CN105426961B - The method for catching user's intention using Intelligent bracelet and smart mobile phone - Google Patents
The method for catching user's intention using Intelligent bracelet and smart mobile phone Download PDFInfo
- Publication number
- CN105426961B CN105426961B CN201510765304.1A CN201510765304A CN105426961B CN 105426961 B CN105426961 B CN 105426961B CN 201510765304 A CN201510765304 A CN 201510765304A CN 105426961 B CN105426961 B CN 105426961B
- Authority
- CN
- China
- Prior art keywords
- mobile phone
- bracelet
- output
- layer
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04M—TELEPHONIC COMMUNICATION
- H04M1/00—Substation equipment, e.g. for use by subscribers
- H04M1/72—Mobile telephones; Cordless telephones, i.e. devices for establishing wireless links to base stations without route selection
- H04M1/724—User interfaces specially adapted for cordless or mobile telephones
- H04M1/72403—User interfaces specially adapted for cordless or mobile telephones with means for local support of applications that increase the functionality
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- User Interface Of Digital Computer (AREA)
- Telephone Function (AREA)
Abstract
Disclosure of the invention is a kind of to catch the method that user is intended to using Intelligent bracelet and smart mobile phone, and the abrasion of mobile phone secondary or physical bond, implementation step are when mainly solving the problems, such as to light mobile phone screen：1. by obtaining characteristic action model with neural metwork training characteristic action data；2. the threshold value of motion gesture feature similarity factor and mobile phone acceleration is set respectively；3. mobile phone is connected with bracelet by bluetooth；4. bracelet monitors gravity sensor on backstage, when the hand for wearing bracelet takes mobile phone, bracelet start recording gravity sensor data；5. the output of characteristic action model, motion gesture feature similarity factor and mobile phone acceleration are calculated respectively；6. when motion gesture feature similarity factor and mobile phone acceleration all meet to be greater than or equal to given threshold value, screen is lighted, otherwise, returns to 4.Invention increases the function of smart mobile phone, reduces the abrasion of mobile phone secondary or physical bond, is brought conveniently to daily life.
Description
Technical field
The invention belongs to field of communication technology, more particularly to a kind of method that user's intention is caught using smart machine,
Available for various smart mobile phones.
Background technology
After first item smart mobile phone in 2001 is issued, smart mobile phone experienced very big conversion, open operation system
System, powerful internet wireless access capability fundamentally change the habits and customs of people invariably.
And 2012, Google glass exposes, and has directly triggered the arrival in " the intelligent wearable device first year ".By feat of can
The portability of wearable device and powerful data acquisition and interaction capabilities, wearable device are just triggering the technological revolution of a new round.
Intelligent bracelet is as using the widest wearable device of most convenient in wearable device, there is provided heart rate test, measurement distance,
The functions such as pedometer, sleep monitor, waterproof, Bluetooth transmission.It can also be matched additionally, due to Intelligent bracelet with smart mobile phone,
It can also realize that some specific functions, such as millet bracelet can realize no password unlock etc..
But either smart mobile phone or Intelligent bracelet, they cannot all identify the intention of people on one's own initiative, when people think
During using mobile phone, it is never because you want to automatically turn on the screen using mobile phone.And sensing built-in in wearable device
Device, it can capture your very small action change, but it identifies the implication less than action forever.And neutral net conduct
Pattern-recognition and the outstanding method of field of intelligent control, can identify the intention of people by training action model, so that
Mobile phone can identify on one's own initiative to be intended to rather than passive etc. to be operated.
The content of the invention
It is an object of the invention to for it is set forth above the problem of, propose it is a kind of by motion bracelet obtain data be input to
It is trained to obtain the identification model of different motion state in BP neural network and then controls smart mobile phone to reduce user
Active operation and reduction mobile phone physical abrasion.
The technical proposal of the invention is realized in this way：
One, technical principles
An important technology of the neutral net as machine learning, from it to simulate the mathematics side of the actual neutral net of the mankind
Since method comes out, huge success has been achieved in fields such as pattern-recognition, intelligent controls.The network topology knot of neural network model
Structure, node feature and learning rules determine its height robustness and fault-tolerant ability, and can fully approach the non-linear of complexity
Relation.
Neutral net is optimised typically by a learning method based on mathematical statistics type, so nerve net
Network is also a kind of practical application of mathematical statistics method, and can be obtained by statistical standard mathematical techniques largely can be with
With function come the partial structurtes space expressed.
The type of processing unit is divided into three classes in network：Input unit, output unit and hidden unit.Input unit receives outer
The signal and data in the world of portion；Output unit realizes the output of system handling result；Hidden unit is to be in output and input unit
Between, it is impossible to the unit observed by its exterior.Interneuronal connection weight reflects the bonding strength between unit, information
Represent and processing is embodied in the connection relation of network processing unit.
The present invention is to obtain characteristic kinematic gesture model, and the characteristic kinematic gesture that will learn by neural network learning
Model is installed in smart mobile phone by the form of mobile phone A pp, and when the hand of wearable motion bracelet acts, sensor starts
Data are obtained, and the data got are sent to mobile phone by bluetooth, whether which is identified by computation model
Meet trained characteristic action, and then judge that user is intended to.When the action meets characteristic action, judge that user has use
The intention of mobile phone, and light mobile phone screen.Using neutral net and combine Intelligent bracelet, smart mobile phone common capture user meaning
The method of figure.
Two, technical solutions
According to above-mentioned principle, technical scheme includes as follows：
(1) characteristic action model training step：
(1.1) bracelet is connected with computer, mobile phone is kept flat, user takes mobile phone to routine use position by use habit
Put, bracelet gravity sensor data during record is taken, the data of acquisition are stored in computer end, sensor is with 100hz frequencies
Rate refresh data, is repeated mobile phone of taking and acts and preserve data, to obtain feature gesture data；
(1.2) it is equal length by the feature gesture data cutout of acquisition, and stores and arrive two-dimensional matrix G={ g1,...,
gt,...,gTIn, gtThe t times data that action obtains of taking wherein is represented, t ∈ [1, T], T represent the total of mobile phone action that take
Number；
(1.3) create the input layer with m input unit, the hidden layer of n hidden unit, p output unit it is defeated
Go out three layers of reverse transmittance nerve network of layer, and two-dimensional matrix G is input in the neutral net and is trained, obtain characteristic action
Model W；
(2) smart mobile phone controls：
(2.1) characteristic action model W is installed on mobile phone, and bracelet and mobile phone is passed through into bluetooth connection；
(2.2) when wear bracelet hand occur feature take action when, bracelet start recording gravity sensor data until
Get m data, by the m data mobile phone is sent to by bluetooth and be input in the input layer of characteristic action model W into
Row calculates, and obtains the output matrix of output layer：R=(y1,...,yk,...,yp), wherein ykRepresent that output layer exports list k-th
The output of member；
(2.3) bracelet motion gesture feature similarity factor is calculatedWherein p is the number of output layer unit；
(2.4) mobile phone acceleration is calculated：Wherein Δ a, Δ b, Δ c add for mobile phone
The velocity sensor increment of component in time interval t on rectangular coordinate system in space, t obtain acceleration sensing twice for mobile phone
The time interval of device data；
(2.4) motion gesture feature similarity coefficient cut-off q is set according to experiment1=0.75, if mobile phone acceleration rate threshold q2=
120, when meeting α ＞ q1, h ＞ q2, and mobile phone is when putting out screen state, to light screen.
The invention has the advantages that：
First, add the function of smart mobile phone.
The present invention is taken action by application neutral net come feature when learning user's routine use mobile phone, is obtained with this
Characteristic action model is taken, identifies whether user has the intention using mobile phone according to characteristic action model, mobile phone moves when taking
Do not meet user's routine use mobile phone take action when, represent that simply mobile phone of taking is not intended to use mobile phone user,
Mobile phone does not light screen, when user take mobile phone action meet routine use mobile phone take action when, represent user
Take away and mobile phone and want to use mobile phone, if mobile phone screen is OFF state, mobile phone automatically turns on the screen, as a result, intelligence
The use that mobile phone is intended to become energy initiative recognition user from the use for being unable to initiative recognition user is intended to, and adds intelligent hand
The function of machine.
Second, reduce the abrasion of mobile phone secondary or physical bond
For the present invention when user wants to use mobile phone, mobile phone only need to be taken routine use position by user by custom, no
Need to click on secondary or physical bond to activate screen, mobile phone energy automatic identification is using being intended to and light screen, so as to reduce mobile phone physics
The abrasion of key, brings conveniently to daily life.
Brief description of the drawings
Fig. 1 realizes flow chart for the present invention's；
Fig. 2 is characteristic action model training sub-process figure in the present invention.
Fig. 3 is neural metwork training sub-process figure in the present invention
Embodiment
With reference to embodiment and attached drawing, further detailed description, but embodiments of the present invention are carried out to the present invention
Not limited to this.
The smart mobile phone that this experiment uses is the smart mobile phone of carrying Android 4.4.4 operating systems and bracelet is to carry Android
4.4.4 the Intelligent bracelet of operating system, under Intel (R) Core (TM) 2Duo CPU 2.3GHz win7 systems
Completed on eclipse4.2.0 platforms.
With reference to Fig. 1, step is as follows for of the invention realizing：
Step 1, characteristic action model W is obtained
It is as follows with reference to Fig. 2, the realization of this step：
1a) obtain characteristic action data
Bracelet is connected with computer, mobile phone is kept flat, user takes mobile phone to routine use position by use habit, remembers
The data of acquisition are stored in computer end, sensor is refreshed with 100hz frequencies by bracelet gravity sensor data during record is taken
Data, are repeated 20 mobile phones of taking and act and preserve data, obtain 20 groups of characteristic action data；
Data processing 1b) is carried out to characteristic action
By 128 before every group of interception of 20 groups of feature gesture data of acquisition, it is 20* that the data after interception are stored in size
In 128 two-dimensional matrix G；
1c) training characteristics action model
With reference to Fig. 3, this example carries out characteristic action model training as follows：
1c1) create an input layer with 128 input units, the hidden layer of 16 hidden units, 4 output lists
Three layers of reverse transmittance nerve network of output layer of member, Studying factors η=0.5 of three layers of reverse transmittance nerve network of setting and expectation
Output valve f=[1,1,1,1], sets error precision ε=0.5, setting input layer to hidden layer and hidden layer to output layer
Initial connection weight is ω between each unit, and initialization ω is the random floating point more than -0.05 less than 0.05；
1c2) using the data for action of taking every time in two-dimensional matrix G it is corresponding with each node of network input layer after as
The input value s of the neutral net, makes input value s try to achieve output layer according to connection weight ω and input value s along network forward-propagating
The real output value r of node；
Error amount α 1c3) is tried to achieve according to real output value r and desired output f, if α≤ε, terminates training, obtains feature
Action model W, otherwise returns to 1c2)；
1c4) make error amount α along network backpropagation, the gradient drop-out value β of error amount α is tried to achieve according to gradient descent method；
1c5) the gradient drop-out value β that error amount is multiplied by with Studying factors η obtains the variation delta ω of connection weight ω, with even
The connection weight after weights ω is updated plus the variation delta ω of connection weight is connect, returns to 1c2).
Step 2, by Bluetooth pairing mobile phone and bracelet, that is, Bluetooth of mobile phone and bracelet bluetooth are opened, carries out Bluetooth pairing.
Step 3, bracelet makes bracelet monitor gravity sensor on backstage by background service.
Step 4, bracelet data are recorded：When there is taking the action of mobile phone in the hand for wearing bracelet, bracelet start recording weight
Force sensor data is until get 128 data；
Step 5, characteristic action model W output matrixes are calculated
Characteristic action model W is stored in mobile phone, 128 data that step 4 is obtained send mobile phone to by bluetooth,
And this 128 data are input in the input layer of characteristic action model W and are calculated, obtain characteristic action model output layer
Output matrix：P=[y1,y2,y3,y4], wherein y1,y2,y3,y44 output units of character pair action model output layer it is defeated
Go out；
Step 6, bracelet motion gesture feature similarity factor is calculated.
Bracelet motion gesture feature similarity factor is calculated according to output matrix P：According to reality
Test, if motion gesture feature similarity coefficient cut-off q1=0.75, by motion gesture feature similarity factor λ and q1Compare size：
If λ ＜ q1Represent that user's process of taking does not meet the action process of taking of routine use mobile phone, then return to step
Rapid 3,
If λ >=q1, then 7 are entered step.
Step 7, mobile phone acceleration is calculated：
The data obtained according to mobile phone acceleration sensor calculate mobile phone acceleration：
Wherein, Δ a is increment of the acceleration transducer in time t in mobile phone screen horizontal direction,
Δ b is increment of the acceleration transducer in time t on mobile phone screen vertical direction,
Δ c is increment of the acceleration transducer in time t on vertical mobile phone screen orientation.
Step 8, screen is lighted
According to experiment, if mobile phone acceleration rate threshold q2=120, step 7 is calculated to mobile phone the acceleration h and q of gained2Compare
Size：
If mobile phone acceleration h ＜ q2, represent that user does not take mobile phone and use position to daily, then return to step
Rapid 3,
If mobile phone acceleration h >=q2, then screen is lighted.
Above description is only example of the present invention, does not form any limitation of the invention.Obviously for this
, all may be without departing substantially from inventive principle, the feelings of structure after present invention and principle has been understood for the professional in field
Under condition, the various modifications and variations in form and details are carried out, but these modifications and variations based on inventive concept still exist
Within the claims of the present invention.
Claims (3)
1. the method for catching user's intention using smart mobile phone and Intelligent bracelet, it is characterised in that including：
(1) characteristic action model training step：
(1.1) bracelet being connected with computer, mobile phone is kept flat, user takes mobile phone to routine use position by use habit,
The data of acquisition are stored in computer end, sensor is with 100hz frequency brushes by bracelet gravity sensor data during record is taken
New data, is repeated mobile phone of taking and acts and preserve data, to obtain characteristic action data；
(1.2) it is equal length by the feature gesture data cutout of acquisition, and stores and arrive two-dimensional matrix G={ g1,...,gΛ,...,
gTIn, wherein gΛThe Λ times data that action obtains of taking is represented, Λ ∈ [1, T], T represent the total degree for mobile phone action of taking；
(1.3) input layer with m input unit, the hidden layer of n hidden unit, the output layer of p output unit are created
Three layers of reverse transmittance nerve network, and two-dimensional matrix G is input in the neutral net and is trained, obtain characteristic action model
W；
(2) smart mobile phone controls：
(2.1) characteristic action model W is installed on mobile phone, and bracelet and mobile phone is passed through into bluetooth connection；
(2.2) bracelet monitor gravity sensor, when wear bracelet hand occur feature take action when, bracelet start recording gravity
Sensing data sends the m data to mobile phone by bluetooth and is input to characteristic action model up to getting m data
Calculated in the input layer of W, obtain the output matrix of output layer：R=(y1,...,yk,...,yp), wherein ykRepresent output
The output of k-th of output unit of layer；
(2.3) bracelet motion gesture feature similarity factor is calculatedWherein p is the number of output layer unit；
(2.4) mobile phone acceleration is calculated：Wherein Δ a, Δ b, Δ c pass for mobile phone acceleration
The sensor increment of component in time interval t on rectangular coordinate system in space, t obtain acceleration transducer data twice for mobile phone
Time interval；
(2.4) motion gesture feature similarity coefficient cut-off q is set according to experiment1=0.75, if mobile phone acceleration rate threshold q2=120, when
Meet λ ＞ q1, h ＞ q2, and mobile phone is when putting out screen state, to light screen.
2. the method according to claim 1 for catching user's intention using smart mobile phone and Intelligent bracelet, wherein step
(1.3) three layers of reverse transmittance nerve network are created in, are carried out as follows：
First, if the input unit number m=128 of input layer, the hidden unit number n=16 of hidden layer, the output of output layer
Unit number p=4,
Secondly, if input layer to hidden layer and hidden layer to initial connection weight between each unit of output layer be ω, initially
It is the random floating point more than -0.05 less than 0.05 to change ω,
Finally, by each unit of all units and hidden layer of input layer with connection weight ω connections, by the institute of hidden layer
There are each of unit and output layer unit with connection weight ω connections, obtain three layers of reverse transmittance nerve network.
3. the method according to claim 1 for catching user's intention using smart mobile phone and Intelligent bracelet, wherein step
(1.3) to the training of neutral net, carry out as follows：
1.3a) Studying factors η=0.5 of three layers of reverse transmittance nerve network of setting and desired output f=[1,1,1,1], if
Error precision ε=0.5, setting input layer to hidden layer and hidden layer are determined to initial connection weight between each unit of output layer
For ω, initialization ω is the random floating point more than -0.05 less than 0.05；
1.3b) this is used as afterwards using the data for action of taking every time in two-dimensional matrix G are corresponding with each node of network input layer
The input value s of neutral net, makes input value s try to achieve output layer section according to connection weight ω and input value s along network forward-propagating
The real output value r of point；
Error amount α 1.3c) is tried to achieve according to real output value r and desired output f, if α≤ε, terminates training, otherwise enters
1.3d)；
1.3d) make error amount α along network backpropagation, the gradient drop-out value β of error amount α is tried to achieve according to gradient descent method；
1.3e) the gradient drop-out value β that error amount is multiplied by with Studying factors η obtains the ω variation delta ω of connection weight, with connection
Weights ω adds the variation delta ω of connection weight, and the connection weight after being updated, returns to 1.3b).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510765304.1A CN105426961B (en) | 2015-11-11 | 2015-11-11 | The method for catching user's intention using Intelligent bracelet and smart mobile phone |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510765304.1A CN105426961B (en) | 2015-11-11 | 2015-11-11 | The method for catching user's intention using Intelligent bracelet and smart mobile phone |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105426961A CN105426961A (en) | 2016-03-23 |
| CN105426961B true CN105426961B (en) | 2018-04-17 |
Family
ID=55505156
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510765304.1A Active CN105426961B (en) | 2015-11-11 | 2015-11-11 | The method for catching user's intention using Intelligent bracelet and smart mobile phone |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105426961B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107438808B (en) * | 2016-10-31 | 2021-05-04 | 深圳市大疆创新科技有限公司 | Method and device for controlling rod amount and related equipment |
| CN108431759A (en) * | 2016-11-21 | 2018-08-21 | 深圳市柔宇科技有限公司 | Electronic device and its display control method |
| CN106354272A (en) * | 2016-11-29 | 2017-01-25 | 广东小天才科技有限公司 | Unlocking method and system for intelligent wearable equipment |
| CN108881627A (en) * | 2018-06-06 | 2018-11-23 | Oppo（重庆）智能科技有限公司 | Screen control method, device, terminal and storage medium |
| CN109346166B (en) * | 2018-11-22 | 2022-04-05 | 广东顺德奥为德科技有限公司 | Deep learning modeling method of intelligent medical bracelet for hospital department |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103124314A (en) * | 2013-03-07 | 2013-05-29 | 上海斐讯数据通信技术有限公司 | Screen switching achieving method and mobile terminal |
| JP2014057129A (en) * | 2012-09-11 | 2014-03-27 | Casio Comput Co Ltd | Mobile device, control method of mobile device, and control program of mobile device |
| CN103927117A (en) * | 2014-03-26 | 2014-07-16 | 广东明创软件科技有限公司 | Method for lighting up screen rapidly and mobile terminal of screen |
| CN104731490A (en) * | 2015-04-20 | 2015-06-24 | 深圳市欧珀通信软件有限公司 | Screen control method and screen control device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013044399A1 (en) * | 2011-09-29 | 2013-04-04 | Swisscom Ag | Automatic flight mode |
-
2015
- 2015-11-11 CN CN201510765304.1A patent/CN105426961B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2014057129A (en) * | 2012-09-11 | 2014-03-27 | Casio Comput Co Ltd | Mobile device, control method of mobile device, and control program of mobile device |
| CN103124314A (en) * | 2013-03-07 | 2013-05-29 | 上海斐讯数据通信技术有限公司 | Screen switching achieving method and mobile terminal |
| CN103927117A (en) * | 2014-03-26 | 2014-07-16 | 广东明创软件科技有限公司 | Method for lighting up screen rapidly and mobile terminal of screen |
| CN104731490A (en) * | 2015-04-20 | 2015-06-24 | 深圳市欧珀通信软件有限公司 | Screen control method and screen control device |
Non-Patent Citations (3)
| Title |
|---|
| Using mobile phones to determine transportation modes;Ｒeddy S etal.;《ACM Transactions on Sensor Networks》;20101231;第6卷(第2期);第13-27页 * |
| 坐标转换在移动用户行为识别中的应用;王忠民 等;《北京邮电大学学报》;20140430;第37卷;第30-34页 * |
| 基于加速度传感器的智能终端手势识别关键技术研究;陈文;《中国优秀硕士学位论文全文数据库 信息科技辑》;20120715 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105426961A (en) | 2016-03-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105426961B (en) | The method for catching user's intention using Intelligent bracelet and smart mobile phone | |
| CN110009052B (en) | Image recognition method, image recognition model training method and device | |
| Laput et al. | Sensing fine-grained hand activity with smartwatches | |
| WO2021082749A1 (en) | Action identification method based on artificial intelligence and related apparatus | |
| Singh et al. | A deeply coupled ConvNet for human activity recognition using dynamic and RGB images | |
| CN108062170A (en) | Multi-class human posture recognition method based on convolutional neural networks and intelligent terminal | |
| WO2020207249A1 (en) | Notification message pushing method and apparatus, and storage medium and electronic device | |
| CN107016342A (en) | A kind of action identification method and system | |
| KR101189078B1 (en) | User identification apparatus and method with means for sensing touch input | |
| WO2021136131A1 (en) | Information recommendation method and related device | |
| CN108683724A (en) | A kind of intelligence children's safety and gait health monitoring system | |
| CN107832784A (en) | A kind of method of image beautification and a kind of mobile terminal | |
| CN106648068A (en) | Method for recognizing three-dimensional dynamic gesture by two hands | |
| CN107996024A (en) | Independent applies safety management | |
| CN109951363A (en) | Data processing method, apparatus and system | |
| CN114450690A (en) | Network model training method, image attribute identification method and device and electronic equipment | |
| KR20200080418A (en) | Terminla and operating method thereof | |
| Tan et al. | Online activity recognition and daily habit modeling for solitary elderly through indoor position-based stigmergy | |
| Raj et al. | Different techniques for human activity recognition | |
| CN110580055B (en) | Action track identification method and mobile terminal | |
| Ata et al. | A robust optimized convolutional neural network model for human activity recognition using sensing devices | |
| Jhang et al. | Sensor based dynamic hand gesture recognition by PairNet | |
| CN111797289A (en) | Model processing method and device, storage medium and electronic equipment | |
| CN111797867A (en) | System resource optimization method and device, storage medium and electronic equipment | |
| KR102415837B1 (en) | A method for estimating costs of cable-pulliing based on artificial neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |