CN113499064A - Wi-Fi perception human body tumbling detection method and system in bathroom scene - Google Patents
Wi-Fi perception human body tumbling detection method and system in bathroom scene Download PDFInfo
- Publication number
- CN113499064A CN113499064A CN202110766943.5A CN202110766943A CN113499064A CN 113499064 A CN113499064 A CN 113499064A CN 202110766943 A CN202110766943 A CN 202110766943A CN 113499064 A CN113499064 A CN 113499064A
- Authority
- CN
- China
- Prior art keywords
- data
- amplitude
- csi
- formula
- module
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 61
- 230000008447 perception Effects 0.000 title claims abstract description 22
- 238000001914 filtration Methods 0.000 claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 19
- 238000003062 neural network model Methods 0.000 claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims abstract description 12
- 230000006870 function Effects 0.000 claims description 29
- 238000000034 method Methods 0.000 claims description 26
- 238000011176 pooling Methods 0.000 claims description 20
- 238000004590 computer program Methods 0.000 claims description 16
- 238000010606 normalization Methods 0.000 claims description 12
- 239000013598 vector Substances 0.000 claims description 10
- 230000004927 fusion Effects 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000012549 training Methods 0.000 claims description 6
- 235000014121 butter Nutrition 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 5
- 238000003860 storage Methods 0.000 claims description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 238000011160 research Methods 0.000 description 9
- 238000013527 convolutional neural network Methods 0.000 description 8
- 238000013528 artificial neural network Methods 0.000 description 7
- 230000006399 behavior Effects 0.000 description 7
- 230000000306 recurrent effect Effects 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 4
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 230000004438 eyesight Effects 0.000 description 2
- 238000007667 floating Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 230000004308 accommodation Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000003287 bathing Methods 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000004040 coloring Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000005021 gait Effects 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 230000000474 nursing effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 239000007921 spray Substances 0.000 description 1
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
- A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
- A61B5/1116—Determining posture transitions
- A61B5/1117—Fall detection
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7221—Determining signal validity, reliability or quality
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/04—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons
- G08B21/0407—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons based on behaviour analysis
- G08B21/043—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons based on behaviour analysis detecting an emergency event, e.g. a fall
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/02—Alarms for ensuring the safety of persons
- G08B21/04—Alarms for ensuring the safety of persons responsive to non-activity, e.g. of elderly persons
- G08B21/0438—Sensor means for detecting
- G08B21/0461—Sensor means for detecting integrated or attached to an item closely associated with the person but not worn by the person, e.g. chair, walking stick, bed sensor
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Surgery (AREA)
- Veterinary Medicine (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Public Health (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Animal Behavior & Ethology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Emergency Management (AREA)
- Gerontology & Geriatric Medicine (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Fuzzy Systems (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Psychology (AREA)
- Social Psychology (AREA)
- Dentistry (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a Wi-Fi perception human body fall detection method and system in a bathroom scene, wherein the system comprises two modules of data processing and fall detection: the data processing module mainly acquires and preprocesses sensing data, and comprises three parts, namely data acquisition, data reconstruction and data filtering; firstly, acquiring original channel state information data under a bathroom environment monitored by a Wi-Fi device of a single transmitter-single receiver and extracting an amplitude frame; secondly, reconstructing the one-dimensional time sequence data into a two-dimensional matrix form, and filtering the two-dimensional matrix form; and the fall detection module inputs the processed sensing data into a designed neural network model to extract features, and then calculates a monitoring result through a Softmax layer.
Description
Technical Field
The invention relates to the field of behavior recognition, in particular to a Wi-Fi perception human body tumbling detection method and system in a bathroom scene.
Background
In recent years, old people often fall down due to wet and slippery ground when taking a bath, and the old people living alone cannot seek help in time after falling down, so that damage to a greater extent is caused.
At present, a plurality of human body falling detection methods exist, and the human body falling detection can be divided into 3 types from the angle of signal acquisition: vision-based fall detection, acoustic-based fall detection, and wearable sensor-based fall detection. The falling detection based on vision is that a picture of human motion is obtained through a camera, and the falling image features are extracted through an image processing algorithm to judge whether the falling occurs, so that the accuracy is high. The method is based on acoustic fall detection, an audio signal generated when a person falls is detected to serve as a criterion for falling, and the misjudgment rate is high due to interference of underwater sound and other sounds in a bathroom. The fall detection based on the wearable sensor is realized by the miniature equipment made of the miniature sensor and the controller, the fall detection is realized by algorithm processing, the product is easy to carry outdoors but inconvenient to carry in a bathroom, and is easy to be interfered by environmental factors, and the detection precision is low.
Wi-Fi perception is a new direction for human body fall detection and identification research by virtue of the advantages of non-contact, easiness in deployment, no influence of light rays, wider perception range and the like. At present, although some achievements have been made in the fall detection identification research based on Wi-Fi perception, some drawbacks also exist.
Scheme 1: method for carrying out Wi-Fi perception human body falling detection and identification by using recurrent neural network
A Recurrent Neural Network (RNN) is a type of Recurrent Neural Network (Recurrent Neural Network) in which sequence data is input, recursion is performed in the direction of evolution of the sequence, and all nodes (Recurrent units) are connected in a chain. At present, in the field of Wi-Fi perception human body fall detection and recognition, research on learning by utilizing a recurrent neural network has been actively developed. The method is represented by the RRS-RNN method. In order to further improve the detection performance, an RNN fall detection method (abbreviated as rrn-RNN) based on a Random Reset Strategy (Random Reset Strategy-RRS) is proposed to reduce the influence of wrong history information in the RNN model on the current behavior judgment. The experimental result shows that the RNN model can effectively realize fall detection, and the identification accuracy rate reaches 88.38%; the identification performance is further improved based on the RRS-RNN method, and the identification accuracy reaches 88.92%.
The method has the following defects: in the existing Wi-Fi perception human body tumble detection and identification research based on the recurrent neural network, most of achievement models are too complex in design, large in parameter calculation amount and low in accuracy, and an application scene is not clearly provided.
Scheme 2: method for detecting and identifying Wi-Fi (wireless fidelity) perception human body tumble by utilizing convolutional neural network
The convolutional neural network is a deep learning method which is started in recent years, and becomes a classification method widely used in various fields by virtue of strong feature extraction capability of the convolutional neural network on multi-dimensional data and automatic parameter adjustment characteristics based on back propagation. In recent years, some research methods based on convolutional neural network algorithm, specifically BLSTM and CBMR methods, have appeared in the field of behavior recognition research of Wi-Fi perception human body fall detection. The BLSTM method processes the collected CSI in a time sequence flow mode, and then a bidirectional LSTM (long-short-term memory) method based on an attention mechanism is adopted to give different weights to different learned characteristics, so that action recognition based on Wi-Fi perception is completed, and a good recognition effect is obtained. By using the convolutional neural network, the method achieves better effects in the aspects of data preprocessing and gait feature extraction.
The method has the following defects: in the existing Wi-Fi perception human body falling detection and recognition research based on the convolutional neural network, most of achievement models are formed by modifying a classical convolutional neural network or a cyclic neural network, the parameter quantity is large, great calculation force needs to be consumed in training, and the special scene that the Wi-Fi perception human body falling detection and recognition research can be applicable to a bathroom is not clearly provided.
Disclosure of Invention
In order to solve the above problems, it is necessary to provide a lightweight Wi-Fi aware human fall detection method and system applied to a bathroom scene.
The invention provides a Wi-Fi perception human body tumbling detection method in a bathroom scene, which comprises the following steps:
collecting original channel state information data H (f) in bathroom environment monitored by Wi-Fi device of single transmitter-single receiverk) Expressed as formula (1);
wherein, H (f)k) CSI, | H (f) of k-th subcarrier of each antenna pair representing Wi-Fi devicek) I and H (f)k) Respectively representing amplitude and phase;
after extracting the amplitude information of the original signal, expressing the CSI sequence of the original signal by using one-dimensional data shown in formula (2);
wherein HkRepresenting the amplitude sequence values over a period of N,indicating the amplitude of the kth subcarrier at time n;
reconstructing the one-dimensional amplitude sequence in the N time period into a two-dimensional CSI amplitude data matrix shown in a formula (3);
wherein HnkIs the CSI amplitude value of the kth subcarrier at time n;
filtering the obtained two-dimensional CSI amplitude data matrix, wherein the sampling of the CSI amplitude data is performedFrequency FsSet to 1000hz, the frequency f of the CSI time series variation is set to 40hz, a butterworth filter is constructed:
calculating the cut-off frequency w from equation (4)c;
Calling a button function in Matlab to calculate coefficient vectors b and a of a button word filter, as shown in formula (5), wherein N is the order of the filter, and low is low-pass filtering;
[b,a]=butter(N,wc,′low′); (5)
constructing a Butterworth filter by using a filtfiltfilt function in Matlab, as shown in a formula (6);
Signal_Filter=filtfilt(b,a,Signal); (6)
b and a are respectively expressed as coefficient vectors of a Butterworth filter system function numerator and a denominator polynomial, and Signal expresses the obtained two-dimensional CSI amplitude data;
and outputting the result after filtering processing to a deep neural network model, performing global average pooling on the result after convolution, inputting the result to a full-connection layer, and classifying the result by using Softmax to obtain a result of fall detection.
Based on the above, the deep neural network model comprises two ConvBN sub-modules for primary feature extraction and two CarConvBNMax modules for feature fusion;
the convBN submodule includes three convolutional layers Conv2D and three Batch Normalization layers, the convolutional layers Conv2D alternating with the Batch Normalization layers;
the CarConvBNMax module comprises a feature fusion layer Concatenate, a convolution layer Conv2D, a Batch standardization layer Batch standardization, and a maximum pooling layer Max Pooling;
two ConvBN submodules are connected in parallel to one convolutional layer Conv2D to the input layer;
each ConvBN submodule is connected with the convolutional layer Conv2D and then connected with a CarConvBNMax module;
two CarConvBNMax modules are connected in parallel and then serve as the output of the deep neural network model.
Based on the above, the expression formula of the Softmax function of the full connection layer is as follows:
in the formula, r is the classification number of y, zyThe result after global mean pooling is shown, and P (y | X) is the posterior probability of the prediction belonging to y category from model input X;
after the posterior probability is obtained, performing minimum training on the loss function by using an Adam optimizer;
the loss function calculation is shown in equation (8):
the invention provides a Wi-Fi perception human body tumbling detection system in a bathroom scene, which comprises: the system comprises a data processing module and a fall detection module, wherein the data processing module comprises a data acquisition module, a data reconstruction module and a data filtering module which are sequentially connected;
the data acquisition module is used for acquiring original channel state information data H (f) monitored by a Wi-Fi device of a single transmitter-single receiver in a bathroom environmentk) Expressed as formula (1);
wherein, H (f)k) CSI, | H (f) of k-th subcarrier of each antenna pair representing Wi-Fi devicek) I and H (f)k) Respectively representing amplitude and phase;
and is also used for representing the CSI sequence of the original signal by one-dimensional data shown in formula (2) after extracting the amplitude information of the original signal;
wherein HkRepresenting the amplitude sequence values over a period of N,indicating the amplitude of the kth subcarrier at time n;
the data reconstruction module is used for reconstructing the one-dimensional amplitude sequence in the N time period into a two-dimensional CSI amplitude data matrix shown in a formula (3);
wherein HnkIs the CSI amplitude value of the kth subcarrier at time n;
the data filtering module is used for filtering the obtained two-dimensional CSI amplitude data matrix, wherein the sampling frequency F of the CSI amplitude data is usedsSet to 1000hz, the frequency f of the CSI time series variation is set to 40hz, a butterworth filter is constructed:
calculating the cut-off frequency w from equation (4)c;
Calling a button function in Matlab to calculate coefficient vectors b and a of a button word filter, as shown in formula (5), wherein N is the order of the filter, and low is low-pass filtering;
[b,a]=butter(N,wc,′low′); (5)
constructing a Butterworth filter by using a filtfiltfilt function in Matlab, as shown in a formula (6);
Signal_Filter=filtfilt(b,a,Signal); (6)
b and a are respectively expressed as coefficient vectors of a Butterworth filter system function numerator and a denominator polynomial, and Signal expresses the obtained two-dimensional CSI amplitude data;
and the falling detection module is used for outputting the result after filtering processing to the deep neural network model, performing global average pooling on the result after convolution, inputting the result to the full connection layer, and classifying the result by using Softmax to obtain the falling detection result.
The third aspect of the present invention provides a terminal, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor implements the step of Wi-Fi perception human body fall detection in a bathroom scene when executing the computer program, and obtains a fall detection result.
In a fourth aspect, the present invention provides a computer readable storage medium having stored thereon computer instructions which, when executed by a processor, perform the steps of Wi-Fi aware human fall detection in a bathroom scenario.
Compared with the prior art, the invention has prominent substantive characteristics and remarkable progress, particularly:
(1) according to the invention, the one-dimensional time sequence data is reconstructed into the two-dimensional multi-carrier data, and then Butterworth filtering is carried out, so that the data characteristic accommodation capacity is enhanced, and the environmental noise is effectively removed.
(2) The deep neural network model can remarkably reduce the parameter scale of the existing behavior recognition model on the basis of remarkably improving the recognition accuracy and generalization capability of the falling behavior of the human body in a Wi-Fi scene.
(3) According to the method, rich and detailed feature extraction is performed by using a Batch Normalization Layer (Batch), a connection Layer (correlation Layer) and a maximum pooling Layer (Max Pooling Layer) in the convolutional neural network, and the number of model parameters and the number of floating point operations are reduced.
Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
The above and/or additional aspects and advantages of the present invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:
FIG. 1 is an overall flow chart of the method of the present invention.
FIG. 2 is a two-dimensional frequency energy comparison graph before and after filtering according to the method of the present invention.
FIG. 3 is a diagram of a deep neural network model architecture in the method of the present invention.
Detailed Description
In order that the above objects, features and advantages of the present invention can be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings. It should be noted that the embodiments and features of the embodiments of the present application may be combined with each other without conflict.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those specifically described herein, and therefore the scope of the present invention is not limited by the specific embodiments disclosed below.
Example 1
As shown in fig. 1-3, the present embodiment provides a Wi-Fi-aware human fall detection system (wishall) in a bathroom scene, including: the system comprises a data processing module and a fall detection module, wherein the data processing module comprises a data acquisition module, a data reconstruction module and a data filtering module which are sequentially connected;
the Wi-Fi perception human body tumbling detection method in the bathroom scene comprises the following steps:
first, a commercial Wi-Fi device generally adopts a mimo (multiple Input multiple output) technology, and CSI data of each antenna pair includes a plurality of pieces of subcarrier information. Acquiring original channel state information data H (f) in bathroom environment monitored by Wi-Fi device of single transmitter-single receiver by data processing modulek) Expressed as formula (1);
wherein, H (f)k) CSI, H (f) representing the kth subcarrierk) I and H (f)k) Representing amplitude and phase respectively, the method uses only the amplitude information therein, since the phase information is susceptible to interference.
Secondly, after extracting the amplitude information of the original signal, expressing the CSI sequence of the original signal by using one-dimensional data shown in formula (2);
wherein HkRepresenting the amplitude sequence values over a period of N,indicating the amplitude of the kth subcarrier at time n;
in order to extract the characteristic information in the subcarriers and among the subcarriers at the same time, the data reconstruction module reconstructs the one-dimensional time sequence data in the N time period into a two-dimensional matrix form shown in a formula (3);
wherein HnkIs the CSI amplitude value of the kth subcarrier at time n. According to the data reconstruction mode shown in the formula (3), a corresponding frequency energy diagram can be generated through proper coloring design. By using the characteristic reconstruction strategy of the frequency energy diagram, the Wi-Fi signal energy attenuation condition can be visually displayed, and more CSI data characteristic information can be contained.
Thirdly, due to multipath effects, the acquired sensing data may contain environmental noise, and the identification performance of the model is affected by ineffective processing. Thus, the obtained two-dimensional CSI amplitude data matrix is data filtered by a butterworth filter constructed by the data filtering module. Since the dynamic noise source of a bathroom scene is primarily shower spray, the frequency response of the Butterworth filter passbandThe curve is smooth, and bathroom noise can be effectively filtered. Specifically, the sampling frequency F of the CSI amplitude datasSet to 1000hz, the frequency f of the CSI time series variation is set to 40hz, a butterworth filter is constructed:
calculating the cut-off frequency w from equation (4)c;
Calling a button function in Matlab to calculate coefficient vectors b and a of a button word filter, as shown in formula (5), wherein N is the order of the filter, and low is low-pass filtering;
[b,a]=butter(N,wc,′low′); (5)
constructing a Butterworth filter by using a filtfiltfilt function in Matlab, as shown in a formula (6);
Signal_Filter=filtfilt(b,a,Signal); (6)
wherein, b and a are respectively expressed as coefficient vectors of a Butterworth filter system function numerator and a denominator polynomial, and Signal expresses reconstructed data, namely the obtained two-dimensional CSI amplitude data.
And then, in a falling detection module, outputting the result after filtering processing to a deep neural network model, performing global average pooling on the result after convolution, inputting the result to a full connection layer, and classifying the result by using Softmax to obtain a falling detection result.
Specifically, the expression formula of the Softmax function of the full connection layer is as follows:
in the formula, r is the classification number of y, zyThe result after global mean pooling is shown, and P (y | X) is the posterior probability of the prediction belonging to y category from model input X;
after the posterior probability is obtained, performing minimum training on the loss function by using an Adam optimizer;
the loss function calculation is shown in equation (8):
example 2
This example differs from example 1 in that: the embodiment provides a specific deep neural network model.
The deep neural network model comprises two ConvBN sub-modules for primary feature extraction and two CarConvBNMax modules for feature fusion;
the ConvBN submodule comprises three convolution layers Conv2D and three Batch Normalization layers Batch Normalization, and a mechanism that the convolution layers Conv2D and the Batch Normalization layers Batch Normalization appear alternately is designed, so that preliminary features and overall features of sensing data can be fully extracted; by using Batch Normalization, the model can be quickly converged, training data can be disturbed, and the use efficiency of the data is improved;
the CarConvBNMax module comprises a feature fusion layer Concatenate, a convolution layer Conv2D, a Batch standardization layer Batch standardization, and a maximum pooling layer Max Pooling;
two ConvBN submodules are connected in parallel to one convolutional layer Conv2D to the input layer;
each ConvBN submodule is connected with the convolutional layer Conv2D and then connected with a CarConvBNMax module;
two CarConvBNMax modules are connected in parallel and then serve as the output of the deep neural network model.
The CarConvBNMax sub-module is arranged between the ConvBN sub-module and the output, and plays a role in starting and stopping in a model. Compared with the ConvBN submodule, the CarConvBNMax submodule is added with a feature fusion layer, i.e. Concatenate and a maximum pooling layer, so that the functions of the ConvBN submodule are enriched, extracted feature information is refined, and the classification precision is improved.
The deep neural network model can remarkably reduce the parameter scale of the existing behavior recognition model on the basis of remarkably improving the recognition accuracy and generalization capability of the falling behavior of the human body in a Wi-Fi scene.
The lightweight and identification accuracy of the model can be obtained by comparing and analyzing with other common models, and the evaluation indexes include running time (Duration), model parameters (parameters), and floating point operations (FLOPs).
Example 3
The embodiment provides a terminal, which comprises a memory, a processor and a computer program which is stored in the memory and can run on the processor, wherein the processor executes the computer program to realize the steps of the Wi-Fi perception human body fall detection method in the bathroom scene described in the embodiments 1 and 2, and obtain the fall detection result.
In particular, the terminal may be used as a nursing terminal. At present, China has already stepped into an aging society. Due to the limited medical resources, the home care of the elderly becomes very important, and the related research is also the current focus. The terminal is used for home care, and can realize the non-inductive intelligent monitoring of the old through the home Wi-Fi, particularly the bathing condition of the old living alone.
It should be understood that in the present embodiment, the Processor may be a Central Processing Unit (CPU), and the Processor may also be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other Programmable logic devices, discrete Gate or transistor logic devices, discrete hardware components, and the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory may include both read-only memory and random access memory, and provides instructions and data to the processor. Some or all of the memory may also include non-volatile random access memory. For example, the memory may also store device type information.
The memory stores a computer program that is executable on the processor. The processor implements the steps in the Wi-Fi aware human fall detection method and system embodiments when executing the computer program. Or, the processor implements the functions of the units in the Wi-Fi aware human fall detection method and system embodiments when executing the computer program.
Example 4
The present embodiments provide a computer readable storage medium having stored thereon computer instructions that, when executed by a processor, implement the above-described lightweight Wi-Fi aware human fall detection method and system applied to bathroom scenes.
The present embodiment provides a computer program product, which when running on a terminal device, enables the terminal device to implement the steps of the Wi-Fi aware human body fall detection method in the bathroom scene described in embodiments 1 and 2 when executed.
In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and reference may be made to the related descriptions of other embodiments for parts that are not described or illustrated in a certain embodiment.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus/terminal and method may be implemented in other ways. For example, the above-described device/terminal embodiments are merely illustrative, and for example, the division of the above-described modules is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated module may be stored in a computer-readable storage medium if it is implemented in the form of a software functional unit and sold or used as a separate product. Based on such understanding, all or part of the flow in the method of the embodiments described above may be implemented by a computer program, which may be stored in a computer-readable storage medium and can implement the steps of the embodiments of the methods described above when the computer program is executed by a processor. The computer program includes computer program code, and the computer program code may be in a source code form, an object code form, an executable file or some intermediate form.
The computer readable medium may include: any entity or device capable of carrying the above-described computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier signal, telecommunications signal, software distribution medium, and the like.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (8)
1. A Wi-Fi perception human body tumbling detection method in a bathroom scene is characterized by comprising the following steps:
collecting original channel state information data H (f) in bathroom environment monitored by Wi-Fi device of single transmitter-single receiverk) Expressed as formula (1);
wherein, H (f)k) CSI, | H (f) of k-th subcarrier of each antenna pair representing Wi-Fi devicek) I and H (f)k) Respectively representing amplitude and phase;
after extracting the amplitude information of the original signal, expressing the CSI sequence of the original signal by using one-dimensional data shown in formula (2);
wherein HkRepresenting the amplitude sequence values over a period of N,indicating the amplitude of the kth subcarrier at time n;
reconstructing the one-dimensional amplitude sequence in the N time period into a two-dimensional CSI amplitude data matrix shown in a formula (3);
wherein HnkIs the CSI amplitude value of the kth subcarrier at time n;
filtering the obtained two-dimensional CSI amplitude data matrix, wherein the sampling frequency F of the CSI amplitude data issSet to 1000hz, the frequency f of the CSI time series variation is set to 40hz, a butterworth filter is constructed:
calculating the cut-off frequency w from equation (4)c;
Calling a button function in Matlab to calculate coefficient vectors b and a of a button word filter, as shown in formula (5), wherein N is the order of the filter, and low is low-pass filtering;
[b,a]=butter(N,wc,′low′); (5)
constructing a Butterworth filter by using a filtfiltfilt function in Matlab, as shown in a formula (6);
Signal_Filter=filtfilt(b,a,Signal); (6)
b and a are respectively expressed as coefficient vectors of a Butterworth filter system function numerator and a denominator polynomial, and Signal expresses the obtained two-dimensional CSI amplitude data;
and outputting the result after filtering processing to a deep neural network model, performing global average pooling on the result after convolution, inputting the result to a full-connection layer, and classifying the result by using Softmax to obtain a result of fall detection.
2. The method of claim 1, wherein the Wi-Fi aware human fall detection method in a bathroom scenario comprises:
the deep neural network model comprises a ConvBN submodule for primary feature extraction and a CarConvBNMax module for feature fusion;
the convBN submodule includes a convolution layer Conv2D alternating with a Batch Normalization layer, the convolution layer Conv2D alternating with a Batch Normalization layer;
the CarConvBNMax module comprises a feature fusion layer Concatenate, a convolution layer Conv2D, a Batch standardization layer Batch standardization, and a maximum pooling layer Max Pooling;
the CarConvBNMax sub-module is interposed between the ConvBN sub-module and the output section.
3. The method of claim 1, wherein the Wi-Fi aware human fall detection method in a bathroom scenario comprises:
the expression formula of the Softmax function of the full connection layer is as follows:
in the formula, r is the classification number of y, zyThe result after global mean pooling is shown, and P (y | X) is the posterior probability of the prediction belonging to y category from model input X;
after the posterior probability is obtained, performing minimum training on the loss function by using an Adam optimizer;
the loss function calculation is shown in equation (8):
4. a Wi-Fi perception human body fall detection system in a bathroom scene, comprising: the system comprises a data processing module and a fall detection module, wherein the data processing module comprises a data acquisition module, a data reconstruction module and a data filtering module which are sequentially connected;
the data acquisition module is used for acquiring original channel state information data H (f) monitored by a Wi-Fi device of a single transmitter-single receiver in a bathroom environmentk) Expressed as formula (1);
wherein, H (f)k) CSI, | H (f) of k-th subcarrier of each antenna pair representing Wi-Fi devicek) I and H (f)k) Respectively representing amplitude and phase;
and is also used for representing the CSI sequence of the original signal by one-dimensional data shown in formula (2) after extracting the amplitude information of the original signal;
wherein HkRepresenting the amplitude sequence values over a period of N,indicating the amplitude of the kth subcarrier at time n;
the data reconstruction module is used for reconstructing the one-dimensional amplitude sequence in the N time period into a two-dimensional CSI amplitude data matrix shown in a formula (3);
wherein HnkIs the CSI amplitude value of the kth subcarrier at time n;
the data filtering module is used for filtering the obtained two-dimensional CSI amplitude data matrix, wherein the sampling frequency F of the CSI amplitude data is usedsSet to 1000hz, the frequency f of the CSI time series variation is set to 40hz, a butterworth filter is constructed:
calculating the cut-off frequency w from equation (4)c;
Calling a button function in Matlab to calculate coefficient vectors b and a of a button word filter, as shown in formula (5), wherein N is the order of the filter, and low is low-pass filtering;
[b,a]=butter(N,wc,′low′); (5)
constructing a Butterworth filter by using a filtfiltfilt function in Matlab, as shown in a formula (6);
Signal_Filter=filtfilt(b,a,Signal); (6)
b and a are respectively expressed as coefficient vectors of a Butterworth filter system function numerator and a denominator polynomial, and Signal expresses the obtained two-dimensional CSI amplitude data;
and the falling detection module is used for outputting the result after filtering processing to the deep neural network model, performing global average pooling on the result after convolution, and finally inputting the result to the full connection layer for classification by using Softmax, so that the falling detection result is obtained.
5. The Wi-Fi aware human fall detection system in a bathroom scene of claim 4, characterized in that:
the deep neural network model comprises two ConvBN sub-modules for primary feature extraction and two CarConvBNMax modules for feature fusion;
the convBN submodule includes three convolutional layers Conv2D and three Batch Normalization layers, the convolutional layers Conv2D alternating with the Batch Normalization layers;
the CarConvBNMax module comprises a feature fusion layer Concatenate, a convolution layer Conv2D, a Batch standardization layer Batch standardization, and a maximum pooling layer Max Pooling;
two ConvBN submodules are connected in parallel to one convolutional layer Conv2D to the input layer;
each ConvBN submodule is connected with the convolutional layer Conv2D and then connected with a CarConvBNMax module;
two CarConvBNMax modules are connected in parallel and then serve as the output of the deep neural network model.
6. The Wi-Fi aware human fall detection system in a bathroom scene of claim 4, characterized in that:
the expression formula of the Softmax function of the full connection layer is as follows:
in the formula, r is the classification number of y, zyThe result after global mean pooling is shown, and P (y | X) is the posterior probability of the prediction belonging to y category from model input X;
after the posterior probability is obtained, performing minimum training on the loss function by using an Adam optimizer;
the loss function calculation is shown in equation (8):
7. a terminal comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that: the processor, when executing the computer program, performs the steps of the Wi-Fi aware human fall detection method in a bathroom scenario of any of claims 1 to 3, resulting in fall detection results.
8. A computer readable storage medium having stored thereon computer instructions, wherein the computer instructions, when executed by a processor, perform the steps of the Wi-Fi aware human fall detection method in a bathroom setting of any of claims 1-3.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110766943.5A CN113499064A (en) | 2021-07-07 | 2021-07-07 | Wi-Fi perception human body tumbling detection method and system in bathroom scene |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110766943.5A CN113499064A (en) | 2021-07-07 | 2021-07-07 | Wi-Fi perception human body tumbling detection method and system in bathroom scene |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113499064A true CN113499064A (en) | 2021-10-15 |
Family
ID=78011977
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110766943.5A Pending CN113499064A (en) | 2021-07-07 | 2021-07-07 | Wi-Fi perception human body tumbling detection method and system in bathroom scene |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113499064A (en) |
Citations (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103606248A (en) * | 2013-09-30 | 2014-02-26 | 广州市香港科大***研究院 | Automatic detection method and system for human body falling-over |
CN105933080A (en) * | 2016-01-20 | 2016-09-07 | 北京大学 | Fall-down detection method and system |
CN106772219A (en) * | 2017-03-08 | 2017-05-31 | 南京大学 | Indoor orientation method based on CSI signals |
CN107331136A (en) * | 2017-05-11 | 2017-11-07 | 深圳市斑点猫信息技术有限公司 | Indoor human body activity detection approach and system based on WiFi |
US20170354349A1 (en) * | 2016-06-10 | 2017-12-14 | The Regents Of The University Of California | Wifi-based person-identification technique for use in smart spaces |
CN107749143A (en) * | 2017-10-30 | 2018-03-02 | 安徽工业大学 | A kind of indoor occupant fall detection system and method through walls based on WiFi signal |
CN108614989A (en) * | 2017-12-20 | 2018-10-02 | 深圳大学 | A kind of old man's intelligent monitor method, system and medium based on WiFi |
CN108761542A (en) * | 2018-05-03 | 2018-11-06 | 西北大学 | A kind of passive type object detection method based on WiFi signal |
CN108833036A (en) * | 2018-05-31 | 2018-11-16 | 湖南大学 | Tumble detection method for human body |
CN109394229A (en) * | 2018-11-22 | 2019-03-01 | 九牧厨卫股份有限公司 | A kind of fall detection method, apparatus and system |
CN110288018A (en) * | 2019-06-24 | 2019-09-27 | 桂林电子科技大学 | A kind of WiFi personal identification method merging deep learning model |
EP3692898A1 (en) * | 2019-02-11 | 2020-08-12 | Nokia Technologies Oy | Sleep/motion determination based on wi-fi signals |
CN111556453A (en) * | 2020-04-27 | 2020-08-18 | 南京邮电大学 | Multi-scene indoor action recognition method based on channel state information and BilSTM |
CN111597877A (en) * | 2020-04-02 | 2020-08-28 | 浙江工业大学 | Fall detection method based on wireless signals |
CN111954250A (en) * | 2020-08-12 | 2020-11-17 | 郑州大学 | Lightweight Wi-Fi behavior sensing method and system |
CN112036433A (en) * | 2020-07-10 | 2020-12-04 | 天津城建大学 | CNN-based Wi-Move behavior sensing method |
CN112330924A (en) * | 2020-11-05 | 2021-02-05 | 北京理工大学 | Falling event detection method based on channel state information in indoor environment |
CN112346050A (en) * | 2020-10-23 | 2021-02-09 | 清华大学 | Fall detection method and system based on Wi-Fi equipment |
CN112545496A (en) * | 2020-11-09 | 2021-03-26 | 金茂智慧科技(广州)有限公司 | Wireless technology identification method for detecting human body tumble |
CN112869734A (en) * | 2021-01-11 | 2021-06-01 | 乐鑫信息科技(上海)股份有限公司 | Wi-Fi human body detection method and intelligent device |
-
2021
- 2021-07-07 CN CN202110766943.5A patent/CN113499064A/en active Pending
Patent Citations (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103606248A (en) * | 2013-09-30 | 2014-02-26 | 广州市香港科大***研究院 | Automatic detection method and system for human body falling-over |
CN105933080A (en) * | 2016-01-20 | 2016-09-07 | 北京大学 | Fall-down detection method and system |
US20170354349A1 (en) * | 2016-06-10 | 2017-12-14 | The Regents Of The University Of California | Wifi-based person-identification technique for use in smart spaces |
CN106772219A (en) * | 2017-03-08 | 2017-05-31 | 南京大学 | Indoor orientation method based on CSI signals |
CN107331136A (en) * | 2017-05-11 | 2017-11-07 | 深圳市斑点猫信息技术有限公司 | Indoor human body activity detection approach and system based on WiFi |
CN107749143A (en) * | 2017-10-30 | 2018-03-02 | 安徽工业大学 | A kind of indoor occupant fall detection system and method through walls based on WiFi signal |
CN108614989A (en) * | 2017-12-20 | 2018-10-02 | 深圳大学 | A kind of old man's intelligent monitor method, system and medium based on WiFi |
CN108761542A (en) * | 2018-05-03 | 2018-11-06 | 西北大学 | A kind of passive type object detection method based on WiFi signal |
CN108833036A (en) * | 2018-05-31 | 2018-11-16 | 湖南大学 | Tumble detection method for human body |
CN109394229A (en) * | 2018-11-22 | 2019-03-01 | 九牧厨卫股份有限公司 | A kind of fall detection method, apparatus and system |
EP3692898A1 (en) * | 2019-02-11 | 2020-08-12 | Nokia Technologies Oy | Sleep/motion determination based on wi-fi signals |
CN110288018A (en) * | 2019-06-24 | 2019-09-27 | 桂林电子科技大学 | A kind of WiFi personal identification method merging deep learning model |
CN111597877A (en) * | 2020-04-02 | 2020-08-28 | 浙江工业大学 | Fall detection method based on wireless signals |
CN111556453A (en) * | 2020-04-27 | 2020-08-18 | 南京邮电大学 | Multi-scene indoor action recognition method based on channel state information and BilSTM |
CN112036433A (en) * | 2020-07-10 | 2020-12-04 | 天津城建大学 | CNN-based Wi-Move behavior sensing method |
CN111954250A (en) * | 2020-08-12 | 2020-11-17 | 郑州大学 | Lightweight Wi-Fi behavior sensing method and system |
CN112346050A (en) * | 2020-10-23 | 2021-02-09 | 清华大学 | Fall detection method and system based on Wi-Fi equipment |
CN112330924A (en) * | 2020-11-05 | 2021-02-05 | 北京理工大学 | Falling event detection method based on channel state information in indoor environment |
CN112545496A (en) * | 2020-11-09 | 2021-03-26 | 金茂智慧科技(广州)有限公司 | Wireless technology identification method for detecting human body tumble |
CN112869734A (en) * | 2021-01-11 | 2021-06-01 | 乐鑫信息科技(上海)股份有限公司 | Wi-Fi human body detection method and intelligent device |
Non-Patent Citations (1)
Title |
---|
段鹏松等: "WiNet:一种适用于无线感知场景的步态识别模型", 《西安交通大学学报》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111445905B (en) | Mixed voice recognition network training method, mixed voice recognition method, device and storage medium | |
CN111091116B (en) | Signal processing method and system for judging arrhythmia | |
CN105943013B (en) | Heart rate detection method and device and intelligent wearable device | |
CN111954250B (en) | Lightweight Wi-Fi behavior sensing method and system | |
CN111242005B (en) | Heart sound classification method based on improved wolf's swarm optimization support vector machine | |
CN109620210B (en) | A kind of electrocardiosignal classification method of the CNN based on from coding mode in conjunction with GRU | |
CN111368627B (en) | Method and system for classifying heart sounds by combining CNN (computer numerical network) with improved frequency wavelet slice transformation | |
CN109871808B (en) | Atrial fibrillation model training and detecting method and device | |
CN111797674B (en) | MI electroencephalogram signal identification method based on feature fusion and particle swarm optimization algorithm | |
CN109567789B (en) | Electrocardiogram data segmentation processing method and device and readable storage medium | |
CN107239684A (en) | A kind of feature learning method and system for ECG identifications | |
CN107530015B (en) | Vital sign analysis method and system | |
CN105147252A (en) | Heart disease recognition and assessment method | |
Ayinala et al. | Low complexity algorithm for seizure prediction using Adaboost | |
CN111588367A (en) | Heart rate detection method and device and computer readable storage medium | |
CN114492501A (en) | Electroencephalogram signal sample expansion method, medium and system based on improved SMOTE algorithm | |
CN111387975A (en) | Electroencephalogram signal identification method based on machine learning | |
CN109189207A (en) | A kind of detection method and device of wearable device wearing state | |
CN113349793A (en) | Arrhythmia classification method based on layered attention model | |
CN109498001B (en) | Sleep quality evaluation method and device | |
CN113920109A (en) | Medical image recognition model training method, recognition method, device and equipment | |
CN114469120A (en) | Multi-scale Dtw-BiLstm-Gan electrocardiosignal generation method based on similarity threshold migration | |
Jain et al. | Rider manta ray foraging optimization-based generative adversarial network and CNN feature for detecting glaucoma | |
CN113128384B (en) | Brain-computer interface software key technical method of cerebral apoplexy rehabilitation system based on deep learning | |
CN111311466B (en) | Safety control method and device |
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: 20211015 |