, espespecially a kind of method and device for realizing mobile human body Activity recognition；This method acquires the 3-axis acceleration data of certain amount basic user difference behavior by being embedded in the sensor of mobile device, and based on the excessively complete matrix dictionary in compressed sensing technology construction basis；New user is demarcated without label new data by transfer learning method, and the excessively complete matrix dictionary of personalization for being suitble to the user is reconstructed using calibrated new data, new user behavior is identified finally by the compressed sensing classifier based on personalized excessively complete matrix dictionary.Mobile human body Activity recognition device based on the above method includes compressed sensing processing unit, without label new data calibration unit and Human bodys' response unit, compressed sensing processing unit and unit is two-way connects without the calibration of label new data, compressed sensing processing unit is connected to Human bodys' response unit.
Description
Technical field
Present document relates to mobile application field, espespecially a kind of method and device for realizing mobile human body Activity recognition.
Background technique
It is quickly universal and high-speed mobile internet continuous with mobile devices such as smart phone, smartwatch/bracelets
Development based on the acceleration transducer acquisition data embedded in mobile device and carries out Human bodys' response and is possibly realized.It is based on
The Human bodys' response technology of mobile device motion monitoring, smart city, in terms of there is very extensive answer
Use prospect.
It carries out mobile human body Activity recognition to need the difficulties solved being the diversification and shifting of mobile device wearing position
The personalization of dynamic user behavior.The mobile human body Activity recognition method having disclosed at present is mainly based upon prior collected sample
Notebook data trains individual generic classifier or the multi-categorizer by fusion carries out Classification and Identification.The method is in user group
Population diversity problem is not can effectively solve when scale constantly expands, so that recognition accuracy be caused to decline.
Summary of the invention
The present invention will provide a kind of method and device for realizing mobile human body Activity recognition, of the existing technology to overcome
Population diversity problem is not can effectively solve, thus the problem of causing recognition accuracy to decline.
In order to reach the purpose of the present invention, present invention provide the technical scheme that
A method of realizing mobile human body Activity recognition, comprising the following steps:
Step 1 is synchronized including two steps:
Step 101 acquires several basic user difference behaviors by being embedded in the acceleration transducer of mobile device
There are label 3-axis acceleration data as basic data, it is excessively complete using these basic datas foundation basis based on compressed sensing technology
Standby matrix dictionary；
Step 102, acquisition basic user except several new user's difference behaviors without label 3-axis acceleration data
As new data, label calibration is carried out to no label new data by clustering algorithm and Feature Correspondence Algorithm, is obtained calibrated
New data；
Step 2 is based on calibrated new data, according to Sparse representation method reconstructed base mistake in compressed sensing technology
Complete matrix dictionary obtains the excessively complete matrix dictionary of personalization for adapting to new user；
Step 3 establishes compressed sensing classifier based on personalized excessively complete matrix dictionary, the movement to needing to identify behavior
The data to be identified of user carry out sparse coefficient and residual computations, obtain recognition result according to least residual value.
Further, in the step 102, method that no label new data is demarcated, comprising the following steps:
Step 1021 carries out cluster operation, choosing without label 3-axis acceleration data to several new user's difference behaviors
Taking can apart from the height that the data that cluster centre radius is less than d (0.01 < d < 0.1) are set as every kind of behavior type in every a kind of data
Letter data；
Step 1022, to the high trust data of every one kind extract mean value, median, variance, standard deviation, maximum value, minimum value,
The features such as root mean square and range, by carrying out these features with the feature for having various species data in the basic data of label
It is calculated with degree；
Label of every class without label new data can be obtained according to matching degree calculated result, that is, demarcate for step 1023
New data afterwards.
According to the device that the method for above-mentioned realization mobile human body Activity recognition constructs, including compressed sensing processing unit, nothing
Label new data demarcates unit and Human bodys' response unit, compressed sensing processing unit and double without label new data calibration unit
To connecting, compressed sensing processing unit is connected to Human bodys' response unit；
Wherein, compressed sensing processing unit, being set as basis has the basic data of label to construct the excessively complete matrix in basis
Dictionary, and reconstructed to obtain the excessively complete matrix dictionary of personalization for adapting to new user according to calibrated new data；
No label new data demarcates unit, is set as carrying out label mark to no label new data according to transfer learning model
It is fixed；
Human bodys' response unit is set as to the mobile subscriber for needing to identify behavior without label 3-axis acceleration data
It is identified to obtain result.
Compared with prior art, the invention has the advantages that
1, it can effectively solve the problems, such as population diversity, improve the recognition accuracy of new user: carry out Activity recognition at present
Method have very much, Activity recognition also may be implemented using only compressed sensing technology, but these methods not can effectively solve
Population diversity problem.Transfer learning method biggest advantage is can to well solve population diversity problem, present invention benefit
Compression sensing method is improved to realize Activity recognition with the thought of transfer learning technology, and Activity recognition accuracy rate can be improved,
The Activity recognition accuracy rate of the new user especially to differ greatly with basic user behavior pattern.
2, a kind of mobile human body Activity recognition device based on compressed sensing transfer learning proposed by the present invention, in use not
It needs to carry out the preprocessing process such as data filtering, does not need to carry out resultant acceleration operation to original 3-axis acceleration data, because
And computation complexity is reduced, advantageously reduce the calculation amount of mobile device CPU, energy saving.
Detailed description of the invention
Fig. 1 is the flow chart for the mobile human body Activity recognition method that the embodiment of the present invention is realized；
Fig. 2 is the flow chart without label new data scaling method that the embodiment of the present invention is realized；
Fig. 3 is the structure journey figure for the mobile human body Activity recognition device that the embodiment of the present invention is realized.
Specific embodiment
Below by specific embodiment combination attached drawing, invention is further described in detail.Wherein different embodiments
Middle similar component uses associated similar element numbers.In the following embodiments, many datail descriptions be in order to
The application is better understood.However, those skilled in the art can recognize without lifting an eyebrow, part of feature
It is dispensed, or can be substituted by other elements, material, method in varied situations.
The embodiment of the present invention is described in detail below in conjunction with attached drawing.
It is a kind of mobile human body Activity recognition method based on compressed sensing transfer learning provided by the invention referring to Fig. 1,
Specifically includes the following steps:
Step 1 is synchronized including following two treatment process:
Step 101 acquires several basic user difference behaviors by being embedded in the acceleration transducer of mobile device
There are label 3-axis acceleration data as basic data.In order to make basic data that there is enough diversity, basic data is acquired
When should cover mobile subscriber as how different types of as possible, the user including different sexes, all ages and classes, different behavioural habits.
Common mobile subscriber's daily behavior include walk, upstairs, downstairs, stand, sit quietly, recumbency etc., the present embodiment is with this six kinds of rows
For identification be illustrated.Identification of the application of the method for the present invention including but not limited to this six kinds of behaviors.Every kind of behavior is needed
Basic user is acquired mobile device is respectively placed in trouser pocket, is placed in packet and is held on three kinds of different locations in hand
Data.Every kind of behavior needs to repeat 10 times in the data acquisition of each different location, and the time acquired every time is 10s,
Sample frequency is 50Hz.In order to reduce the influence of experimental noise, for 10s data collected, intermediate 6s data is only taken to make
For the data finally used.
If sometime the original 3-axis acceleration data of certain behavior are vx、vy、vz, these three data are constituted one-dimensional
Observation vector v (v ∈ R3).N observation vector in certain period is connected and composed into one-dimensional vector V (V ∈ R againN, N=3n).If pair
The K class difference behavior of dry basic user carries out M respectivelyi(i=1,2 ..., K) secondary sampling, each class behavior all obtain N number of base
Plinth data.These basic data linearly connecteds are got up may make up a matrix A=[A1,A2,...,Ai,...,Ak]∈RN×M,
Wherein AiIndicate the submatrix that all sample datas for belonging to the i-th class behavior are constituted, Ai∈RN×Mi(M=M1+M2+...+MK).It presses
According to Sparse representation theory, matrix A is to describe the excessively complete matrix dictionary in basis of different user behavior.
Step 102, in a comparable manner acquire basic user except several new user's difference behaviors without label
3-axis acceleration data are as new data.The method for carrying out label calibration to these new datas is as shown in Figure 2, comprising: will own
New data carries out cluster operation, chooses in every a kind of data and sets apart from the data that cluster centre radius is less than d (0.01 < d < 0.1)
It is set to the high trust data of every kind of behavior type.Mean value, median, variance, standard deviation, most are extracted to the high trust data of every one kind
The features such as big value, minimum value, root mean square and range, by by these features and having a various species data in the basic data of label
Feature matched, label of every class without label new data, that is, calibrated new data can be obtained.
Step 2, according to Sparse representation method in compressed sensing technology these new datas respectively as belonging to difference
The new samples data submatrix of behavior is added in the excessively complete matrix dictionary A in basis built, obtains for adapting to new user
The excessively complete matrix dictionary of propertyization.
Step 3, according to compressive sensing theory, for data to be identified, that is, the data sample y of some behavior to be identified
∈RN×1, meet y=A α, wherein α=[0 ..., 0, αi,1,αi,2,...,αi,Mi,0,...,0]T∈RM×1.If row to be identified
It is to belong to the i-th class behavior for sample y, then the M in its coefficient vector α only with respective behavior position in dictionary matrix AiIt is a
Element is not zero, and the element in other positions is zero.Row only corresponding with sample to be identified in solution vector α ideally
It is not zero for the element of position, the element of other positions is zero, it is possible to the nonzero element directly from solution vector
Judge type belonging to the behavior in position.But due to the influence of the factors such as measurement noise, solution error, acquired solution vector
In will also tend to several nonzero elements occur in other positions.In such a case, it is possible to calculate behavior vector y to be measured with it is each
The residual values r of all training sample vectors corresponding to class behavior:Wherein AiIt is i-th
Class behavior submatrix corresponding in personalized excessively complete matrix dictionary A,It indicates according to AiPosition in A from solution to
AmountThe subvector of the corresponding position of middle taking-up.Assuming that shared K class behavior, then can calculate K residual values, choose wherein
Behavior classification corresponding to minimum value exports recognition result as classification belonging to sample to be tested behavior.It can based on the above method
It establishes compressed sensing classifier and treats the data of identification behavior and identified.
Referring to Fig. 3, the mobile human body Activity recognition device of realization provided by the invention is above-mentioned to be learned based on compressed sensing migration
Habit mobile human body Activity recognition method building device, including compressed sensing processing unit, without label new data demarcate unit
With Human bodys' response unit, compressed sensing processing unit and without the calibration of label new data, unit is two-way connects, at compressed sensing
Reason unit is connected to Human bodys' response unit；
Wherein, compressed sensing processing unit, being set as basis has the basic data of label to construct the excessively complete matrix in basis
Dictionary, and reconstructed to obtain the excessively complete matrix dictionary of personalization for adapting to new user according to calibrated new data；
No label new data demarcates unit, is set as carrying out label mark to no label new data according to transfer learning model
It is fixed；
Human bodys' response unit is set as to the mobile subscriber for needing to identify behavior without label 3-axis acceleration data
It is identified to obtain result.
Device will be described in more detail below:
Described compressed sensing processing unit, being set as basis has the basic data of label to construct the excessively complete matrix in basis
Dictionary, and reconstructed to obtain the excessively complete matrix dictionary of personalization for adapting to new user according to calibrated new data.The present embodiment
In compressed sensing processing unit constructed in Matlab simulated environment, used basic user data and new user data difference
The data set of UCI HAR data set and oneself acquisition from University of California.
Described demarcates unit without label new data, is set as marking no label new data according to transfer learning model
Label calibration.Constructing in Matlab simulated environment in the present embodiment without label new data calibration unit, uses K-Means algorithm
It is clustered, carries out characteristic matching using decision Tree algorithms.
Described Human bodys' response unit is set as accelerating the mobile subscriber for needing to identify behavior without three axis of label
Degree is according to being identified to obtain result.Human bodys' response unit in the present embodiment constructs in Matlab simulated environment, right
Compressed sensing minimum l1The solution of norm has used Stanford University's L1magic software package, be respectively adopted gaussian random matrix and
Bernoulli Jacob's random matrix projects complete matrix dictionary and data to be identified as observing matrix.
Optionally, other software environments and calculation method also can be used in the building of above each unit.
For those skilled in the art, under the premise of not departing from principle belonging to the present invention, may be used also
To make several improvements and modifications, these modifications and embellishments should also be considered as the scope of protection of the present invention.
Claims (3)
1. a kind of method for realizing mobile human body Activity recognition, which comprises the following steps:
Step 1 is synchronized including two steps:
Step 101 has mark by be embedded in that the acceleration transducer of mobile device acquires several basic user difference behaviors
3-axis acceleration data are signed as basic data, the excessively complete square in basis is established using these basic datas based on compressed sensing technology
Battle array dictionary；
Step 102, acquisition basic user except several new user's difference behaviors without label 3-axis acceleration data conduct
New data carries out label calibration to no label new data by clustering algorithm and Feature Correspondence Algorithm, obtains calibrated new number
According to；
Step 2 is based on calibrated new data, excessively complete according to Sparse representation method reconstructed base in compressed sensing technology
Matrix dictionary obtains the excessively complete matrix dictionary of personalization for adapting to new user；
Step 3 establishes compressed sensing classifier based on personalized excessively complete matrix dictionary, to the mobile subscriber for needing to identify behavior
Data to be identified carry out sparse coefficient and residual computations, recognition result is obtained according to least residual value.
2. the mobile human body Activity recognition method based on compressed sensing transfer learning according to claim 1, which is characterized in that
In the step 102, method that no label new data is demarcated, comprising the following steps:
Step 1021 carries out cluster operation without label 3-axis acceleration data to several new user's difference behaviors, chooses every
The credible number of height of every kind of behavior type is set as in a kind of data apart from the data that cluster centre radius is less than d (0.01 < d < 0.1)
According to；
Step 1022 extracts mean value to the high trust data of every one kind, is median, variance, standard deviation, maximum value, minimum value, square
The features such as root and range, by the way that these features are carried out matching degree with the feature for having various species data in the basic data of label
It calculates；
Label of every class without label new data can be obtained according to matching degree calculated result for step 1023, that is, calibrated
New data.
3. realizing the device of the method building of mobile human body Activity recognition according to claim 1, which is characterized in that including pressure
Contracting perceives processing unit, without label new data calibration unit and Human bodys' response unit, compressed sensing processing unit and without mark
Unit is two-way connects for the calibration of label new data, and compressed sensing processing unit is connected to Human bodys' response unit；
Wherein, compressed sensing processing unit, being set as basis has the basic data of label to construct the excessively complete matrix dictionary in basis,
And it is reconstructed to obtain the excessively complete matrix dictionary of personalization for adapting to new user according to calibrated new data；
No label new data demarcates unit, is set as carrying out label calibration to no label new data according to transfer learning model；
Human bodys' response unit is set as carrying out the mobile subscriber for needing to identify behavior without label 3-axis acceleration data
Identification obtains result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910752716.XA CN110472576A (en) | 2019-08-15 | 2019-08-15 | A kind of method and device for realizing mobile human body Activity recognition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910752716.XA CN110472576A (en) | 2019-08-15 | 2019-08-15 | A kind of method and device for realizing mobile human body Activity recognition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110472576A true CN110472576A (en) | 2019-11-19 |
Family
ID=68511352
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910752716.XA Pending CN110472576A (en) | 2019-08-15 | 2019-08-15 | A kind of method and device for realizing mobile human body Activity recognition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110472576A (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103544507A (en) * | 2013-10-15 | 2014-01-29 | 中国矿业大学 | High spectral data dimension reduction method based on pairwise constraint discriminant analysis-nonnegative sparse divergence |
| CN104850865A (en) * | 2015-06-05 | 2015-08-19 | 三峡大学 | Real-time compression tracking method of multi-characteristic transfer learning |
| CN105095863A (en) * | 2015-07-14 | 2015-11-25 | 西安电子科技大学 | Similarity-weight-semi-supervised-dictionary-learning-based human behavior identification method |
| CN106815806A (en) * | 2016-12-20 | 2017-06-09 | 浙江工业大学 | Single image SR method for reconstructing based on compressed sensing and SVR |
| CN107273908A (en) * | 2017-07-06 | 2017-10-20 | 清华大学深圳研究生院 | A kind of target identification method based on compressed sensing |
| CN107316049A (en) * | 2017-05-05 | 2017-11-03 | 华南理工大学 | A kind of transfer learning sorting technique based on semi-supervised self-training |
| CN108685570A (en) * | 2017-04-12 | 2018-10-23 | 中国科学院微电子研究所 | Method, device and system for processing over-complete dictionary |
| CN109117793A (en) * | 2018-08-16 | 2019-01-01 | 厦门大学 | Direct-push high Resolution Range Profile Identification of Radar method based on depth migration study |
| CN109993188A (en) * | 2018-01-02 | 2019-07-09 | ***通信有限公司研究院 | Data label recognition methods, Activity recognition method and device |
-
2019
- 2019-08-15 CN CN201910752716.XA patent/CN110472576A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103544507A (en) * | 2013-10-15 | 2014-01-29 | 中国矿业大学 | High spectral data dimension reduction method based on pairwise constraint discriminant analysis-nonnegative sparse divergence |
| CN104850865A (en) * | 2015-06-05 | 2015-08-19 | 三峡大学 | Real-time compression tracking method of multi-characteristic transfer learning |
| CN105095863A (en) * | 2015-07-14 | 2015-11-25 | 西安电子科技大学 | Similarity-weight-semi-supervised-dictionary-learning-based human behavior identification method |
| CN106815806A (en) * | 2016-12-20 | 2017-06-09 | 浙江工业大学 | Single image SR method for reconstructing based on compressed sensing and SVR |
| CN108685570A (en) * | 2017-04-12 | 2018-10-23 | 中国科学院微电子研究所 | Method, device and system for processing over-complete dictionary |
| CN107316049A (en) * | 2017-05-05 | 2017-11-03 | 华南理工大学 | A kind of transfer learning sorting technique based on semi-supervised self-training |
| CN107273908A (en) * | 2017-07-06 | 2017-10-20 | 清华大学深圳研究生院 | A kind of target identification method based on compressed sensing |
| CN109993188A (en) * | 2018-01-02 | 2019-07-09 | ***通信有限公司研究院 | Data label recognition methods, Activity recognition method and device |
| CN109117793A (en) * | 2018-08-16 | 2019-01-01 | 厦门大学 | Direct-push high Resolution Range Profile Identification of Radar method based on depth migration study |
Non-Patent Citations (4)
| Title |
|---|
| 宋辉等: "基于稀疏矩阵字典的移动用户行为识别方法", 《计算机应用研究》 * |
| 庄福振等: "基于混合正则化的无标签领域的归纳迁移学习", 《科学通报》 * |
| 贺炎等: "用户无关的多分类器融合行为识别模型", 《西安邮电大学学报》 * |
| 黄敏等: "基于过完备字典稀疏表示的云分类研究", 《郑州轻工业学院学报（自然科学版）》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107153871B (en) | Falling detection method based on convolutional neural network and mobile phone sensor data | |
| Zhu et al. | Efficient human activity recognition solving the confusing activities via deep ensemble learning | |
| Yang et al. | Distributed segmentation and classification of human actions using a wearable motion sensor network | |
| Henpraserttae et al. | Accurate activity recognition using a mobile phone regardless of device orientation and location | |
| US9060714B2 (en) | System for detection of body motion | |
| CN101561868B (en) | Human motion emotion identification method based on Gauss feature | |
| CN105224104B (en) | Pedestrian movement's state identification method based on smart mobile phone grip mode | |
| CN109685148A (en) | Multi-class human motion recognition method and identifying system | |
| CN109726662A (en) | Multi-class human posture recognition method based on convolution sum circulation combination neural net | |
| CN112183314B (en) | Expression information acquisition device, expression recognition method and system | |
| CN111954250B (en) | Lightweight Wi-Fi behavior sensing method and system | |
| CN109886068A (en) | Action behavior recognition methods based on exercise data | |
| CN108764123A (en) | Intelligent recognition human body sleep posture method based on neural network algorithm | |
| CN109765996A (en) | Insensitive gesture detection system and method are deviated to wearing position based on FMG armband | |
| CN109976526A (en) | A kind of sign Language Recognition Method based on surface myoelectric sensor and nine axle sensors | |
| CN105869354B (en) | A kind of Falls Among Old People detection method based on attractor propagation algorithm | |
| Ding et al. | Energy efficient human activity recognition using wearable sensors | |
| Ngo et al. | OU-ISIR wearable sensor-based gait challenge: Age and gender | |
| CN104269025A (en) | Wearable type single-node feature and position selecting method for monitoring outdoor tumble | |
| CN110175642A (en) | A kind of chrysanthemum similarity calculation method based on PCA dimensionality reduction and feature binary | |
| CN111178288A (en) | Human body posture recognition method and device based on local error layer-by-layer training | |
| CN109086667A (en) | Similar active recognition methods based on intelligent terminal | |
| CN110414590A (en) | Physical activity recognition methods based on Intelligent mobile equipment and convolutional neural networks | |
| Luqian et al. | Human activity recognition using time series pattern recognition model-based on tsfresh features | |
| CN110008847A (en) | A kind of stroke recognition methods based on convolutional neural networks |
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: 20191119 |