CN108988968A - Human behavior detection method, device and terminal device - Google Patents
Human behavior detection method, device and terminal device Download PDFInfo
- Publication number
- CN108988968A CN108988968A CN201810842914.0A CN201810842914A CN108988968A CN 108988968 A CN108988968 A CN 108988968A CN 201810842914 A CN201810842914 A CN 201810842914A CN 108988968 A CN108988968 A CN 108988968A
- Authority
- CN
- China
- Prior art keywords
- human body
- time series
- information
- csi time
- csi
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Electromagnetism (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The present invention is suitable for behavioral value technical field, provides a kind of human behavior detection method, device and terminal device.The described method includes: receiving the wireless signal reflected by human body, and obtain the CSI time series of the wireless signal；The CSI time series is pre-processed, and feature extraction is carried out to the pretreated CSI time series, obtains human body behavioural characteristic；The human body behavioural characteristic is input in classifier and is classified, determines human body behavior according to classification results.The present invention can be avoided present in traditional human body behavioral value method that time stability is poor, can not separate the defects detection human body behavior of multi-path signal, control layer information realization human testing is replaced with the information of physical layer, improves the precision and accuracy rate of human body behavioral value.
Description
Technical field
The invention belongs to behavioral value technical fields more particularly to a kind of human behavior detection method, device and terminal to set
It is standby.
Background technique
Human behavior detection technique wide range of services is in the fields such as medical treatment & health and wisdom security protection.It is right in medical treatment & health field
The daily behavior of old man and child are monitored, and are reduced and are waited hazardous acts to damage personnel because falling down.It is led in wisdom security protection
Domain carries out real-time behavior monitoring to the personnel within the scope of wireless network coverage area, and note abnormalities behavior, and system is obtained by alarm
Real-time Feedback is obtained, to achieve the purpose that security protection.
Traditional human behavior detection technique is based primarily upon sensor, vision and wearable device to realize.Based on sensing
Personnel's detection technique of device needs to dispose big quantity sensor in designated place, and mobility is poor, and maintenance cost is relatively high；Base
It is high in requirement of the detection technique to light of camera, it fails in dark conditions, detection accuracy is low；It is set based on wearable
Standby detection technique requires user's body-worn monitoring device, increases burden for users, and comfort is poor, in addition, the technology for
The consciousness requirement of target person wearable device is high, is not suitable for the special screnes such as intrusion detection, suspect's tracking.
Later, with the development of wireless network, all standing is done step-by-step in wireless network.Wireless network signal is via mobile human body etc.
Reflection, scattering etc. can occur when barrier and form multipath superposed signal, cause wireless signal to generate by analyzing human body behavior
Corresponding change can carry out mobile personnel detection.Existing wireless behavioral value technology is by receiving wireless signal and utilizing reception
Signal strength instruction detects human body behavior, however, the signal in this detection mode is the Overlay of multipath transmisstion,
A plurality of signal propagation path cannot be distinguished one by one, can only carry out the behavioral value of coarse grain information, and time stability is poor, so that
The accuracy of human body behavioral value is low.
Summary of the invention
In view of this, the embodiment of the invention provides a kind of human behavior detection method, device and terminal device, to solve
Human body behavioral value method based on signal strength in the prior art, the time stability of signal is poor, and is folded using multipath
Plus signal detection, so that the problem that human body behavioral value rate is low.
The first aspect of the embodiment of the present invention provides a kind of human behavior detection method, comprising:
The wireless signal reflected by human body is received, and obtains CSI (the Channel State of the wireless signal
Information, channel state information) time series；
The CSI time series is pre-processed, and feature is carried out to the pretreated CSI time series and is mentioned
It takes, obtains human body behavioural characteristic；
Classified by classifier to the human body behavioural characteristic, determines human body behavior according to classification results.
Optionally, the CSI time series includes temporal information and channel state information；
The CSI time series for obtaining the wireless signal specifically includes:
The channel state information in the CSI time series is obtained using orthogonal frequency division multiplexing method；
The channel state information includes multiple subcarrier informations.
Optionally, it is described by the CSI time series carry out pretreatment specifically include:
The Information abnormity value in the CSI time series is detected, the Information abnormity value is deleted；
The position that the Information abnormity value is deleted in the CSI time series carries out interpolation processing；
CSI time series after progress interpolation processing is filtered, the pretreated CSI time sequence is obtained
Column.
Optionally, the Information abnormity value in the detection CSI time series specifically includes:
Status information preset range is set；
Judge the value of information in the channel state information whether in the status information preset range；
The value of information not in the status information preset range is detected as the Information abnormity value.
Optionally, described that feature extraction is carried out to the pretreated CSI time series, obtain human body behavioural characteristic tool
Body includes:
The pretreated CSI time series is subjected to dimension-reduction treatment, and determines feature principal component and preset quantity
Main feature vector；
The first-order difference mean value for calculating each main feature vector, calculates the variance of the feature principal component, and by institute
The ratio for stating variance and the first-order difference mean value is set as characteristics of human body's value；
Characteristics of human body's value of preset quantity forms the human body behavioural characteristic.
Optionally, the first-order difference mean value is calculated to specifically include:
Wherein, N is the number of the subcarrier, viFor the corresponding main feature vector.
Optionally, after being input to the human body behavioural characteristic and being classified in classifier, further includes:
Decision is carried out to the classification results using data fusion method, human body behavior is determined according to the result of decision.
The second aspect of the embodiment of the present invention provides a kind of human behavior detection device, comprising:
Data obtaining module for receiving the wireless signal reflected by human body, and obtains the CSI time of the wireless signal
Sequence；
Characteristic extracting module, for pre-processing the CSI time series, and when to the pretreated CSI
Between sequence carry out feature extraction, obtain human body behavioural characteristic；
Activity recognition module is classified for the human body behavioural characteristic to be input in classifier, is tied according to classification
Fruit determines human body behavior.
The third aspect of the embodiment of the present invention provides a kind of terminal device, including memory, processor and is stored in
In the memory and the computer program that can run on the processor, which is characterized in that described in the processor executes
It realizes when computer program such as the step of any of the above-described the method.
The fourth aspect of the embodiment of the present invention provides a kind of computer readable storage medium, the computer-readable storage
Media storage has computer program, comprising: the side as described in any of the above-described is realized when the computer program is executed by processor
The step of method.
Human body behavioral value method, apparatus and terminal device in the embodiment of the present invention are existing compared with prior art to be had
Beneficial effect is: firstly, by receiving the wireless signal reflected by human body, and the CSI time series of the wireless signal is obtained, it uses
The CSI time series of physical layer replaces the coarse grain informations such as the signal strength of control layer, mentions for subsequent feature extraction and detection
For more fine-grained information, human body behavioral value precision is improved；Secondly, the CSI time series is pre-processed, reduce
Noise in CSI time series improves the accuracy that CSI time series carries out feature extraction, further increases human body behavior inspection
The accuracy rate of survey；Finally, being classified by the way that the human body behavioural characteristic to be input in classifier, determined according to classification results
Human body behavior further improves the precision and accuracy rate of human body pedestrian detection.
Detailed description of the invention
It to describe the technical solutions in the embodiments of the present invention more clearly, below will be to embodiment or description of the prior art
Needed in attached drawing be briefly described, it should be apparent that, the accompanying drawings in the following description is only of the invention some
Embodiment for those of ordinary skill in the art without any creative labor, can also be according to these
Attached drawing obtains other attached drawings.
Fig. 1 is the implementation process schematic diagram of human behavior detection method provided in an embodiment of the present invention；
Fig. 2 is the implementation process schematic diagram of step S102 in Fig. 1；
Fig. 3 is the implementation process schematic diagram of step S201 in Fig. 2；
Fig. 4 is another implementation process schematic diagram of step S102 in Fig. 1；
Fig. 5 is a kind of implement scene figure provided in an embodiment of the present invention；
Fig. 6 is the schematic diagram of the channel state information variation provided in an embodiment of the present invention under unmanned scene；
Fig. 7 is the schematic diagram of the channel state information variation provided in an embodiment of the present invention under someone's mobile context；
Fig. 8 is the structural block diagram of human behavior detection device provided in an embodiment of the present invention；
Fig. 9 is the schematic diagram of terminal device provided in an embodiment of the present invention.
Specific embodiment
In being described below, for illustration and not for limitation, the tool of such as particular system structure, technology etc is proposed
Body details, to understand thoroughly the embodiment of the present invention.However, it will be clear to one skilled in the art that there is no these specific
The present invention also may be implemented in the other embodiments of details.In other situations, it omits to well-known system, device, electricity
The detailed description of road and method, in case unnecessary details interferes description of the invention.
In order to illustrate technical solutions according to the invention, the following is a description of specific embodiments.
Embodiment one
One embodiment implementation process schematic diagram of human behavior detection method is provided referring to Fig. 1, details are as follows:
Step S101 receives the wireless signal reflected by human body, and obtains the CSI time series of the wireless signal.
Optionally, the present embodiment uses MIMO (Multiple-Input Multiple-Output, multiple-input and multiple-output)
Technology, i.e. wireless signal transmitting terminal and receiving end use multiple transmitting antennas and multiple receiving antennas respectively, available multiple
Wireless signal, it can the CSI time series of the comprehensive received wireless signal of acquisition, further in CSI time series
Channel state information signature analysis reduces human body behavior False Rate, improves detection accuracy.
Optionally, the CSI time series is obtained using Atheros series network interface card or Intel5300 network interface card.
Optionally, the CSI time series may include temporal information and channel state information.
The channel state information refers to the channel state information of wireless signal physical layer, can be expressed as CSI, main to wrap
Include the frequency domain information of signal, such as amplitude and phase etc..Wherein, temporally information carries out rearranging CSI channel state information
Time series.
In practical application, signal receiver can receive multiple data packets, each data packet when receiving wireless signal
To include temporal information and channel state information, wherein the channel state information in data packet can be arranged according to temporal information
Column form the CSI time series.
Optionally, it can detecte out whether received wireless signal has loss of data phenomenon according to temporal information, such as lose
Packet phenomenon, the present embodiment can also carry out interpolation processing to the data for packet loss occur, guarantee the correctness of information.
Under a concrete application scene, wireless signal, wireless signal receiver pair are emitted using wireless signal transmitter
The signal of human body reflection is received.As shown in figure 5, wireless transmitter emits wireless signal, pass through accessible through path
And wireless signal receiver is reached by the reflection path of target person, wherein through path is sighting distance (Line of
Sight, LOS) path, reflection path is the path non line of sight (Non Line of Sight, NLOS).Target person is moved through
Path changes in journey, and corresponding change occurs for channel state information.The present embodiment be by detected wireless signals transmitter and
The channel state information of signal between wireless signal receiver detects in scene whether to have mobile personnel.
It is the variation schematic diagram of unmanned scene lower channel status information referring to Fig. 6 and Fig. 7, Fig. 6, Fig. 7 is under someone's scene
The schematic diagram of channel state information variation.As can be seen from Figures 6 and 7, channel state information is almost unchanged in unmanned situation,
And when someone walks about in scene, channel state information amplitude of variation obviously becomes larger.
The signal strength information that traditional media access control layer is replaced using the channel state information of physical layer, can solve
Certainly time stability is poor in conventional method, can not separate the defects of multipath signal, realizes the human body behavioral value of higher precision.
In another embodiment, the CSI time series for obtaining the wireless signal specifically includes:
The channel state information in the CSI time series is obtained using orthogonal frequency division multiplexing method.
Specifically, orthogonal frequency division multiplexing method mainly breaks a channel into several orthogonal sub-channels, by high-speed data signal
It is converted into parallel low speed sub-data flow, is modulated to and is transmitted on each of the sub-channels, it can obtains the CSI time sequence
Channel state information and corresponding temporal information in column, refine the Information Granularity of acquisition, improve the accuracy of human bioequivalence.
Optionally, the channel state information includes multiple subcarrier informations.
Specifically, the received wireless signal may include multiple subcarriers, the channel shape in each CSI time series
State information may include multiple subcarrier informations.The number of the subcarrier determines by the bandwidth of current receiving device, for example,
The signal receiver of 20MHZ can receive the wireless signal including 56 subcarriers, and the signal receiver of 40MHZ can connect
Packet receiving includes the wireless signal of 114 subcarriers.
Optionally, the channel state information includes the amplitude information of multiple subcarriers.
Specifically, temporal information refers to the time series of the amplitude information of subcarrier；Amplitude information refers to corresponding subcarrier
Amplitude.
Optionally, the channel state information can also include: the phase information of multiple subcarriers.
The present embodiment can also according to the phase informations of multiple subcarriers and temporal information as a CSI time series into
Row step S102 to S103, the signal reflected from the direction of the phase information of signal carrier human body is analyzed, such as feature
Extraction, feature training and tagsort, further determine that human body behavior, improve human body behavioral value precision and accuracy.
Optionally, the channel state information can also include: the phase information and amplitude information of multiple subcarriers.
The present embodiment can also be according to the phase information of multiple subcarriers, amplitude information and temporal information as one
CSI time series carries out step S102 to S103, and human body is reflected in the direction of phase information and amplitude information from signal carrier
Signal analyzed, such as feature extraction, feature training and tagsort further determine that human body behavior, improve human body row
For detection accuracy and accuracy.
Step S102 pre-processes the CSI time series, and to the pretreated CSI time series into
Row feature extraction obtains human body behavioural characteristic.
Due to the interference inside the variation and equipment of outside environmental elements, collected original CSI time sequence will lead to
Column include much noise and abnormal point, cannot be used directly for mobile personnel detection, also can be to human body behavioral value precision and accurate
Degree impacts.So the present embodiment first pre-processes the CSI time series of acquisition, the abnormal point in information is removed
With noise etc., then feature extraction is carried out, and then the feature extracted is more representative, and then improves human body behavioral value precision.
Referring to fig. 2, in another embodiment, the CSI time series is carried out in step S102 pretreated specific
Realization process is as follows:
Step S201 detects the Information abnormity value in the CSI time series, and the Information abnormity value is deleted.
In practical application, due to signal receiver or external environment problem, so that the CSI time of the wireless signal obtained
Channel state information in sequence influences the accuracy extracted to human body behavioural characteristic, so the present embodiment is first there are exceptional value
The Information abnormity value in the CSI time series is detected, is then deleted.
Optionally, rejecting outliers can be carried out using Hampel filter.
Hampel filter can identify the position that exceptional value occurs in monitoring data, and use least square supporting vector
Machine regression model realizes monitoring CSI time sequence based on the method for recursive prediction for the exceptional value in detection CSI time series
The analysis processing of Information abnormity value in column.
Step S202, the position that the Information abnormity value is deleted in the CSI time series carry out interpolation processing.
In practical application, after the exceptional value of certain position in CSI time series is deleted, it may cause information and lose or lack
Phenomenon is lost, the corresponding relationship of channel state information and temporal information can be upset, and then will affect entire CSI time series.Therefore,
The present embodiment also needs the position for deleting Information abnormity value in CSI time series to carry out interpolation processing, and missing data is supplemented,
Guarantee the accuracy of the information sequence of CSI time series.
In practical application, multiple data packets of wireless signal can be received when receiving wireless signal simultaneously, but due to communicating matter
The problem of amount, can be there is a phenomenon where data-bag lost when receiving wireless signal, and data packet when wireless signal is transmitted
All it is continuously, if data packet therein is lost, the transmission of the CSI time series in data packet just will appear sky
Packet loss is caused in hole.Optionally, so the present embodiment carries out interpolation processing for packet loss phenomenon, guarantee the transmission of CSI time series
Quality, and then improve the accuracy of human body behavioral value.Wherein, the data packet includes CSI time series.
Optionally, the method for carrying out interpolation processing can be handled using Newton interpolating method.
Newton interpolating method calculate it is simpler, i.e., when increasing additional interpolation point, operation result before can use with
Operand is reduced, Lagrange's interpolation is better than, it can accelerates to carry out pretreated speed, Jin Erjia to CSI time series
Fast human body behavioral value speed.
Step S203 is filtered the CSI time series after progress interpolation processing, obtains pretreated described
CSI time series.
In the field of wireless communication, wireless signal is highly susceptible to variation and the signal receiver of outside environmental elements
Internal interference, and then cause the CSI time series of collected wireless signal containing much noise, it cannot be used directly for mobile human
Member's detection.
Optionally, the method being filtered to the CSI time series after progress interpolation processing can be gone using small echo
It makes an uproar method.Specifically, Wavelet-denoising Method mainly includes two methods of wavelet transformation and wavelet threshold processing, wavelet transformation is to noisy
Acoustic intelligence carries out wavelet transformation, and the wavelet coefficient obtained to transformation is handled, and noise wherein included is removed, finally to processing
Wavelet coefficient afterwards carries out wavelet inverse transformation, the information after being denoised；Wavelet threshold processing is first to carry out small wavelength-division to information
Solution, obtains scale coefficient, then carries out threshold process to scale coefficient, finally carries out the data after wavelet reconstruction is denoised again
Information.The data information noise of Wavelet-denoising Method processing is low, denoises accurate information.
Optionally, using wavelet threshold processing method when being filtered to the CSI time series after progress interpolation processing.Example
Such as, CSI time series first can be subjected to two-layer decomposition according to wavelet function, is then carried out using unbiased possibility predication threshold method
Denoising, is finally again reconstructed information, and the CSI time series after denoise is to get to the pretreated CSI time
Sequence.
Referring to Fig. 3, in another embodiment, the Information abnormity value in the CSI time series is detected in step S201
Specific implementation process include:
Status information preset range is arranged in step S301.
Optionally, setting status information preset range may include:
The median μ and median absolute deviation σ for calculating the channel state information set section [+3 σ of μ -3 σ, μ] to
The status information preset range.
Wherein, the calculating process of median absolute deviation may include: first to seek the median of the value of information in channel state information,
Then all values of information are subtracted into all absolute values that median obtains and forms one channel status sample, then seek channel status
The median of sample, the median of channel status sample are median absolute deviation.
For example, channel state information X={ 2387964 }, median 6, each letter in channel state information
Breath value subtracts median and seeks absolute value, and the channel status sample that all absolute values form one is { 4321302 }, letter
The median of road state sample is 2, so the median absolute deviation of channel state information X={ 2387964 } is 2.
Whether step S302 judges the value of information in the channel state information in the status information preset range.
Optionally, judge each value of information in the channel state information whether status information preset range [μ -3 σ,
+ 3 σ of μ] in.Wherein, μ is the median of temporal information, and σ is the median absolute deviation of temporal information.
The value of information not in the status information preset range is detected as the Information abnormity by step S303
Value.
Optionally, each value of information is judged whether in status information preset range [+3 σ of μ -3 σ, μ], by letter of not being at state
The value of information in breath preset range [+3 σ of μ -3 σ, μ] is determined as Information abnormity value, and is deleted.
Referring to fig. 4, in another embodiment, the pretreated CSI time series is carried out in step S102 special
Sign is extracted, and the detailed process for obtaining human body behavioural characteristic includes:
The pretreated CSI time series is carried out dimension-reduction treatment by step S401, and determines feature principal component and pre-
If the main feature vector of quantity.
In practical application, the dimension of the CSI time series of received wireless signal is very big, and information dimension ambassador must calculate multiple
It is miscellaneous, and influence human body behavioral value speed.To reduce computation complexity, CSI time series is first carried out dimensionality reduction by the present embodiment
Processing.
Optionally, the method for carrying out dimensionality reduction to CSI time series can carry out dimensionality reduction using Principal Component Analysis, or adopt
With core principle component analysis method dimensionality reduction.
Illustratively, pretreated CSI time series is carried out Data Dimensionality Reduction by Principal Component Analysis.Specifically, first right
Pretreated CSI time series is normalized, and obtains normalization CSI time series, then normalizes CSI time sequence
Column carry out autocorrelation calculation and obtain correlation matrix, and obtain preceding K maximum characteristic values and corresponding feature vector, and K most
The big corresponding feature vector of characteristic value is the default main feature vector measured in advance in the present embodiment；According to K main feature vectors and
Normalization CSI time series seeks corresponding principal component matrix, that is, determines the feature principal component, realizes Data Dimensionality Reduction.Wherein,
K is the preset quantity.
K maximum characteristic values may include: to calculate the characteristic value of correlation matrix before obtaining, by each characteristic value carry out from
Small arrangement is arrived greatly, K maximum characteristic values before obtaining, while obtaining corresponding feature vector, i.e. K maximum characteristic values pair
The feature vector answered is the main feature vector of preset quantity.Wherein, K is less than the matrix columns of CSI time series.
Optionally, K value (preset quantity) can be inputted by user and be set, can also be by tribute of the characteristic value in global feature
Rate decision is offered, the corresponding feature vector of characteristic value for having more than default contribution rate is then used as the main feature vector of preset quantity.
For example, the CSI time series after normalization is information matrix H.Wherein, each column of information matrix H represent single
The range value that subcarrier changes over time, that is, each column of information matrix H represent the width of single sub-carrier information change at any time
Angle value；Every a line represents different sub-carrier in the range value of synchronization.The columns n of information matrix H is equal to the quantity of subcarrier,
The line number m of information matrix H is equal to the length of temporal information.For example, signal receiver uses the bandwidth of 40MHz in 5G frequency range,
The CSI time series that signal receiver receives a wireless signal, which can be, collects 50 data each second, obtains 10 seconds
Data, then m=10*50, n=114.Correspondingly, the dimension of information matrix H is exactly 500*114.
Autocorrelation operation is carried out to information matrix H, acquires correlation matrix, i.e. correlation matrix C=HT× H, wherein HTIt is information
The transposed matrix of matrix H, the dimension of correlation matrix C are n × n.Eigenvalues Decomposition is carried out to correlation matrix C, K maximum before obtaining
Characteristic value and corresponding feature vector, for example, K=6.Wherein it is determined that K main feature vectors be expressed as v1, v2... ...,
vk, the length of each main feature vector is n.By information matrix H and main feature vector seek principal component matrix P (feature it is main at
Point), specifically, according to formula P=H × viObtain feature principal component, i.e., by using Principal Component Analysis by CSI time sequence
The dimension of time series is down to K dimension in column.
Step S402 calculates the first-order difference mean value of each main feature vector, calculates the side of the feature principal component
Difference, and characteristics of human body's value is set by the ratio of the variance and the first-order difference mean value.
Referring to Fig. 6 and Fig. 7, it can be seen that the channel state information in unmanned situation is almost unchanged, and when someone is on the scene
When walking about in scape, channel state information amplitude of variation obviously becomes larger.It is, principal component changes greatly, phase as someone in scene
Answer ground variance yields very big, meanwhile, main feature vector variation is gentle, and correspondingly first-order difference mean value is small；On the contrary, when nothing in scene
When people, principal component variation is small, and variance yields is small, meanwhile, main feature vector variation is random, and first-order difference mean value is big.So this implementation
Example chooses the ratio of variance and first-order difference mean value as characteristic value.
Optionally, the first-order difference mean value is calculated to specifically include:
Wherein, N is the number of the subcarrier, viFor the corresponding main feature vector.
Optionally, the variance is calculated to specifically include:
Wherein, L is the number of the subcarrier, piFor the vector of the corresponding feature principal component,For corresponding institute
State the mean value of the vector of feature principal component.
Characteristics of human body's value of step S403, preset quantity form the human body behavioural characteristic.
Specifically, the main feature vector of preset quantity is sought first-order difference mean value, and seek the feature principal component
Variance, set characteristics of human body's value for the ratio of variance and first-order difference mean value, i.e., human body behavioural characteristic is by the present count
Characteristics of human body's value of amount forms.
Step S103 classifies to the human body behavioural characteristic by classifier, determines human body row according to classification results
For.
Optionally, the classifier in the present embodiment selects BP (Error Back Propagation Training, error
Back-propagation algorithm) neural network.BP neural network is widely used in classification problem.
BP neural network mainly includes two processes of backpropagation of the propagated forward and error of signal, i.e. calculating error is defeated
It is carried out when out by from the direction for being input to output, and adjusts weight and threshold value and then carried out from the direction for being output to input.
Specifically, input feature vector acts on output node by hidden layer when propagated forward, by nonlinear transformation, produce
Raw output signal is transferred to the backpropagation of error if the classification situation of reality output is not consistent with the classification of desired output
Journey；Error-duration model is that output error is passed through to hidden layer to give all units of each layer to the layer-by-layer anti-pass of input layer, and by error distribution,
Using from the error signal that each layer obtains as adjustment each unit weight foundation, by adjusting the connection of input node and hidden node
The linking intensity and classification thresholds for connecing intensity, hidden node and output node decline error along gradient direction, i.e., according to ladder
It spends descent method and adjusts error.
Optionally, gradient descent method can be by regulating gradient decline step-length, in conjunction with adjust learning training rate,
The paces for updating each convolutional layer are different, and then improve the precision for adjusting error；Gradient descent method can also include small quantities of
It measures gradient and declines (Mini-batch gradient descent) method, update the gradient calculated according to human body behavioural characteristic every time
First-order difference mean value, according to the first-order difference mean value de-regulation error, and then improve the precision for adjusting error；In addition, terraced
Spending descent method can also include momentum (Momentum) optimization and Nesterov momentum (Nie Sijieluofu momentum) method, to reduce
Convergence is accelerated in the upheaval of gradient.
In above-mentioned BP neural network, the study of propagated forward and error-duration model to the characteristic information of input Jing Guo repeatedly is instructed
Practice, determines weight corresponding with minimal error and classification thresholds, illustrate that BP neural network model has trained, while training
Disaggregated model is more accurate.
The assorting process of the present embodiment is mainly, first to the letter under the channel status feature and unmanned scene under someone's scene
Road state feature is marked, then that the channel status feature under the channel status feature and unmanned scene under someone's scene is defeated
Enter and be trained into BP neural network, that is, the letter under the channel status feature and unmanned scene under a part of someone's scene
Road state feature is as training set, the training in BP neural network, after training, the CSI time sequence that then will obtain again
Column carry out feature extraction, and obtained people's behavioural characteristic is input to trained BP neural network as test set, trained
BP neural network can voluntarily handle the smallest letter by non-linear conversion of output error to the input information of similar sample
Breath, that is, classification results.
Specifically, obtaining the corresponding CSI time series of different ambient conditions forms training dataset.In acquisition process
In, the human body behavioural characteristic in the CSI time series of wireless signal is manually marked, that is, indicates that human body behavioural characteristic is corresponding
Which kind of ambient condition.For someone's ambient condition, available target person carries out movable CSI under different location and orientation
Time series.CSI time series under the different scenes concentrated for training data, can use above-mentioned steps S102 and step
Information preprocessing method and feature extracting method are handled in S103, are obtained corresponding human body behavioural characteristic and are directly inputted to BP
Neural network is trained.Then, mobile personnel detection is carried out using trained BP neural network as classifier.
It is above-mentioned to be classified using BP neural network to human body behavioural characteristic, that is, utilize the reversed biography of propagated forward and error
Repetition training and test are broadcast, classification results are obtained, determines human body behavior according to classification results, so that tagsort is more accurate,
Classification results are more accurate.
In another embodiment, it after being input to the human body behavioural characteristic and being classified in classifier, also wraps
It includes:
Decision is carried out to the classification results using data fusion method, human body behavior is determined according to the result of decision.
Optionally, data fusion method may include D-S evidence theory method.D-S evidence theory method has processing not
The ability for determining information, as a kind of uncertain reasoning method, the satisfaction that is mainly characterized by of D-S evidence theory compares Bayesian probability
By weaker condition, there is the ability of directly expression " uncertain " and " not knowing ".Using D-S evidence theory method to above-mentioned more
The classification results of antenna carry out decision, and the accuracy of identification of human body behavioral value can be improved, and reduce error rate.
Above-noted persons' behavioral value method first passes through the wireless signal for receiving and being reflected by human body, and obtains the wireless communication
Number CSI time series, the coarse grain informations such as the signal strength of control layer are replaced with the CSI time series of physical layer, for below
Feature extraction and detection more fine-grained information is provided, improve human body behavioral value precision；Secondly, by the CSI time
Sequence is pre-processed, and the noise in CSI time series is reduced, and improves the accuracy that CSI time series carries out feature extraction, into
The accuracy rate of one step raising human body behavioral value；Divided finally by the human body behavioural characteristic to be input in classifier
Class determines human body behavior according to classification results, further improves the precision and accuracy rate of human body pedestrian detection.
It will be understood by those skilled in the art that in above-described embodiment the size of the serial number of each step be not meant to execute it is suitable
Sequence it is successive, the execution of each process sequence should be determined by its function and internal logic, the implementation without coping with the embodiment of the present invention
Process constitutes any restriction.
Embodiment two
Corresponding to human behavior detection method described in foregoing embodiments one, shown in the embodiment of the present invention two in Fig. 8
The structural block diagram of human behavior detection device.For ease of description, only the parts related to this embodiment are shown.
The device includes: data obtaining module 110, characteristic extracting module 120 and Activity recognition module 130.
Data obtaining module 110 is used to receive by the wireless signal of human body reflection, and when obtaining the CSI of the wireless signal
Between sequence.
Characteristic extracting module 120 is used to pre-process the CSI time series, and to the pretreated CSI
Time series carries out feature extraction, obtains human body behavioural characteristic.
Activity recognition module 130 is classified for the human body behavioural characteristic to be input in classifier, according to classification
As a result human body behavior is determined.
In above-noted persons' behavioral value device, data obtaining module 110 receives the wireless signal reflected by human body, and obtains
The CSI time series of the wireless signal replaces the coarsenesses such as the signal strength of control layer to believe with the CSI time series of physical layer
Breath provides more fine-grained information for subsequent feature extraction and detection, improves human body behavioral value precision；Secondly, feature
Extraction module 120 pre-processes the CSI time series, reduces the noise in CSI time series, improves CSI time sequence
Column carry out the accuracy of feature extraction, further increase the accuracy rate of human body behavioral value；Finally, Activity recognition module 130 is logical
It crosses for the human body behavioural characteristic to be input in classifier and classify, determine human body behavior according to classification results, further mention
The high precision and accuracy rate of human body pedestrian detection.
Embodiment three
Fig. 9 is the schematic diagram for the terminal device 100 that the embodiment of the present invention three provides.As shown in figure 9, the terminal of the embodiment
Equipment 100 includes: processor 140, memory 150 and is stored in the memory 150 and can be on the processor 140
The computer program 151 of operation, such as the program of human behavior detection method.The processor 140 is executing the computer
The step in above-noted persons' behavioral value embodiment of the method is realized when program 151, such as step 101 shown in FIG. 1 is to 103.Or
Person, the processor 140 realize the function of each module/unit in above-mentioned each Installation practice when executing the computer program 151
Can, such as the function of module 110 to 130 shown in Fig. 8.
Illustratively, the computer program 151 can be divided into one or more module/units, it is one or
Multiple module/the units of person are stored in the memory 150, and are executed by the processor 140, to complete the present invention.Institute
Stating one or more module/units can be the series of computation machine program instruction section that can complete specific function, the instruction segment
For describing implementation procedure of the computer program 151 in the terminal device 100.For example, the computer program 151
Data obtaining module, characteristic extracting module and Activity recognition module can be divided into, each module concrete function is as follows:
Data obtaining module is used to receive the wireless signal reflected by human body, and obtains the CSI time of the wireless signal
Sequence.
Characteristic extracting module is used to pre-process the CSI time series, and to the pretreated CSI time
Sequence carries out feature extraction, obtains human body behavioural characteristic.
Activity recognition module is classified for the human body behavioural characteristic to be input in classifier, according to classification results
Determine human body behavior.
The terminal device 100 can be the calculating such as desktop PC, notebook, palm PC and cloud server and set
It is standby.The terminal device 100 may include, but be not limited only to processor 140, memory 150.Those skilled in the art can manage
Solution, Fig. 9 is only the example of terminal device 100, does not constitute the restriction to terminal device 100, may include more than illustrating
Or less component, certain components or different components are perhaps combined, such as the terminal device 100 can also include defeated
Enter output equipment, network access equipment, bus etc..
Alleged processor 140 can be central processing unit (Central Processing Unit, CPU), can also be
Other general processors, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit
(Application Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field-
Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,
Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processor
Deng.
The memory 150 can be the internal storage unit of the terminal device 100, such as terminal device 100 is hard
Disk or memory.The memory 150 is also possible to the External memory equipment of the terminal device 100, such as the terminal device
The plug-in type hard disk being equipped on 100, intelligent memory card (Smart Media Card, SMC), secure digital (Secure
Digital, SD) card, flash card (Flash Card) etc..Further, the memory 150 can also both include the terminal
The internal storage unit of equipment 100 also includes External memory equipment.The memory 150 for store the computer program with
And other programs and data needed for the terminal device 100.The memory 150 can be also used for temporarily storing defeated
Out or the data that will export.
It is apparent to those skilled in the art that for convenience of description and succinctly, only with above-mentioned each function
Can unit, model division progress for example, in practical application, can according to need and by above-mentioned function distribution by different
Functional unit, module are completed, i.e., the internal structure of described device is divided into different functional unit or module, more than completing
The all or part of function of description.Each functional unit in embodiment, module can integrate in one processing unit, can also
To be that each unit physically exists alone, can also be integrated in one unit with two or more units, it is above-mentioned integrated
Unit both can take the form of hardware realization, can also realize in the form of software functional units.In addition, each function list
Member, the specific name of module are also only for convenience of distinguishing each other, the protection scope being not intended to limit this application.Above system
The specific work process of middle unit, module, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, is not described in detail or remembers in some embodiment
The part of load may refer to the associated description of other embodiments.
Those of ordinary skill in the art may be aware that list described in conjunction with the examples disclosed in the embodiments of the present disclosure
Member and algorithm steps can be realized with the combination of electronic hardware or computer software and electronic hardware.These functions are actually
It is implemented in hardware or software, the specific application and design constraint depending on technical solution.Professional technician
Each specific application can be used different methods to achieve the described function, but this realization is it is not considered that exceed
The scope of the present invention.
In embodiment provided by the present invention, it should be understood that disclosed device/terminal device and method, it can be with
It realizes by another way.For example, device described above/terminal device embodiment is only schematical, for example, institute
The division of module or unit is stated, only a kind of logical function partition, there may be another division manner in actual implementation, such as
Multiple units or components can be combined or can be integrated into another system, or some features can be ignored or not executed.Separately
A bit, shown or discussed mutual coupling or direct-coupling or communication connection can be through some interfaces, device
Or the INDIRECT COUPLING or communication connection of unit, it can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme
's.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated module/unit be realized in the form of SFU software functional unit and as independent product sale or
In use, can store in a computer readable storage medium.Based on this understanding, the present invention realizes above-mentioned implementation
All or part of the process in example method, can also instruct relevant hardware to complete, the meter by computer program
Calculation machine program can be stored in a computer readable storage medium, the computer program when being executed by processor, it can be achieved that on
The step of stating each embodiment of the method.Wherein, the computer program includes computer program code, the computer program generation
Code can be source code form, object identification code form, executable file or certain intermediate forms etc..The computer-readable medium
It may include: any entity or device, recording medium, USB flash disk, mobile hard disk, magnetic that can carry the computer program code
Dish, CD, computer storage, read-only memory (ROM, Read-Only Memory), random access memory (RAM,
Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium etc..It should be noted that described
The content that computer-readable medium includes can carry out increasing appropriate according to the requirement made laws in jurisdiction with patent practice
Subtract, such as in certain jurisdictions, according to legislation and patent practice, computer-readable medium do not include be electric carrier signal and
Telecommunication signal.
Embodiment described above is merely illustrative of the technical solution of the present invention, rather than its limitations；Although referring to aforementioned reality
Applying example, invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each
Technical solution documented by embodiment is modified or equivalent replacement of some of the technical features；And these are modified
Or replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution should all
It is included within protection scope of the present invention.
Claims (10)
1. a kind of human behavior detection method characterized by comprising
The wireless signal reflected by human body is received, and obtains the channel state information CSI time series of the wireless signal；
The CSI time series is pre-processed, and feature extraction is carried out to the pretreated CSI time series, is obtained
To human body behavioural characteristic；
Classified by classifier to the human body behavioural characteristic, determines human body behavior according to classification results.
2. human behavior detection method as described in claim 1, which is characterized in that the CSI time series includes time letter
Breath and channel state information；
The CSI time series for obtaining the wireless signal specifically includes:
The channel state information in the CSI time series is obtained using orthogonal frequency division multiplexing method；
The channel state information includes multiple subcarrier informations.
3. human behavior detection method as claimed in claim 2, which is characterized in that described to carry out the CSI time series
Pretreatment specifically includes:
The Information abnormity value in the CSI time series is detected, the Information abnormity value is deleted；
The position that the Information abnormity value is deleted in the CSI time series carries out interpolation processing；
CSI time series after progress interpolation processing is filtered, the pretreated CSI time series is obtained.
4. human behavior detection method as claimed in claim 3, which is characterized in that in the detection CSI time series
Information abnormity value specifically include:
Status information preset range is set；
Judge the value of information in the channel state information whether in the status information preset range；
The value of information not in the status information preset range is detected as the Information abnormity value.
5. human behavior detection method as claimed in claim 2, which is characterized in that it is described to the pretreated CSI when
Between sequence carry out feature extraction, obtain human body behavioural characteristic and specifically include:
The pretreated CSI time series is subjected to dimension-reduction treatment, and determines the main spy of feature principal component and preset quantity
Levy vector；
The first-order difference mean value for calculating each main feature vector, calculates the variance of the feature principal component, and by the side
It is poor to be set as characteristics of human body's value with first-order difference mean value ratio；
Characteristics of human body's value of preset quantity forms the human body behavioural characteristic.
6. human behavior detection method as claimed in claim 5, which is characterized in that calculate the first-order difference mean value and specifically wrap
It includes:
Wherein, N is the number of the subcarrier, viFor the corresponding main feature vector.
7. such as human behavior detection method described in any one of claims 1 to 6, which is characterized in that by the human body row
It is characterized after being input to and being classified in classifier, further includes:
Decision is carried out to the classification results using data fusion method, human body behavior is determined according to the result of decision.
8. a kind of human behavior detection device characterized by comprising
Data obtaining module for receiving the wireless signal reflected by human body, and obtains the CSI time sequence of the wireless signal
Column；
Characteristic extracting module, for pre-processing the CSI time series, and to the pretreated CSI time sequence
Column carry out feature extraction, obtain human body behavioural characteristic；
Activity recognition module is classified for the human body behavioural characteristic to be input in classifier, true according to classification results
Determine human body behavior.
9. a kind of terminal device, including memory, processor and storage are in the memory and can be on the processor
The computer program of operation, which is characterized in that the processor realizes such as claim 1 to 7 when executing the computer program
The step of any one the method.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In when the computer program is executed by processor the step of any one of such as claim 1 to 7 of realization the method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810842914.0A CN108988968A (en) | 2018-07-27 | 2018-07-27 | Human behavior detection method, device and terminal device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810842914.0A CN108988968A (en) | 2018-07-27 | 2018-07-27 | Human behavior detection method, device and terminal device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108988968A true CN108988968A (en) | 2018-12-11 |
Family
ID=64551353
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810842914.0A Pending CN108988968A (en) | 2018-07-27 | 2018-07-27 | Human behavior detection method, device and terminal device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108988968A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109905399A (en) * | 2019-03-14 | 2019-06-18 | 哈尔滨工程大学 | A kind of social media individual abnormal user detection method to be developed based on self network structure |
| CN109998513A (en) * | 2019-01-25 | 2019-07-12 | 芯海科技（深圳）股份有限公司 | The error handling method of HRV signal is calculated in human body impedance signal |
| CN110011741A (en) * | 2019-03-29 | 2019-07-12 | 河北工程大学 | Personal identification method and device based on wireless signal |
| CN110300399A (en) * | 2019-06-24 | 2019-10-01 | 西北大学 | A kind of short distance multi-user concealed communication method and system based on Wi-Fi network interface card |
| CN110737201A (en) * | 2019-10-11 | 2020-01-31 | 珠海格力电器股份有限公司 | monitoring method, device, storage medium and air conditioner |
| CN110751115A (en) * | 2019-10-24 | 2020-02-04 | 北京金茂绿建科技有限公司 | Non-contact human behavior identification method and system |
| CN110855382A (en) * | 2019-12-02 | 2020-02-28 | 北京邮电大学 | Method and device for detecting human body activity data based on CSI (channel State information), and electronic equipment |
| CN111753686A (en) * | 2020-06-11 | 2020-10-09 | 深圳市三旺通信股份有限公司 | CSI-based people number identification method, device, equipment and computer storage medium |
| CN112839327A (en) * | 2021-01-21 | 2021-05-25 | 河北工程大学 | Personnel validity detection method and device based on WiFi signals |
| CN113726457A (en) * | 2021-07-28 | 2021-11-30 | 深圳市联洲国际技术有限公司 | Human body action recognition method and device, storage medium and network equipment |
| CN115586581A (en) * | 2022-12-02 | 2023-01-10 | 荣耀终端有限公司 | Personnel detection method and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104732208A (en) * | 2015-03-16 | 2015-06-24 | 电子科技大学 | Video human action reorganization method based on sparse subspace clustering |
| CN105719368A (en) * | 2014-12-04 | 2016-06-29 | 台达电子工业股份有限公司 | Personnel detection system and personnel detection method |
-
2018
- 2018-07-27 CN CN201810842914.0A patent/CN108988968A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105719368A (en) * | 2014-12-04 | 2016-06-29 | 台达电子工业股份有限公司 | Personnel detection system and personnel detection method |
| CN104732208A (en) * | 2015-03-16 | 2015-06-24 | 电子科技大学 | Video human action reorganization method based on sparse subspace clustering |
Non-Patent Citations (4)
| Title |
|---|
| 史白等: "基于信道状态信息的非合作式室内人体运动检测", 《计算机应用》 * |
| 王正杰等: "基于CSI的行为识别研究综述", 《计算机工程与应用》 * |
| 肖玲等: "基于WiFi 信号的人体动作识别系统", 《北京邮电大学学报》 * |
| 鲁勇等: "基于ＷｉＦｉ信号的人体行为感知技术研究综述", 《计算机学报》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109998513A (en) * | 2019-01-25 | 2019-07-12 | 芯海科技（深圳）股份有限公司 | The error handling method of HRV signal is calculated in human body impedance signal |
| CN109998513B (en) * | 2019-01-25 | 2022-04-19 | 芯海科技（深圳）股份有限公司 | Error processing method for calculating HRV signal in human body impedance signal |
| CN109905399A (en) * | 2019-03-14 | 2019-06-18 | 哈尔滨工程大学 | A kind of social media individual abnormal user detection method to be developed based on self network structure |
| CN109905399B (en) * | 2019-03-14 | 2021-06-01 | 哈尔滨工程大学 | Social media individual abnormal user detection method based on self network structure evolution |
| CN110011741A (en) * | 2019-03-29 | 2019-07-12 | 河北工程大学 | Personal identification method and device based on wireless signal |
| CN110300399B (en) * | 2019-06-24 | 2020-07-21 | 西北大学 | Close-range multi-user covert communication method and system based on Wi-Fi network card |
| CN110300399A (en) * | 2019-06-24 | 2019-10-01 | 西北大学 | A kind of short distance multi-user concealed communication method and system based on Wi-Fi network interface card |
| CN110737201A (en) * | 2019-10-11 | 2020-01-31 | 珠海格力电器股份有限公司 | monitoring method, device, storage medium and air conditioner |
| CN110737201B (en) * | 2019-10-11 | 2020-10-09 | 珠海格力电器股份有限公司 | Monitoring method and device, storage medium and air conditioner |
| CN110751115B (en) * | 2019-10-24 | 2021-01-01 | 北京金茂绿建科技有限公司 | Non-contact human behavior identification method and system |
| CN110751115A (en) * | 2019-10-24 | 2020-02-04 | 北京金茂绿建科技有限公司 | Non-contact human behavior identification method and system |
| CN110855382A (en) * | 2019-12-02 | 2020-02-28 | 北京邮电大学 | Method and device for detecting human body activity data based on CSI (channel State information), and electronic equipment |
| CN111753686A (en) * | 2020-06-11 | 2020-10-09 | 深圳市三旺通信股份有限公司 | CSI-based people number identification method, device, equipment and computer storage medium |
| CN112839327A (en) * | 2021-01-21 | 2021-05-25 | 河北工程大学 | Personnel validity detection method and device based on WiFi signals |
| CN112839327B (en) * | 2021-01-21 | 2022-08-16 | 河北工程大学 | Personnel validity detection method and device based on WiFi signals |
| CN113726457A (en) * | 2021-07-28 | 2021-11-30 | 深圳市联洲国际技术有限公司 | Human body action recognition method and device, storage medium and network equipment |
| CN115586581A (en) * | 2022-12-02 | 2023-01-10 | 荣耀终端有限公司 | Personnel detection method and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108988968A (en) | Human behavior detection method, device and terminal device | |
| Zou et al. | Towards occupant activity driven smart buildings via WiFi-enabled IoT devices and deep learning | |
| Zou et al. | Deepsense: Device-free human activity recognition via autoencoder long-term recurrent convolutional network | |
| EP3411670B1 (en) | Systems and methods for sensing an environment with wireless communication signals | |
| Hsieh et al. | Deep learning-based indoor localization using received signal strength and channel state information | |
| US11902857B2 (en) | Handling concept drift in Wi-Fi-based localization | |
| US20220327714A1 (en) | Motion Engine | |
| Ding et al. | A WiFi-based smart home fall detection system using recurrent neural network | |
| Zou et al. | Robust WiFi-enabled device-free gesture recognition via unsupervised adversarial domain adaptation | |
| CN106846729A (en) | A kind of fall detection method and system based on convolutional neural networks | |
| CN109447162B (en) | Real-time behavior recognition system based on Lora and Capsule and working method thereof | |
| CN114201988A (en) | Satellite navigation composite interference signal identification method and system | |
| Hnoohom et al. | An Efficient ResNetSE Architecture for Smoking Activity Recognition from Smartwatch. | |
| Jannat et al. | Efficient Wi-Fi-based human activity recognition using adaptive antenna elimination | |
| Bokhari et al. | DGRU based human activity recognition using channel state information | |
| CN109728863A (en) | Personnel activity's duration estimation method, device and terminal device | |
| Shang et al. | LSTM-CNN network for human activity recognition using WiFi CSI data | |
| CN115964670B (en) | Spectrum anomaly detection method | |
| CN114818788A (en) | Tracking target state identification method and device based on millimeter wave perception | |
| CN112488019A (en) | Fall detection method and device based on posture recognition, electronic equipment and storage medium | |
| Hu et al. | ResFi: WiFi-enabled device-free respiration detection based on deep learning | |
| CN113453180B (en) | Intelligent detection method and system for human body tumble and information data processing terminal | |
| Lolla et al. | WiFi motion detection: A study into efficacy and classification | |
| Kianoush et al. | Device-free crowd sensing in dense WiFi MIMO networks: Channel features and machine learning tools | |
| CN117056717A (en) | Behavior detection model training method, detection method, device and storage medium |
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: 20181211 |