CN115913415B - WIFI signal action recognition method and device based on RIS assistance and storage medium - Google Patents
WIFI signal action recognition method and device based on RIS assistance and storage medium Download PDFInfo
- Publication number
- CN115913415B CN115913415B CN202211401369.4A CN202211401369A CN115913415B CN 115913415 B CN115913415 B CN 115913415B CN 202211401369 A CN202211401369 A CN 202211401369A CN 115913415 B CN115913415 B CN 115913415B
- Authority
- CN
- China
- Prior art keywords
- data
- channel state
- action recognition
- wifi
- signal
- 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
- 230000009471 action Effects 0.000 title claims abstract description 162
- 238000000034 method Methods 0.000 title claims description 64
- 238000003062 neural network model Methods 0.000 claims abstract description 11
- 238000001514 detection method Methods 0.000 claims description 99
- 238000012549 training Methods 0.000 claims description 56
- 238000012545 processing Methods 0.000 claims description 38
- 238000012937 correction Methods 0.000 claims description 36
- 238000007781 pre-processing Methods 0.000 claims description 29
- 230000033001 locomotion Effects 0.000 claims description 18
- 238000004590 computer program Methods 0.000 claims description 8
- 230000007547 defect Effects 0.000 claims description 8
- 238000004891 communication Methods 0.000 claims description 5
- 230000008054 signal transmission Effects 0.000 claims description 4
- 238000007405 data analysis Methods 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 3
- 230000000877 morphologic effect Effects 0.000 claims description 3
- 230000008447 perception Effects 0.000 abstract description 16
- 238000005516 engineering process Methods 0.000 abstract description 4
- 239000000523 sample Substances 0.000 description 18
- 230000008569 process Effects 0.000 description 15
- 238000013461 design Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000011160 research Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000002950 deficient Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000005429 filling process Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000009877 rendering Methods 0.000 description 2
- 230000002238 attenuated effect Effects 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Radar Systems Or Details Thereof (AREA)
Abstract
On one hand, the invention aims at the problem of limited perception range in the WIFI perception technology, introduces an RIS intelligent reflecting surface into the scene to improve the problem of rapid attenuation of signals on a reflecting link by utilizing the capability of RIS passive beam forming, thereby realizing the enhancement of the signal perception capability in the reflecting link, effectively widening the limit of the WIFI perception boundary and further acquiring more channel information of objects in a reflecting area, and further accurately realizing the action recognition of the objects in the reflecting area; on the other hand, the invention utilizes the double-flow convolution enhancement neural network model to extract the characteristics from two dimensions of time and channel, and utilizes the two terms of amplitude and phase of input data to classify to obtain the recognition result, thus, more characteristics are utilized to conduct action classification, and the recognition accuracy can be further improved.
Description
Technical Field
The invention belongs to the field of action recognition, and particularly relates to a WIFI signal action recognition method and device based on RIS assistance and a storage medium.
Background
The communication perception integration is an important field of development of the next generation 6G mobile communication network, and is realized in various scenes (including health medical treatment, remote monitoring, intelligent families and other scenes), so that the communication network has more intelligent characteristics; in a plurality of 6G communication perception integrated schemes, WIFI wireless perception is an important technical means for realizing indoor multi-object, personnel and environment perception in the future, wherein the WIFI wireless perception has more research and application in the aspects of vital sign monitoring, man-machine interaction, activity recognition, indoor positioning and tracking and the like due to the advantages of low cost, universality, non-contact, non-invasion and the like.
At present, although many scholars and institutions develop many technical researches and mechanism exploration on indoor perception of WIFI signals, the perception of WIFI still faces many challenges; currently, the channel state information (CSI, channel State Information) sensing task for WIFI mainly has the following problems, namely a sensing range problem, a sensing limit problem, a position dependent problem and an automatic segmentation problem of continuous action, wherein the sensing range problem is a main factor affecting the wireless sensing effect of WIFI.
In the practical application process, the WIFI sensing range is mainly concentrated in the boundary of the first Fresnel zone, so that the indoor sensing capability of the WIFI is limited, particularly when a sensed object is not in a WIFI signal direct link or the first Fresnel zone, the WIFI signal can reach the receiving end only by reflection, and the strength of the signal is greatly attenuated after the signal is reflected, so that only a very small part of reflected signals can be received by the receiving end; therefore, for the WIFI signal on the reflection path, the channel information that can be captured becomes very small, which results in that the object in the reflection area or the scattering area is difficult to be identified by the WIFI signal on the reflection path; therefore, how to realize the motion recognition of objects in the reflection area becomes a current research hotspot in the technical field of WIFI wireless sensing.
Disclosure of Invention
The invention aims to provide a WIFI signal action recognition method and device based on RIS assistance and a storage medium, which are used for solving the problem that in the prior art, the action of an object in a reflection area is difficult to recognize through a WIFI signal on a reflection path.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
in a first aspect, a method for identifying a WIFI signal action based on RIS assistance is provided, including:
The method comprises the steps that an RIS reflecting end receives an initial WIFI detection signal which is sent by a WIFI signal sending end and used for detecting actions of a target human body, and an optimal beam forming codebook of the initial WIFI detection signal is determined, wherein the RIS reflecting end is arranged on a reflecting link of the initial WIFI detection signal;
the RIS reflecting end performs beam forming on the initial WIFI detection signal by using the optimal beam forming codebook to obtain an enhanced WIFI detection signal;
the RIS reflecting end sends the enhanced WIFI detection signal to the CSI action recognition end;
the CSI action recognition terminal receives the enhanced WIFI detection signal sent by the RIS reflection terminal, wherein the target human body is positioned between the WIFI signal transmission terminal and the CSI action recognition terminal;
the CSI action recognition terminal obtains channel state data of the enhanced WIFI detection signal based on the enhanced WIFI detection signal, wherein each piece of channel data in the channel state data comprises amplitude data and phase data;
and the CSI action recognition end inputs the channel state data into an action recognition model so as to perform action recognition on the target human body based on the action recognition model and the channel state data, and obtain an action recognition result of the target human body.
Based on the above disclosure, the RIS intelligent reflecting surface is introduced into the WIFI wireless sensing scene, namely the RIS intelligent reflecting surface is arranged on a reflecting link of an initial WIFI detection signal emitted to a target human body, and beam forming is carried out on the initial WIFI detection signal based on the RIS intelligent reflecting surface so as to enhance the sensing capability of the signal in the reflecting link and obtain an enhanced WIFI detection signal, and then the enhanced WIFI detection signal is sent to the CSI action recognition end through the RIS intelligent reflecting surface and is subjected to action recognition through the action recognition model; through the design, the RIS intelligent reflecting surface is introduced into the scene aiming at the problem of limited sensing range in the WIFI sensing technology, so that the problem of rapid attenuation of signals on a reflecting link is solved by utilizing the RIS passive beam forming capability, the enhancement of the signal sensing capability in the reflecting link is realized, the limit of the WIFI sensing boundary is effectively widened, and further more channel information of objects in a reflecting area is acquired.
In one possible design, before inputting the channel state data into the action recognition model, the method further comprises:
judging whether the channel state data is defect data or not;
if yes, carrying out data filling on the channel state data to obtain preprocessed channel state data after filling;
carrying out phase correction on the preprocessed channel state data to obtain corrected channel state data;
and denoising the corrected channel state data by using a morphological filtering algorithm to obtain denoising channel state data after denoising is completed, so that the denoising channel state data is input into an action recognition model to obtain an action recognition result of the target human body.
Based on the above disclosure, the method and the device sequentially perform data filling, phase correction and denoising processing on the channel state data before performing action recognition, so that invalid data can be removed, the accuracy of the data is ensured, and the accuracy of the action recognition is improved.
In one possible design, the enhanced WIFI sounding signal includes a number of different channels of subcarrier signals, and the channel state data includes channel data of the number of different channels of subcarrier signals;
Wherein determining whether the channel state data is defect data comprises:
judging whether the channel data of any subcarrier signal in the channel state data is defect data or not;
if yes, carrying out interpolation processing on the channel data of any subcarrier signal by using an interpolation method to obtain preprocessed channel data of any subcarrier signal; or (b)
And carrying out data zero padding processing on the channel data of any subcarrier signal to obtain the preprocessed channel data of any subcarrier signal, and after the channel data interpolation processing or the data zero padding processing of all subcarrier signals are finished, utilizing the preprocessed channel data of all subcarrier signals to form the preprocessed channel state data.
In one possible design, phase correcting the preprocessed channel state data to obtain corrected channel state data includes:
for the preprocessing channel data of the ith subcarrier signal in the preprocessing channel state data, carrying out phase correction by adopting the following formula (1) to obtain corrected channel data of the ith subcarrier signal after the phase correction;
In the above formula (1), phi i The pre-processed channel data representing the ith subcarrier signal,correction channel data, phi, representing the ith subcarrier signal t Preprocessing channel data, phi, representing the last subcarrier signal of a number of subcarrier signals 1 Preprocessing channel data, k, representing a first subcarrier signal of a number of subcarrier signals t Index sequence number k corresponding to last subcarrier signal 1 Represents the index sequence number, phi, corresponding to the first subcarrier signal j The preprocessed channel data representing the jth subcarrier signal, n being the total number of subcarrier signals, k i An index sequence number corresponding to the i-th subcarrier signal, and i=1, 2, n;
and adding 1 to i until i is equal to n, obtaining n pieces of correction channel data, and forming the correction channel state data by using the n pieces of correction channel data, wherein the initial value of i is 1.
In one possible design, before receiving the enhanced WIFI detection signal sent by the RIS reflector and used for detecting the motion of the target human body, the method further includes:
acquiring a training data set, wherein the training data set comprises a plurality of data subsets corresponding to different human body actions, and each data subset comprises a plurality of channel state sample data corresponding to enhanced WIFI detection sample signals;
Performing data preprocessing on the training data set to obtain a preprocessed training data set, wherein the preprocessing comprises data filling processing, phase correction processing and denoising processing;
and training a double-flow convolution enhanced neural network model by taking the preprocessed training data set as input and taking an action recognition result corresponding to each piece of channel state sample data in the preprocessed training data set as output so as to obtain the action recognition model after training is finished.
In a second aspect, a device for identifying a WIFI signal action based on RIS assistance is provided, taking the device as a CSI action identifying end as an example, including:
the first receiving unit is used for receiving an enhanced WIFI detection signal which is sent by the RIS reflecting end and used for detecting the action of a target human body, wherein the enhanced WIFI detection signal is obtained by carrying out beam forming on an initial WIFI detection signal sent by the WIFI signal transmitting end after the RIS reflecting end receives the initial WIFI detection signal, and the target human body is positioned between the WIFI signal transmitting end and the CSI action identifying end;
the data analysis unit is used for obtaining channel state data of the enhanced WIFI detection signal based on the enhanced WIFI detection signal;
And the action recognition unit is used for inputting the channel state data into an action recognition model so as to perform action recognition on the target human body based on the action recognition model and the channel state data, and obtain an action recognition result of the target human body.
In a third aspect, another device for identifying WIFI signal actions based on RIS assistance is provided, taking the device as a RIS reflecting end as an example, including:
the second receiving unit is used for receiving an initial WIFI detection signal which is sent by the WIFI signal transmitting end and used for detecting the action of a target human body, and determining an optimal beam forming codebook of the initial WIFI detection signal, wherein the RIS reflecting end is arranged on a reflecting link of the initial WIFI detection signal;
the beam forming unit is used for carrying out beam forming on the initial WIFI detection signal by utilizing the optimal beam forming codebook to obtain an enhanced WIFI detection signal;
and the sending unit is used for sending the enhanced WIFI detection signal to the CSI action recognition end, so that the CSI action recognition end obtains channel state data of the enhanced WIFI detection signal after receiving the enhanced WIFI detection signal, and inputs the channel state data into the action recognition model so as to obtain an action recognition result of the target human body based on the action recognition model and the channel state data.
In a fourth aspect, a third device for identifying WIFI signal actions based on RIS assistance is provided, taking the device as an electronic device, and the device includes a memory, a processor and a transceiver, which are connected in communication in sequence, where the memory is used to store a computer program, the transceiver is used to send and receive a message, and the processor is used to read the computer program, and execute the method for identifying WIFI signal actions based on RIS assistance according to the first aspect or any one of the possible designs of the first aspect.
In a fifth aspect, a storage medium is provided, where instructions are stored, and when the instructions are executed on a computer, the method for identifying a WIFI signal action based on the RIS according to any one of the first aspect or the first aspect may be implemented.
In a sixth aspect, there is provided a computer program product comprising instructions which, when run on a computer, cause the computer to perform the RIS-based assisted WIFI signal action recognition method as in the first aspect or any one of the possible designs of the first aspect.
The beneficial effects are that:
(1) Aiming at the problem of limited perception range in the WIFI perception technology, the RIS intelligent reflecting surface is introduced into the scene, so that the problem of rapid attenuation of signals on a reflecting link is improved by utilizing the RIS passive beam forming capability, the enhancement of the signal perception capability in the reflecting link is realized, the limit of the WIFI perception boundary is effectively widened, and more channel information of objects in a reflecting area is obtained.
(2) The channel state data used by the invention comprises amplitude data and phase data, so that the method is equivalent to classifying the amplitude and the phase of the signal by using the motion recognition model, and therefore, the method extracts more characteristic data of the CSI and has higher motion recognition accuracy.
Drawings
Fig. 1 is a schematic architecture diagram of a WIFI signal motion recognition system based on RIS assistance according to an embodiment of the present invention;
fig. 2 is a schematic step flow diagram of a method for identifying WIFI signal actions based on RIS assistance according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a CSI action recognition end according to an embodiment of the present invention;
FIG. 4 is a schematic structural diagram of a RIS reflective end according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the present invention will be briefly described below with reference to the accompanying drawings and the description of the embodiments or the prior art, and it is obvious that the following description of the structure of the drawings is only some embodiments of the present invention, and other drawings can be obtained according to these drawings without inventive effort to a person skilled in the art. It should be noted that the description of these examples is for aiding in understanding the present invention, but is not intended to limit the present invention.
It will be understood that, although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another element. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments of the present invention.
It should be understood that for the term "and/or" that may appear herein, it is merely one association relationship that describes an associated object, meaning that there may be three relationships, e.g., a and/or B, may represent: a alone, B alone, and both a and B; for the term "/and" that may appear herein, which is descriptive of another associative object relationship, it means that there may be two relationships, e.g., a/and B, it may be expressed that: a alone, a alone and B alone; in addition, for the character "/" that may appear herein, it is generally indicated that the context associated object is an "or" relationship.
Examples:
referring to fig. 1, a system architecture is provided for the present application, which may include, but is not limited to: the system comprises a WIFI signal transmitting end, a CSI action recognition end and at least one RIS (Reconfigurable Intelligent Surface, intelligent reflecting surface) reflecting end, wherein a target human body (namely a human body to be detected) is positioned between the WIFI signal transmitting end and the CSI action recognition end (namely in a reflecting area or a scattering area of an initial WIFI detection signal sent by the WIFI signal transmitting end), and each RIS reflecting end is positioned on a reflecting link of the initial WIFI detection signal; based on the initial WIFI detection signals sent by the WIFI signal transmitting end can be received by each RIS reflecting end, and beam forming is carried out on the initial WIFI detection signals to obtain enhanced WIFI detection signals; then, the received signal can be sent to a CSI action recognition end; the CSI action recognition terminal uses the method provided by the embodiment to perform action recognition on the received enhanced WIFI detection signal, so as to obtain the action recognition result of the target human body; through the design, the RIS intelligent reflecting surface is introduced into the WIFI wireless sensing scene, so that the problem of abrupt attenuation of signals on a reflecting link is solved by utilizing the RIS passive beam forming capability, the sensing capability of the signals in the reflecting link can be effectively enhanced, the limit of the WIFI sensing boundary can be effectively widened, and the action recognition of objects in a reflecting area can be accurately realized.
Referring to fig. 2, in this embodiment, the method may be, but not limited to, running on the RIS reflection side and/or the CSI action recognition side, where, for example, the CSI action recognition side may be, but not limited to, a personal computer (personal computer, PC), a tablet computer, a smart phone, or a personal digital assistant (personal digital assistant, PDA), etc., it is to be understood that the foregoing execution body is not limited to the embodiment of the present application, and accordingly, the running steps of the method may be, but not limited to, as shown in steps S1 to S6 below.
S1, an RIS (RIS) reflecting end receives an initial WIFI detection signal which is sent by a WIFI signal transmitting end and used for detecting actions of a target human body, and determines an optimal beam forming codebook of the initial WIFI detection signal, wherein the RIS reflecting end is arranged on a reflecting link of the initial WIFI detection signal; in particular applications, for example, a traversal algorithm may be used to determine the optimal beamforming codebook of the initial WIFI probe signal, alternatively, the traversal algorithm may be used to perform codebook optimization by a greedy algorithm; after determining the optimal beamforming codebook of the initial WIFI detection signal, the codebook may be used to perform beamforming on the initial WIFI detection signal to enhance the sensing capability thereof, as shown in step S2 below.
S2, the RIS reflecting end performs beam forming on the initial WIFI detection signal by utilizing the optimal beam forming codebook to obtain an enhanced WIFI detection signal; when the method is specifically applied, the initial WIFI detection signal is subjected to beam forming by utilizing the optimal beam forming codebook, namely, the phase of the signal is adjusted, so that the radiation energy of the signal is concentrated, and the CSI action recognition end is ensured to be positioned at a position with stronger signal; therefore, the RIS is utilized to realize the beam forming of the WIFI signal on the reflection path, more signals can be focused on the receiving end (namely the CSI action recognition end) on the reflection path, and therefore the channel perception capability on the reflection path is enhanced.
In this embodiment, the enhanced WIFI detection signal may include, but is not limited to: a number of different channel subcarrier signals, such as subcarrier signals comprising 30 different channels; of course, the number of channels may be specifically set according to actual use, and is not limited to the foregoing examples.
After the beam forming of the signal on the reflection link is completed, the formed signal can be sent to the CSI action recognition end, so that the CIS action recognition end can recognize the action of a target human body by means of the received signal; the operation recognition process may be, but is not limited to, those shown in steps S3 to S6.
S3, the RIS reflecting end sends the enhanced WIFI detection signal to the CSI action recognition end.
S4, the CSI action recognition end receives the enhanced WIFI detection signal sent by the RIS reflection end, wherein the target human body is located between the WIFI signal transmission end and the CSI action recognition end.
S5, the CSI action recognition terminal obtains channel state data of the enhanced WIFI detection signal based on the enhanced WIFI detection signal, wherein each piece of channel data in the channel state data comprises amplitude data and phase data; in a specific application, since the enhanced WIFI probe signal has been described as including a plurality of subcarrier signals of different channels, the signal state data obtained in this step includes channel data of each subcarrier signal, and each channel data includes amplitude and phase data of a corresponding subcarrier signal; in this way, the present embodiment is equivalent to using the amplitude and phase of the signal to realize the motion recognition.
In this embodiment, in order to reduce the interference of the invalid data on the motion recognition and thereby improve the accuracy of the motion recognition, a data preprocessing step is further provided, as shown in the following steps S51 to S54.
S51, judging whether the channel state data are defect data or not; in specific application, the defect data is data which is not received completely or is missing; since the foregoing has described that the channel state data includes channel data of a plurality of subcarrier signals, each channel data needs to be determined to be missing, and any channel data is described as an example, that is, whether the channel data of any subcarrier signal in the channel state data is missing data is determined; specifically, assuming that the channel state data includes 30 pieces of channel data, each piece of channel data includes 1000 pieces of data, at this time, if the CSI action recognition terminal receives 990 pieces of data only for any piece of channel data, then determining that any piece of channel data is defective data; and judging that the received data is defective data, and filling the data is needed, so that the integrity of the data is ensured; wherein the data population process is as shown in step S52 below.
S52, if yes, carrying out data filling on the channel state data to obtain preprocessed channel state data after filling; in the embodiment, interpolation processing may be performed on the channel data of any subcarrier signal by using an interpolation method to obtain pre-processed channel data of any subcarrier signal, or data zero padding processing may be performed on the channel data of any subcarrier signal to obtain pre-processed channel data of any subcarrier signal; in this embodiment, the interpolation rule is equivalent to interpolating incomplete data into complete data (i.e. interpolating data with 1000 length), and the data zero-filling process is to zero-fill the channel data of any subcarrier signal to reach 1000 length; thus, by adopting the same method, after the channel data of all the subcarrier signals are filled with data, complete preprocessing channel state data can be formed.
After the data of each channel data in the channel state data is filled, the phase correction processing of the data is needed to obtain accurate phase information; the phase correction process is shown in the following step S53.
S53, carrying out phase correction on the preprocessed channel state data to obtain corrected channel state data; in particular applications, the following steps S53a and S53b may be used, but are not limited to, to achieve phase correction of the preprocessed channel state data.
S53a, carrying out phase correction on the preprocessed channel data of the ith subcarrier signal in the preprocessed channel state data by adopting the following formula (1) so as to obtain corrected channel data of the ith subcarrier signal after the phase correction.
In the above formula (1), phi i The pre-processed channel data representing the ith subcarrier signal,correction channel data, phi, representing the ith subcarrier signal t Preprocessing channel data, phi, representing the last subcarrier signal of a number of subcarrier signals 1 Preprocessing channel data, k, representing a first subcarrier signal of a number of subcarrier signals t Index sequence number k corresponding to last subcarrier signal 1 Represents the index sequence number, phi, corresponding to the first subcarrier signal j The preprocessed channel data representing the jth subcarrier signal, n being the total number of subcarrier signals, k i The index sequence number corresponding to the i-th subcarrier signal is indicated, and i=1, 2.
In the present embodiment, phi i The phase of the pre-processed channel data of the i-th subcarrier signal is actually represented,then the phase corresponding to the correction channel data representing the ith subcarrier signal and the same is true t And phi 1 The essence of this is also the phase, but the value of n may be, but not limited to, 30, and the index sequence number corresponding to 30 subcarrier signals (the 30 subcarrier signals are subcarriers corresponding to 20MHz bandwidth specified by IEEE 802.11n (wireless transmission standard protocol)) is: -28, -26, -24, -22, -20, -18, -16, -14, -12, -10, -8, -6, -4, -2, -1,1,3,5,7,9, 11,13, 15, 17, 19, 21, 23, 25, 27, 28; thus, k t Take the value of 28, k 1 The value is-28.
Thus, with the foregoing formula (1), the phase correction can be performed on each of the preprocessed channel data, so as to ensure the accuracy of the phase information, wherein the cyclic processing procedure of the phase correction is as follows in step S53b.
S53b, adding 1 to i until i is equal to n, and obtaining n pieces of correction channel data to form the correction channel state data by using n pieces of correction signal data, wherein the initial value of i is 1.
Thus, through the foregoing steps S53a and S53b, the pre-processed channel data of each subcarrier signal may be phase corrected, so as to recover the correct phase information, so as to provide an accurate data basis for the subsequent action recognition.
After the data correction of each of the pre-processed channel data is completed, the denoising process may be performed on each of the pre-processed channel data, as shown in step S54 below.
S54, denoising the corrected channel state data by using a morphological filtering algorithm to obtain denoising channel state data after denoising is completed, so that the denoising channel state data is input into an action recognition model to obtain an action recognition result of the target human body; in specific application, denoising each piece of preprocessed channel data to remove high-frequency noise points in the data; of course, after all the preprocessed channel data are denoised, denoised channel state data can be composed.
After the preprocessing of the data is completed, the preprocessed data can be input into the action recognition model, so that the action recognition of the target human body is realized; the action recognition process is shown in the following step S6.
S6, inputting the channel state data into an action recognition model by the CSI action recognition end so as to perform action recognition on the target human body based on the action recognition model and the channel state data, and obtaining an action recognition result of the target human body; in this embodiment, for example, but not limited to, a trained dual-flow convolutional enhanced neural network model is used as an action recognition model, and in a specific recognition process, each piece of data input into the model in this embodiment includes an amplitude and a phase, so that the method is equivalent to using the attention mechanism of the model to perform feature extraction in two dimensions of time and channels, so that action recognition is performed based on the amplitude and the phase, and therefore, the method classifies feature data of multiple dimensions, and can improve accuracy of model recognition.
One of the training methods of the dual-flow convolution enhanced neural network model is disclosed as follows, and steps S01 to S03 are described below.
S01, acquiring a training data set, wherein the training data set comprises a plurality of data subsets corresponding to different human body actions, and each data subset comprises channel state sample data corresponding to a plurality of enhanced WIFI detection sample signals; in a specific application, the method can be used for acquiring data subsets corresponding to 6 human actions, wherein the 6 human actions can be used for acquiring the data subsets corresponding to the 6 human actions in turn: walking, running, bending over, falling, standing and sitting up; optionally, the frequency of data corresponding to each action is 200 times/second, each action can be sampled for different people 200 times, and the sampling time length is 5s each time, so that the total of six data subsets is 1200 pieces of channel state sample data; further, the length of each piece of channel state sample data may be, but not limited to, 1000, each piece of channel state sample data includes channel sample data corresponding to 30 subcarrier sample signals of different channels, and each piece of channel sample data may be, but not limited to, including the amplitude and the phase of the corresponding subcarrier sample signal, so that each piece of channel state sample data has 240-dimensional data in total, and therefore, the size of each piece of channel state sample data is a data matrix of 240×1000 substantially; of course, the sample parameters of the training data set may be specifically set according to practical use, and are not limited to the foregoing examples.
In this embodiment, data preprocessing is also required for the training data set, as shown in step S02 below.
S02, carrying out data preprocessing on the training data set to obtain a preprocessed training data set, wherein the preprocessing comprises data filling processing, phase correction processing and denoising processing; in specific applications, the process of the data filling process, the phase correction process and the denoising process of the training data set can be referred to in the steps S51 to S54, and will not be described herein.
After the data preprocessing of the training data set is completed, training of the model may be performed as shown in step S03 below.
S03, training a double-flow convolution enhanced neural network model by taking the preprocessed training data set as input and taking an action recognition result corresponding to each piece of channel state sample data in the preprocessed training data set as output so as to obtain the action recognition model after training is finished; in specific application, the HAT (double-flow convolution enhanced neural network model) divides the original CSI input into a channel flow and a time sequence flow, and respectively processes the channel flow and the time flow through two parallel neural modules, so that feature extraction is performed in two dimensions of time and channel, and motion recognition is realized by utilizing multiple features; thus, the accuracy of motion recognition of the model can be improved.
According to the RIS-assisted WIFI signal action recognition method described in detail in the steps S1-S6, on one hand, the RIS intelligent reflecting surface is introduced into a WIFI wireless sensing scene, so that the problem of sharp attenuation of signals on a reflecting link is solved by utilizing the RIS passive beam forming capability, thus, the sensing capability of the signals in the reflecting link can be effectively enhanced, the limit of the WIFI sensing boundary can be effectively widened, and the action recognition of objects in a reflecting area can be accurately realized; on the other hand, the invention utilizes the double-flow convolution enhancement neural network model to extract the characteristics from two dimensions of time and channel, and utilizes the two terms of amplitude and phase of input data to classify to obtain the recognition result, thus, more characteristics are utilized to conduct action classification, and the recognition accuracy can be further improved.
As shown in fig. 3, a hardware device for implementing the method for identifying a WIFI signal action based on RIS assistance in the first aspect of the present embodiment is provided in a second aspect of the present embodiment, and includes:
the first receiving unit is used for receiving an enhanced WIFI detection signal which is sent by the RIS reflecting end and used for detecting the action of a target human body, wherein the enhanced WIFI detection signal is obtained by carrying out beam forming on an initial WIFI detection signal after the RIS reflecting end receives the initial WIFI detection signal sent by the WIFI signal transmitting end, and the target human body is located between the WIFI signal transmitting end and the CSI action recognition end.
And the data analysis unit is used for obtaining the channel state data of the enhanced WIFI detection signal based on the enhanced WIFI detection signal.
And the action recognition unit is used for inputting the channel state data into an action recognition model so as to perform action recognition on the target human body based on the action recognition model and the channel state data, and obtain an action recognition result of the target human body.
The working process, working details and technical effects of the device provided in this embodiment may refer to the first aspect of the embodiment, and are not described herein again.
As shown in fig. 4, a third aspect of the present embodiment provides another hardware device for implementing the method for identifying a WIFI signal action based on RIS assistance in the first aspect of the present embodiment, which includes:
the second receiving unit is used for receiving an initial WIFI detection signal which is sent by the WIFI signal transmitting end and used for detecting the action of a target human body, and determining an optimal beam forming codebook of the initial WIFI detection signal, wherein the RIS reflecting end is arranged on a reflecting link of the initial WIFI detection signal.
And the beam forming unit is used for carrying out beam forming on the initial WIFI detection signal by utilizing the optimal beam forming codebook to obtain an enhanced WIFI detection signal.
And the sending unit is used for sending the enhanced WIFI detection signal to the CSI action recognition end, so that the CSI action recognition end obtains channel state data of the enhanced WIFI detection signal after receiving the enhanced WIFI detection signal, and inputs the channel state data into the action recognition model so as to obtain an action recognition result of the target human body based on the action recognition model and the channel state data.
The working process, working details and technical effects of the device provided in this embodiment may refer to the first aspect of the embodiment, and are not described herein again.
As shown in fig. 5, in a fourth aspect of the present embodiment, a third device for identifying WIFI signal actions based on RIS assistance is provided, where the device is an electronic device, and includes: the system comprises a memory, a processor and a transceiver which are sequentially and communicatively connected, wherein the memory is used for storing a computer program, the transceiver is used for receiving and transmitting messages, and the processor is used for reading the computer program and executing the WIFI signal action recognition method based on RIS assistance according to the first aspect of the embodiment.
By way of specific example, the Memory may include, but is not limited to, random access Memory (randomaccess Memory, RAM), read Only Memory (ROM), flash Memory (Flash Memory), first-in-first-out Memory (First Input First Output, FIFO) and/or first-in-last-out Memory (First In Last Out, FILO), etc.; in particular, the processor may include one or more processing cores, such as a 4-core processor, an 8-core processor, or the like. The processor may be implemented in at least one hardware form of DSP (Digital Signal Processing ), FPGA (Field-Programmable Gate Array, field programmable gate array), PLA (Programmable LogicArray ), and may also include a main processor and a coprocessor, where the main processor is a processor for processing data in an awake state, and is also called CPU (Central Processing Unit ); a coprocessor is a low-power processor for processing data in a standby state.
In some embodiments, the processor may be integrated with a GPU (Graphics Processing Unit, image processor) for taking charge of rendering and rendering of content required to be displayed by the display screen, for example, the processor may not be limited to a microprocessor employing a model number of STM32F105 family, a reduced instruction set computer (reduced instruction set computer, RISC) microprocessor, an X86 or other architecture processor, or a processor integrating an embedded neural network processor (neural-network processing units, NPU); the transceiver may be, but is not limited to, a wireless fidelity (WIFI) wireless transceiver, a bluetooth wireless transceiver, a general packet radio service technology (General Packet Radio Service, GPRS) wireless transceiver, a ZigBee protocol (low power local area network protocol based on the ieee802.15.4 standard), a 3G transceiver, a 4G transceiver, and/or a 5G transceiver, etc. In addition, the device may include, but is not limited to, a power module, a display screen, and other necessary components.
The working process, working details and technical effects of the electronic device provided in this embodiment may refer to the first aspect of the embodiment, and are not described herein again.
A fifth aspect of the present embodiment provides a storage medium storing instructions including the method for identifying a WIFI signal action based on an RIS assistance according to the first aspect of the present embodiment, that is, the storage medium stores instructions, and when the instructions are executed on a computer, the method for identifying a WIFI signal action based on an RIS assistance according to the first aspect is executed.
The storage medium refers to a carrier for storing data, and may include, but is not limited to, a floppy disk, an optical disk, a hard disk, a flash memory, a flash disk, and/or a memory stick (memory stick), where the computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable devices.
The working process, working details and technical effects of the storage medium provided in this embodiment may refer to the first aspect of the embodiment, and are not described herein again.
A sixth aspect of the present embodiment provides a computer program product comprising instructions which, when run on a computer, cause the computer to perform the RIS-assisted WIFI signal action recognition method according to the first aspect of the embodiment, wherein the computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus.
Finally, it should be noted that: the foregoing description is only of the preferred embodiments of the invention and is not intended to limit the scope of the invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (9)
1. The WIFI signal action recognition method based on RIS assistance is characterized by being applied to a CSI action recognition terminal, wherein the method comprises the following steps:
Receiving an enhanced WIFI detection signal which is transmitted by a RIS reflection end and used for detecting the action of a target human body, wherein the enhanced WIFI detection signal is obtained by determining an optimal beam forming codebook of an initial WIFI detection signal after the RIS reflection end receives the initial WIFI detection signal transmitted by a WIFI signal transmission end, and carrying out beam forming on the initial WIFI detection signal by utilizing the optimal beam forming codebook, and the target human body is positioned between the WIFI signal transmission end and the CSI action recognition end;
obtaining channel state data of the enhanced WIFI detection signal based on the enhanced WIFI detection signal, wherein each piece of channel data in the channel state data comprises amplitude data and phase data;
inputting the channel state data into an action recognition model to perform action recognition on the target human body based on the action recognition model and the channel state data so as to obtain an action recognition result of the target human body;
before receiving the enhanced WIFI detection signal for detecting the motion of the target human body, which is sent by the RIS reflecting end, the method further comprises the following steps:
acquiring a training data set, wherein the training data set comprises a plurality of data subsets corresponding to different human body actions, and each data subset comprises a plurality of channel state sample data corresponding to enhanced WIFI detection sample signals;
Performing data preprocessing on the training data set to obtain a preprocessed training data set, wherein the preprocessing comprises data filling processing, phase correction processing and denoising processing;
and training a double-flow convolution enhanced neural network model by taking the preprocessed training data set as input and taking an action recognition result corresponding to each piece of channel state sample data in the preprocessed training data set as output so as to obtain the action recognition model after training is finished.
2. The method of claim 1, wherein prior to inputting the channel state data into the action recognition model, the method further comprises:
judging whether the channel state data is defect data or not;
if yes, carrying out data filling on the channel state data to obtain preprocessed channel state data after filling;
carrying out phase correction on the preprocessed channel state data to obtain corrected channel state data;
and denoising the corrected channel state data by using a morphological filtering algorithm to obtain denoising channel state data after denoising is completed, so that the denoising channel state data is input into an action recognition model to obtain an action recognition result of the target human body.
3. The method of claim 2, wherein the enhanced WIFI sounding signal comprises subcarrier signals of a number of different channels, and wherein the channel state data comprises channel data of subcarrier signals of a number of different channels, respectively;
wherein determining whether the channel state data is defect data comprises:
judging whether the channel data of any subcarrier signal in the channel state data is defect data or not;
if yes, carrying out interpolation processing on the channel data of any subcarrier signal by using an interpolation method to obtain preprocessed channel data of any subcarrier signal; or (b)
And carrying out data zero padding processing on the channel data of any subcarrier signal to obtain the preprocessed channel data of any subcarrier signal, and after the channel data interpolation processing or the data zero padding processing of all subcarrier signals are finished, utilizing the preprocessed channel data of all subcarrier signals to form the preprocessed channel state data.
4. A method according to claim 3, wherein phase correcting the pre-processed channel state data to obtain corrected channel state data comprises:
For the preprocessing channel data of the ith subcarrier signal in the preprocessing channel state data, carrying out phase correction by adopting the following formula (1) to obtain corrected channel data of the ith subcarrier signal after the phase correction;
（1）
in the above-mentioned formula (1),preprocessing channel data representing the ith subcarrier signal, < > and>correction channel data representing the ith subcarrier signal, < >>Preprocessing channel data representing the last subcarrier signal of a number of subcarrier signals, +.>Preprocessing channel data representing a first subcarrier signal of a plurality of subcarrier signals, ">Index sequence number corresponding to the last subcarrier signal, < +.>Index sequence number corresponding to the first subcarrier signal, is indicated>Preprocessing channel data representing jth subcarrier signal, and>is the total number of subcarrier signals, +.>An index sequence number corresponding to the i-th subcarrier signal, and i=1, 2, n;
and adding 1 to i until i is equal to n, obtaining n pieces of correction channel data, and forming the correction channel state data by using the n pieces of correction channel data, wherein the initial value of i is 1.
5. The WIFI signal action recognition method based on RIS assistance is characterized by being applied to a RIS reflecting end, wherein the method comprises the following steps:
Receiving an initial WIFI detection signal sent by a WIFI signal transmitting end and used for detecting actions of a target human body, and determining an optimal beam forming codebook of the initial WIFI detection signal, wherein the RIS reflecting end is arranged on a reflecting link of the initial WIFI detection signal;
performing beam forming on the initial WIFI detection signal by using the optimal beam forming codebook to obtain an enhanced WIFI detection signal;
the enhanced WIFI detection signal is sent to a CSI action recognition end, so that the CSI action recognition end obtains channel state data of the enhanced WIFI detection signal after receiving the enhanced WIFI detection signal, and the channel state data is input into an action recognition model so as to obtain an action recognition result of the target human body based on the action recognition model and the channel state data;
the motion recognition model is obtained by training the following method:
acquiring a training data set, wherein the training data set comprises a plurality of data subsets corresponding to different human body actions, and each data subset comprises a plurality of channel state sample data corresponding to enhanced WIFI detection sample signals;
performing data preprocessing on the training data set to obtain a preprocessed training data set, wherein the preprocessing comprises data filling processing, phase correction processing and denoising processing;
And training a double-flow convolution enhanced neural network model by taking the preprocessed training data set as input and taking an action recognition result corresponding to each piece of channel state sample data in the preprocessed training data set as output so as to obtain the action recognition model after training is finished.
6. WIFI signal action recognition device based on RIS is supplementary, characterized in that includes:
the first receiving unit is used for receiving an enhanced WIFI detection signal which is sent by the RIS reflecting end and used for detecting the action of a target human body, wherein the enhanced WIFI detection signal is obtained by determining an optimal beam forming codebook of an initial WIFI detection signal after the RIS reflecting end receives the initial WIFI detection signal sent by the WIFI signal transmitting end and carrying out beam forming on the initial WIFI detection signal by utilizing the optimal beam forming codebook, and the target human body is positioned between the WIFI signal transmitting end and the WIFI signal action recognition device based on RIS assistance;
the data analysis unit is used for obtaining channel state data of the enhanced WIFI detection signal based on the enhanced WIFI detection signal, wherein each piece of channel data in the channel state data comprises amplitude data and phase data;
The motion recognition unit is used for inputting the channel state data into a motion recognition model so as to perform motion recognition on the target human body based on the motion recognition model and the channel state data, and obtain a motion recognition result of the target human body;
before receiving the enhanced WIFI detection signal which is sent by the RIS reflecting end and is used for detecting the action of the target human body, the method further comprises the following steps:
acquiring a training data set, wherein the training data set comprises a plurality of data subsets corresponding to different human body actions, and each data subset comprises a plurality of channel state sample data corresponding to enhanced WIFI detection sample signals;
performing data preprocessing on the training data set to obtain a preprocessed training data set, wherein the preprocessing comprises data filling processing, phase correction processing and denoising processing;
and training a double-flow convolution enhanced neural network model by taking the preprocessed training data set as input and taking an action recognition result corresponding to each piece of channel state sample data in the preprocessed training data set as output so as to obtain the action recognition model after training is finished.
7. WIFI signal action recognition device based on RIS is supplementary, characterized in that includes:
The second receiving unit is used for receiving an initial WIFI detection signal which is sent by the WIFI signal transmitting end and used for detecting the action of a target human body, and determining an optimal beam forming codebook of the initial WIFI detection signal, wherein the WIFI signal action recognition device based on RIS assistance is arranged on a reflection link of the initial WIFI detection signal;
the beam forming unit is used for carrying out beam forming on the initial WIFI detection signal by utilizing the optimal beam forming codebook to obtain an enhanced WIFI detection signal;
the sending unit is used for sending the enhanced WIFI detection signal to the CSI action recognition end so that the CSI action recognition end can obtain channel state data of the enhanced WIFI detection signal after receiving the enhanced WIFI detection signal, and inputting the channel state data into the action recognition model so as to obtain an action recognition result of the target human body based on the action recognition model and the channel state data;
the motion recognition model is obtained by training the following method:
acquiring a training data set, wherein the training data set comprises a plurality of data subsets corresponding to different human body actions, and each data subset comprises a plurality of channel state sample data corresponding to enhanced WIFI detection sample signals;
Performing data preprocessing on the training data set to obtain a preprocessed training data set, wherein the preprocessing comprises data filling processing, phase correction processing and denoising processing;
and training a double-flow convolution enhanced neural network model by taking the preprocessed training data set as input and taking an action recognition result corresponding to each piece of channel state sample data in the preprocessed training data set as output so as to obtain the action recognition model after training is finished.
8. WIFI signal action recognition device based on RIS is supplementary, characterized in that includes: the system comprises a memory, a processor and a transceiver which are connected in sequence in communication, wherein the memory is used for storing a computer program, the transceiver is used for receiving and transmitting messages, and the processor is used for reading the computer program and executing the WIFI signal action recognition method based on RIS assistance according to any one of claims 1-4 or claim 5.
9. A storage medium having instructions stored thereon that, when executed on a computer, perform the RIS-based WIFI signal action recognition method of any one of claims 1 to 4 or claim 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211401369.4A CN115913415B (en) | 2022-11-09 | 2022-11-09 | WIFI signal action recognition method and device based on RIS assistance and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211401369.4A CN115913415B (en) | 2022-11-09 | 2022-11-09 | WIFI signal action recognition method and device based on RIS assistance and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115913415A CN115913415A (en) | 2023-04-04 |
| CN115913415B true CN115913415B (en) | 2024-02-02 |
Family
ID=86475442
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211401369.4A Active CN115913415B (en) | 2022-11-09 | 2022-11-09 | WIFI signal action recognition method and device based on RIS assistance and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115913415B (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105933080A (en) * | 2016-01-20 | 2016-09-07 | 北京大学 | Fall-down detection method and system |
| CN109902554A (en) * | 2019-01-09 | 2019-06-18 | 天津大学 | A kind of recognition methods of the sign language based on commercial Wi-Fi |
| KR20210077516A (en) * | 2019-12-17 | 2021-06-25 | 서울대학교산학협력단 | Method of Recognizing Daily Activities of Living-alone Person Using Channel State Information between WiFi Transmitter and WiFi Receiver |
| WO2021160189A1 (en) * | 2020-02-14 | 2021-08-19 | 重庆邮电大学 | Csi method for recognizing human fall in wi-fi interference environment |
| CN113378718A (en) * | 2021-06-10 | 2021-09-10 | 中国石油大学(华东) | Action identification method based on generation of countermeasure network in WiFi environment |
| CN113572546A (en) * | 2021-06-16 | 2021-10-29 | 西北大学 | Method for recognizing human body activity by using DenseNet based on CSI signal |
| WO2021239067A1 (en) * | 2020-05-29 | 2021-12-02 | 华为技术有限公司 | Cooperative communication method and communication apparatus |
| CN113743374A (en) * | 2021-10-11 | 2021-12-03 | 南京邮电大学 | Personnel identity identification method based on channel state information respiration perception |
| CN114024630A (en) * | 2021-11-24 | 2022-02-08 | 北京大学 | Gesture recognition method based on wireless signals |
| CN114241491A (en) * | 2021-11-22 | 2022-03-25 | 南京邮电大学 | Handwritten letter recognition method based on lightweight deep learning network |
| CN114973330A (en) * | 2022-06-16 | 2022-08-30 | 深圳大学 | Cross-scene robust personnel fatigue state wireless detection method and related equipment |
| CN114999643A (en) * | 2022-06-01 | 2022-09-02 | 暨南大学 | WiFi-based intelligent monitoring method for old people |
-
2022
- 2022-11-09 CN CN202211401369.4A patent/CN115913415B/en active Active
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105933080A (en) * | 2016-01-20 | 2016-09-07 | 北京大学 | Fall-down detection method and system |
| CN109902554A (en) * | 2019-01-09 | 2019-06-18 | 天津大学 | A kind of recognition methods of the sign language based on commercial Wi-Fi |
| KR20210077516A (en) * | 2019-12-17 | 2021-06-25 | 서울대학교산학협력단 | Method of Recognizing Daily Activities of Living-alone Person Using Channel State Information between WiFi Transmitter and WiFi Receiver |
| WO2021160189A1 (en) * | 2020-02-14 | 2021-08-19 | 重庆邮电大学 | Csi method for recognizing human fall in wi-fi interference environment |
| WO2021239067A1 (en) * | 2020-05-29 | 2021-12-02 | 华为技术有限公司 | Cooperative communication method and communication apparatus |
| CN113747465A (en) * | 2020-05-29 | 2021-12-03 | 华为技术有限公司 | Cooperative communication method and communication device |
| CN113378718A (en) * | 2021-06-10 | 2021-09-10 | 中国石油大学(华东) | Action identification method based on generation of countermeasure network in WiFi environment |
| CN113572546A (en) * | 2021-06-16 | 2021-10-29 | 西北大学 | Method for recognizing human body activity by using DenseNet based on CSI signal |
| CN113743374A (en) * | 2021-10-11 | 2021-12-03 | 南京邮电大学 | Personnel identity identification method based on channel state information respiration perception |
| CN114241491A (en) * | 2021-11-22 | 2022-03-25 | 南京邮电大学 | Handwritten letter recognition method based on lightweight deep learning network |
| CN114024630A (en) * | 2021-11-24 | 2022-02-08 | 北京大学 | Gesture recognition method based on wireless signals |
| CN114999643A (en) * | 2022-06-01 | 2022-09-02 | 暨南大学 | WiFi-based intelligent monitoring method for old people |
| CN114973330A (en) * | 2022-06-16 | 2022-08-30 | 深圳大学 | Cross-scene robust personnel fatigue state wireless detection method and related equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115913415A (en) | 2023-04-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210158533A1 (en) | Image processing method and apparatus, and storage medium | |
| US11551333B2 (en) | Image reconstruction method and device | |
| US20200242812A1 (en) | Encoding and decoding a stylized custom graphic | |
| US11200395B2 (en) | Graphic code recognition method and apparatus, terminal, and storage medium | |
| JP7331171B2 (en) | Methods and apparatus for training image recognition models, methods and apparatus for recognizing images, electronic devices, storage media, and computer programs | |
| CN111401406B (en) | Neural network training method, video frame processing method and related equipment | |
| CN111047602A (en) | Image segmentation method and device and terminal equipment | |
| CN114611705A (en) | Data processing method, training method for machine learning, and related device and equipment | |
| US20190385073A1 (en) | Visual recognition via light weight neural network | |
| CN111028237A (en) | Image segmentation method and device and terminal equipment | |
| US11948088B2 (en) | Method and apparatus for image recognition | |
| US20220327834A1 (en) | Action localization method, device, electronic equipment, and computer-readable storage medium | |
| CN112487217A (en) | Cross-modal retrieval method, device, equipment and computer-readable storage medium | |
| WO2024083121A1 (en) | Data processing method and apparatus | |
| CN115343704A (en) | Gesture recognition method of FMCW millimeter wave radar based on multi-task learning | |
| US20220058824A1 (en) | Method and apparatus for image labeling, electronic device, storage medium, and computer program | |
| CN111325709A (en) | Wireless capsule endoscope image detection system and detection method | |
| CN112232804B (en) | Code scanning payment method and device and electronic equipment | |
| CN116580290B (en) | Unmanned aerial vehicle identification method, unmanned aerial vehicle identification device and storage medium | |
| CN115913415B (en) | WIFI signal action recognition method and device based on RIS assistance and storage medium | |
| Wang et al. | A COVID‐19 CXR image recognition method based on MSA‐DDCovidNet | |
| Huang et al. | Swin-YOLO for Concealed Object Detection in Millimeter Wave Images | |
| CN111950637B (en) | Ultraviolet detection method, device, skin detector and readable storage medium | |
| CN111429414B (en) | Artificial intelligence-based focus image sample determination method and related device | |
| CN116482680B (en) | Body interference identification method, device, system 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |