Wireless network-based human posture detection method, apparatus, device, and storage medium

Abstract

To provide a wireless network-based posture detection method, device, electronic apparatus, and storage medium.SOLUTION: A method includes the steps of: obtaining channel state information of a wireless network, collected by a signal collection device; denoising the channel state information and extracting features from the denoised channel state information to obtain target feature information corresponding to the channel state information; and inputting the target feature information into a pre-trained posture detection model and outputting a target posture tag corresponding to the channel state information. The use of wireless networks to reflect human body data is achieved without the requirement for deploying devices for the entire scene, which reduces detection cost. Moreover, by combining deep learning and the sensing of wireless network signals, it is possible to improve the convenience and accuracy of human posture identification and detection.SELECTED DRAWING: Figure 1

Description

【Technical Field】 【0001】 This application relates to the technical field of human body posture detection, and particularly to a posture detection method, apparatus, electronic device, and storage medium based on a wireless network. 【Background Art】 【0002】 As smart home automation technology develops rapidly and is widely applied to people's lives and work, for example, when applied to the scene of home care for the elderly, in order to improve the timeliness of care, the importance of analyzing the position and behavior state of the human body also increases accordingly. 【0003】 Conventionally, when detecting and analyzing the position and behavior state of the human body, infrared rays are often used as sensor signals. However, due to the insufficient permeability of infrared signals, it is necessary to install sensors in each space. Also, if the human body does not move for a certain period of time, it may not be possible to detect it, and it can only be detected by waving at the sensor after a certain period of time. Therefore, the human body posture cannot be accurately and timely detected. Especially in the scene of home care for the elderly, when dangerous behaviors such as falling occur, the posture cannot be timely detected to arouse attention, resulting in safety problems. 【0004】 Therefore, currently, there is an urgent need for a human body posture detection method that improves the convenience and accuracy of detection. 【Summary of the Invention】 【Problems to be Solved by the Invention】 【0005】 The embodiments of this application aim to provide a posture detection method, apparatus, electronic device, and storage medium based on a wireless network to solve the technical problem that the detection of human body posture behavior in the related art is not convenient and accurate. 【Means for Solving the Problems】 【0006】 In the first embodiment, the attitude detection method based on a wireless network according to the embodiment of the present application is: The steps include: acquiring channel status information of a wireless network collected by a signal acquisition device; The steps include: performing noise reduction processing on the channel state information, performing feature extraction on the noise-reduced channel state information to obtain target feature information corresponding to the channel state information; The process includes the steps of inputting the target feature information into a pre-trained posture detection model and outputting and acquiring a target posture tag corresponding to the channel state information. 【0007】 In the second embodiment, the attitude detection device based on a wireless network according to the embodiment of the present application is: A data acquisition module that acquires channel status information of a wireless network collected by a signal acquisition device, A data processing module that performs noise reduction processing on the channel state information, performs feature extraction on the noise-reduced channel state information, and obtains target feature information corresponding to the channel state information. The system includes a posture determination module that inputs the aforementioned target feature information into a pre-trained posture detection model and outputs and acquires a target posture tag corresponding to the channel state information. 【0008】 In a third embodiment, the electronic device according to the embodiment of the present application includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor, upon executing the computer program, realizes a step of the attitude detection method based on a wireless network as described in any one of the above paragraphs. 【0009】 In the fourth embodiment, the computer-readable storage medium according to the embodiment of the present application stores a computer program that, when executed by a processor, realizes the steps of the wireless network-based attitude detection method described in any one of the above paragraphs. [Effects of the Invention] 【0010】 The embodiment of this application provides a posture detection method, apparatus, electronic device, and storage medium based on a wireless network. When detecting human posture in an environment, it acquires channel state information (CSI data) of the wireless network collected by a signal acquisition device, then performs noise reduction processing on the channel state information, performs feature extraction after noise reduction processing to acquire target feature information corresponding to the channel state information, and finally predicts and acquires a posture tag corresponding to the channel state information using a pre-trained posture detection model. In the process of detecting and identifying posture, it realizes the realization of human body data using a wireless network, and because of the characteristics of the wireless network itself, it is not necessary to place equipment in all scenes, thereby reducing detection costs. Furthermore, by combining deep learning and detection of wireless network signals, it improves the convenience and accuracy of identifying and detecting human posture. [Brief explanation of the drawing] 【0011】 [Figure 1] This is a schematic diagram showing the steps of a wireless network-based attitude detection method according to an embodiment of the present invention. [Figure 2] This is a schematic diagram showing the steps for acquiring target feature information according to an embodiment of the present invention. [Figure 3] This is a schematic diagram showing the steps of the training process for a posture detection model according to an embodiment of the present invention. [Figure 4] This is a schematic diagram showing the steps for processing training data according to an embodiment of the present invention. [Figure 5] This is a schematic diagram of the signal pattern of a specific set of historical channel state information according to an embodiment of the present application. [Figure 6] This is a schematic diagram of the signal pattern after clipping processing has been performed on a specific set of historical channel state information according to an embodiment of the present application. [Figure 7] This is a schematic diagram of a spectrogram of a pre-processed specific set of historical channel state information according to an embodiment of the present application. [Figure 8] It is a schematic diagram showing the step of obtaining target feature information according to an embodiment of the present application. [Figure 9] It is a schematic configuration diagram of a posture detection device based on a wireless network according to an embodiment of the present application. [Figure 10] It is a schematic configuration diagram of an electronic device according to an embodiment of the present application. [Figure 11] It is another schematic configuration diagram of an electronic device according to an embodiment of the present application. 【Mode for Carrying Out the Invention】 【0012】 Hereinafter, referring to the drawings in the embodiments of the present application, the technical means in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present application, not all of them. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor belong to the protection scope of the present application. 【0013】 In addition, each step described in the embodiment of the method of the present disclosure may be executed in a different order and / or may be executed in parallel. Also, the embodiment of the method may include additional steps and / or may omit the execution of the steps shown. The scope of the present disclosure is not limited in this regard. 【0014】 The term "including" and its variants used in this specification are open inclusion, that is, "including but not limited to these". The term "based on" means "at least partially based on". The term "one embodiment" represents "at least one embodiment", the term "another embodiment" represents "at least one another embodiment", and the term "other embodiments" represents "at least some embodiments". Related definitions of other terms are given in the following description. 【0015】 In related technologies, when detecting and analyzing the position and behavior state of the human body, infrared rays are often used as sensor signals. However, since the permeability of infrared signals is not sufficient, it is necessary to install sensors in each space. Also, if the human body does not move for a certain period of time, it may not be possible to detect it, and it can only be detected by waving at the sensor after a certain period of time. Therefore, when detecting the human body posture, not only is it necessary to arrange devices over a wide range, but after the arrangement, detection and analysis cannot be carried out in a timely and effective manner, and the detection and analysis of the human body posture cannot be carried out conveniently, timely and accurately. Also, the cost is high. 【0016】 In order to solve the technical problems existing in related technologies, the embodiments of the present application provide a posture detection method based on a wireless network. FIG. 1 is a schematic diagram showing the steps of a posture detection method based on a wireless network according to an embodiment of the present application. As shown in FIG. 1, the method includes steps 101 to 103. 【0017】 In step 101, the channel state information of the wireless network collected by the signal collection device is obtained. 【0018】 In one embodiment, when performing analysis and processing of the human body posture, the channel state information of the wireless network collected by the signal collection device is obtained, and further, by analyzing and processing the channel state information, the posture corresponding to the currently collected channel state information is determined. 【0019】 In actual applications, the entire environment is preset, and corresponding signal transmitters and signal receivers are provided in the environment. Then, by continuously pinging the signal transmitter using the corresponding signal source, the signal transmitter returns the signal to the signal receiver, and by performing analysis and processing on the signal received by the signal receiver, the posture of the human body in the current environment is determined. 【0020】 For example, if the signal transmitter is a router and the signal receiver is the Bcm43455 chip mounted on a Raspberry Pi 4B, in the implementation process, a Raspberry Pi Pico W is used as a ping signal source. By continuously pinging the router with the Raspberry Pi Pico, the router returns a Wi-Fi signal at a constant refresh rate. Then, the Nexmon CSI tool captures the Wi-Fi signal returned from the router, which specifically contains CSI signals due to human movement, to use as relevant data for human posture detection. 【0021】 Furthermore, human posture includes, but is not limited to, sitting still, standing, sitting, walking, and falling postures. Also, to detect human posture in the environment, if there are no people in the environment, the posture is considered "empty." 【0022】 In step 102, denoising is performed on the channel state information, and feature extraction is performed on the denoised channel state information to obtain target feature information corresponding to the channel state information. 【0023】 In one embodiment, channel state information in the environment is collected, and then related analysis is performed. Specifically, noise reduction is performed on the channel state information, followed by feature extraction after noise reduction to obtain target feature information corresponding to the channel state information, and this target feature information is used as the basis for subsequent attitude detection and judgment. 【0024】 Specifically, the collected channel state information is generated based on the application of wireless detection technology based on wireless networks. Due to the influence of the environment and other factors, the collected channel state information may contain unwanted information such as noise. Therefore, before further analysis, it is necessary to denoise the channel state information to remove the unwanted information, and after the denoising is complete, feature extraction is performed to obtain target feature information for subsequent analysis. 【0025】 Figure 2 is a schematic diagram showing the steps for acquiring target feature information according to an embodiment of the present application. Furthermore, as shown in Figure 2, these steps include steps 201 to 203. 【0026】 In step 201, clipping is performed on the channel state information to obtain the clipped channel state information. In step 202, it is determined whether or not noise is present in the clipped channel state information, and if it is determined that noise is present, a noise reduction process is performed to obtain the noise-reduced channel state information. In step 203, feature extraction is performed on the denoised channel state information to obtain target feature information corresponding to the channel state information. 【0027】 For example, when collecting channel status information from a wireless network, the channel status information itself contains a certain amount of noise data. Therefore, after acquiring the channel status information, clipping is performed on the channel status information to remove the normal noise contained in it, and noise-free channel status information is obtained. Then, it is determined whether or not other noise is contained in the noise-free channel status information. If noise is contained, the noise removal process is performed again to obtain noise-free channel status information, and finally, feature information is extracted from the noise-free channel status information to obtain target feature information corresponding to the channel status information. 【0028】 Typically, when collecting channel status information (i.e., CSI data) from a wireless network, the collected channel status information is data containing 64 subcarriers, with a numerical range of 0 to 35000, and a small number of these subcarriers are normal noise. For example, subcarriers 1-2, 28-36, and 64 are noise subcarriers. In this case, when clipping is performed, the noise subcarriers are clipped, and 52 subcarriers with complete operation suspended are obtained. 【0029】 After the clipping process for the channel state information is completed, the normal noise contained within it is removed. However, since the channel state information may contain ultra-low frequency noise, a judgment process is performed on the clipped channel state information to determine whether or not other noise is present, thereby obtaining completely noise-free channel state data. 【0030】 If further processing is required for clipped channel state information, a Fast Fourier Transform is performed on the channel state information, and spectral analysis is used to identify the ultra-low frequency noise component of the clipped channel state information. The standard deviation of the ultra-low frequency noise component data is compared with a set threshold to determine whether or not it is noise. If it is determined that ultra-low frequency noise is present, the sequence containing the ultra-low frequency noise is removed, and data is added using the data from the preceding sequence among its adjacent sequences to ensure data integrity. 【0031】 In step 103, target feature information is input into a pre-trained posture detection model, and a target posture tag corresponding to the channel state information is output and obtained. 【0032】 In one embodiment, after completing the processing of channel state information and acquiring target feature information, the target feature information is input into a pre-trained posture detection model, and posture tags corresponding to the channel state information are output and acquired. Based on the posture settings described above, posture tags include, but are not limited to, tags such as sitting still, standing, sitting, walking, falling, and unmanned environment. Furthermore, tags can be modified and added or removed according to the actual situation and actual needs. 【0033】 For example, the posture detection model currently in use is pre-trained based on relevant historical data, which is historical channel status information of a wireless network. The posture detection model is trained and adjusted by acquiring the corresponding historical data in advance and performing processing and feature extraction on the historical data. Furthermore, once training is complete, the trained posture detection model for posture determination is obtained. 【0034】 Figure 3 is a schematic diagram showing the steps of the training process for a posture detection model according to an embodiment of the present invention. As shown in Figure 3, the process includes steps 301 to 303. 【0035】 Step 301 includes the history channel status information of the wireless network and posture tags, and acquires training data corresponding to one posture tag for each history channel status information. In step 302, preprocessing and feature extraction are performed on the historical channel state information to obtain target feature information corresponding to the historical channel state information. In step 303, the posture detection model to be trained is trained based on posture tags and target feature information, and once training is complete, the trained posture detection model is obtained. 【0036】 In one embodiment, when training to acquire a posture detection model, the associated training data is acquired, which includes wireless network history channel state information and posture tags, with each history channel state information corresponding to one posture tag. Subsequently, preprocessing and feature extraction are performed on the history channel state information to acquire target feature information corresponding to the history channel state information. The data from which feature extraction is performed is the preprocessed history channel state information. Finally, the posture detection model to be trained is trained based on the posture tags and target feature information. Furthermore, upon completion of training, the trained posture detection model is acquired. The condition for completion of training may be that the number of training iterations reaches a set threshold, or it may be that the set conditions are met by adjusting the parameters. 【0037】 Based on the aforementioned environmental conditions, the acquired historical channel status information consists of Wi-Fi signals returned from the router, captured within a certain time period, including changes in the CSI signal due to human movement. Each set of motion data generates a single binary file (bin file). For stationary postures, each posture tag contains at least 250 data points, and each data point is collected for 20 seconds. Each set of data contains 64 subcarriers. 【0038】 Furthermore, similar to performing posture detection analysis using a trained posture detection model, it is necessary to perform relevant preprocessing on the historical channel state information after acquiring it, specifically the same as the processing process described in step 102 and related embodiments above. Specifically, after acquiring training data, clipping, denoising, and feature extraction processing are performed on each set of data in the training data to obtain target feature information corresponding to each set of data. 【0039】 Specifically, when processing training data to obtain target feature information, you can refer to Figure 4. Figure 4 is a schematic diagram showing the steps for processing training data according to an embodiment of the present invention, and these steps include steps 401 to 403. 【0040】 In step 401, clipping is performed on the historical channel state information to obtain the clipped historical channel state information. In step 402, denoising is performed on the clipped historical channel state information to obtain the denoised historical channel state information. In step 403, feature extraction is performed on the denoised historical channel state information to obtain target feature information corresponding to the historical channel state information. 【0041】 For example, when processing historical channel state information, first, clipping is performed on the historical channel state information to obtain the clipped historical channel state information. Then, denoising is performed on the clipped historical channel state information to obtain the denoised historical channel state information. Finally, feature information is extracted to obtain target feature information corresponding to each set of data in the historical channel state information, and each target feature information is associated with the pose tag of the corresponding data. 【0042】 Refer to Figure 5, which uses a specific set of historical channel state information as an example. Figure 5 is a schematic diagram of the signal pattern of a specific set of historical channel state information according to an embodiment of the present application. When clipping is performed, normal noise included in the historical channel state information is removed, and the normal noise is the protruding portion in Figure 5. Specifically, analysis can determine that the normal noise corresponds to the data of the 1st to 2nd, 28th to 36th, and 64th subcarriers, and when clipping is performed, the normal noise (1st to 2nd, 28th to 36th, and 64th subcarriers) is removed to obtain historical channel state information of 52 subcarriers including complete operation data. Figure 6 is a schematic diagram of the signal pattern of a clipped specific set of historical channel state information according to an embodiment of the present application. As shown in Figure 6, the data included in this case is signal data including complete operation. 【0043】 Therefore, when performing clipping processing on historical channel state information, the process includes the steps of: performing a first noise analysis on the historical channel state information to obtain a first noise contained in the historical channel state information; and clipping the first noise in the historical channel state information to obtain historical channel state information that does not contain the first noise. 【0044】 In other words, the noise contained in the history channel state information is analyzed and identified, the primary noise contained within it is determined, and then clipping is performed on the primary noise to obtain history channel state information that does not contain the primary noise. Specifically, normal noise (primary noise, 1st to 2nd subcarriers, 28th to 36th subcarriers, and 64th subcarrier) is clipped in the history channel data of 64 subcarriers to obtain history channel state information for 52 subcarriers. 【0045】 Furthermore, after the removal of normal noise in the history channel state information is complete, filtering may be performed to suppress unsmooth high-frequency noise in the signal and low-frequency noise that shifts the entire signal upward, thereby ensuring the smoothing level of each set of history channel state information. Specifically, a Butterworth bandpass filter may be used to filter the clipped history channel state information, with the order set to 1st order and the cutoff frequency set to 0.05Hz to 0.25Hz. Further filtering will smooth the signal data while retaining signal data due to human movement. Figure 6 can be used to show the signal pattern of a specific set of filtered history channel state information. 【0046】 Then, when processing, if the ultra-low frequency noise component of the historical channel state information is to be removed, the presence or absence of ultra-low frequency noise is determined using the Fast Fourier Transform and spectral analysis methods, and if it is determined to be present, noise removal processing is performed. As an example, in Figure 6, the ultra-low frequency noise is the smoother line in Figure 6, but of course, not all channel state information contains ultra-low frequency noise, so when processing, it is first determined whether or not ultra-low frequency noise is present, and then, if it is determined to be present, removal processing is performed. Specifically, this includes the steps of performing a second noise analysis on the clipped historical channel state information, removing the second noise obtained from the analysis if it is determined that noise is present, and performing data padding on the historical channel state information from which the second noise has been removed to obtain noise-removed historical channel state information, and if it is determined that there is no noise after the second noise analysis, converting the clipped historical channel state information into noise-removed historical channel state information. 【0047】 During processing, a Fast Fourier Transform calculation is performed on the clipped and filtered historical channel state information to obtain a spectrogram corresponding to the historical channel state information. Figure 7 is a schematic diagram of the spectrogram of a specific set of historical channel state information according to an embodiment of the present invention. As can be seen from Figure 7, the noise frequency decreases to almost 0 in sequence 2, and the normal signal has a higher value in sequence 1 than in sequences 0 and 2. At this time, by calculating the standard deviation of the sequence and checking based on a set threshold, it is possible to determine whether or not it is streaky noise, i.e., very low frequency noise. 【0048】 The specific calculation and decision-making methods are as follows: 1. If fft_std010 > 0.29, check whether the standard deviation of the first 10 frequency sequences is greater than 0.29. 2. If fft_std010 < 0.275, check whether the standard deviation of the first 10 frequency sequences is less than 0.275. 3. If fft_std02 < 0.35, check whether the standard deviation of the first two frequency sequences is less than 0.35. 【0049】 After processing based on the above method, the history channel state information shown in Figure 6 becomes a signal that does not contain ultra-low frequency noise, specifically as shown in Figure 7. Figure 7 is a schematic diagram of the signal pattern when preprocessing of a specific set of history channel state information according to the embodiment of the present application is completed. 【0050】 Furthermore, after noise reduction processing of the historical channel state information is completed, target feature information is extracted. Figure 8 is a schematic diagram showing the steps for obtaining target feature information according to an embodiment of the present invention. Specifically, as shown in Figure 8, the steps include steps 801 to 804. 【0051】 In step 801, based on the posture tags, classification processing is performed on the denoised historical channel state information to obtain tag data corresponding to each posture tag, and the maximum amount of tag data is determined. In step 802, sample synthesis processing is performed based on the tag data to obtain synthesized data corresponding to each tag data. In step 803, additional data for each tag is determined from the composite data based on the maximum data volume, and category data corresponding to each posture tag is obtained based on the additional data and tag data, with each category data corresponding to one posture tag. In step 804, feature extraction is performed on the categorical data to obtain target feature information corresponding to the historical channel state information. 【0052】 In one embodiment, when extracting target feature information, a data synthesis method is used to ensure consistency in the amount of training data for different poses in order to avoid biased fitting of the model to one or more poses during the training process. Specifically, after completing processing on the history channel state information and obtaining denoised history channel state information, classification processing is performed based on pose tags to determine tag data corresponding to each pose tag, the amount of data contained in each tag data is identified, and the maximum amount of data is obtained. Subsequently, sample data synthesis is performed on each tag data to obtain synthesized data containing a certain amount of data corresponding to each tag data, and then additional data for each tag data is determined from the synthesized data based on the maximum amount of data, and further category data corresponding to each pose tag in the history channel state information is obtained. Finally, target feature information extraction is performed on the category data to obtain target feature information corresponding to the history channel state information. 【0053】 For example, before extracting target feature information, sample data synthesis ensures that the amount of training data corresponding to each posture tag matches, and further avoids overfitting during the training process. 【0054】 When synthesizing sample data, the pose with the largest data volume is determined, and this maximum data volume is used as the criterion for adding data corresponding to each pose tag. For example, if the maximum data volume is 200, then after sample synthesis and data addition are complete, the data volume corresponding to each pose tag will be 200. During the sample data synthesis process, the SMOTE algorithm can be used to generate synthesized samples. 【0055】 Specifically, for each selected small number of category samples (training data corresponding to other posture tags other than the one with the largest data volume), the k closest samples are found, one sample is randomly selected from the k samples, and a new synthetic sample is generated using the following formula. 【number】 Here, x new This is a synthetic sample, and x chosen This is a selected small number of category samples, and x nearest δ is a randomly selected sample from the k nearest samples, where δ∈[0,1]. Oversampling ensures that a small number of samples are equivalent to the others, and the best way to ensure data consistency is to refer to the largest data amount as the data amount. 【0056】 Furthermore, after the target feature information extraction is complete, the constructed posture detection model is trained. The posture detection model can be built using a deep learning model of a long-term memory network (LSTM), extracts features from the processed CSI data, and identifies human posture through feature extraction by the model and detection and classification of the actual data. 【0057】 In the training process, forward propagation is performed first, allowing the model to process each input data, transmit information through each layer of the network, and finally generate a corresponding output. Then, the cross-entropy loss is calculated to quantify the difference between the model's prediction and the actual tag. 【0058】 The formula for cross-entropy loss is as follows: 【number】 p is the actual category, and q is the distribution after model prediction. 【0059】 Then, backpropagation is performed, and in the backpropagation process, the loss obtained by gradient descent is backpropagated throughout the entire network to adjust the model weights and deviations. Then, the Adam optimizer is used to automatically adjust the model weights and deviations and minimize the loss function. During training, the performance of the machine learning model is evaluated using 10-fold cross-validation, and the performance metrics obtained over 10 iterations are averaged to obtain the final performance evaluation of the model, and a decision is made whether or not to acquire the trained pose detection model after training is complete. 【0060】 Based on the above, this application discloses a posture detection method based on a wireless network. When detecting human posture in an environment, channel state information (CSI data) of the wireless network is acquired by a signal acquisition device, noise reduction processing is then performed on the channel state information, feature extraction is performed after noise reduction processing to acquire target feature information corresponding to the channel state information, and finally, a posture tag corresponding to the channel state information is output and acquired by predicting with a pre-trained posture detection model. In the process of detecting and identifying posture, human body data is realized using a wireless network, and because it is not necessary to place equipment in all scenes due to the characteristics of the wireless network itself, detection costs are reduced, and the convenience and accuracy of identifying and detecting human posture are improved by combining deep learning and detection of wireless network signals. 【0061】 Based on the methods described in the above embodiment, this embodiment will be further described in terms of a wireless network-based attitude detection device, which may be implemented as an independent entity or integrated into an electronic device, such as a terminal, and the terminal may include a mobile phone, a tablet computer, and the like. 【0062】 Figure 9 is a schematic diagram of the wireless network-based attitude detection device according to an embodiment of the present application. As shown in Figure 9, the wireless network-based attitude detection device 900 according to an embodiment of the present application is A data acquisition module 901 that acquires channel status information of a wireless network collected by a signal acquisition device, A data processing module 902 performs noise reduction on channel state information, extracts features from the noise-reduced channel state information, and obtains target feature information corresponding to the channel state information. The system includes a posture determination module 903 that inputs target feature information into a pre-trained posture detection model and outputs and acquires a target posture tag corresponding to channel state information. 【0063】 In one embodiment, the data processing module 902 further, The channel state information is clipped, and the clipped channel state information is obtained. Determine whether or not noise is present in the clipped channel state information, and if it is determined that noise is present, perform noise reduction processing to obtain the noise-reduced channel state information. Feature extraction is performed on the denoised channel state information to obtain target feature information corresponding to the channel state information. 【0064】 In one embodiment, the posture detection device 900 based on a wireless network further includes a model training module, which is The system includes wireless network history channel status information and posture tags, and acquires training data corresponding to one posture tag for each history channel status information. Preprocessing and feature extraction are performed on the historical channel state information to obtain target feature information corresponding to the historical channel state information. Based on posture tags and target feature information, the posture detection model to be trained is trained, and once training is complete, the trained posture detection model is retrieved. 【0065】 In one embodiment, the model training module further includes: Clipping is performed on the historical channel state information, and the clipped historical channel state information is obtained. The clipped historical channel state information is subjected to noise reduction processing to obtain the noise-reduced historical channel state information. Feature extraction is performed on the denoised historical channel state information to obtain target feature information corresponding to the historical channel state information. 【0066】 In one embodiment, the model training module further includes: The first noise analysis is performed on the historical channel state information to obtain the first noise contained in the historical channel state information. The first noise is clipped from the historical channel state information to obtain historical channel state information that does not contain the first noise. 【0067】 In one embodiment, the model training module further includes: A second noise analysis is performed on the clipped historical channel state information. If a second noise analysis is performed and it is determined that noise is present, the second noise obtained from the analysis is removed, and data padding is performed on the historical channel state information from which the second noise has been removed to obtain the noise-removed historical channel state information. If a second noise analysis is performed and it is determined that no noise exists, the clipped historical channel state information is treated as the denoised historical channel state information. 【0068】 In one embodiment, the model training module further includes: Based on the posture tags, classification processing is performed on the denoised historical channel state information to obtain tag data corresponding to each posture tag, and the maximum amount of tag data is determined. Based on the tag data, a sample synthesis process is performed to obtain the synthesized data corresponding to each tag data. Based on the maximum data volume, additional data for each tag is determined from the composite data, and based on the additional data and tag data, category data corresponding to each posture tag is obtained, with each category data corresponding to one posture tag. Feature extraction is performed on categorical data to obtain target feature information corresponding to the historical channel state information. 【0069】 Figure 10 is a schematic diagram of an electronic device according to an embodiment of the present application. As shown in Figure 10, the electronic device may be a mobile terminal, such as a smartphone or tablet computer. As shown in Figure 10, the electronic device 1000 includes a processor 1001 and a memory 1002. The processor 1001 is electrically connected to the memory 1002. 【0070】 The processor 1001 is the control center of the electronic device 1000, connected to all parts of the entire electronic device by various interfaces and lines, and performs various functions and data processing of the electronic device 1000 by executing or loading application programs stored in memory 1002 and retrieving data stored in memory 1002, thereby monitoring the electronic device 1000 as a whole. 【0071】 In this embodiment, the processor 1001 of the electronic device 1000 loads commands corresponding to the processes of one or more application programs into the memory 1002 according to the following steps, and the processor 1001 executes the application programs stored in the memory 1002, thereby realizing any of the steps of the attitude detection method based on the wireless network according to the above embodiment. 【0072】 Since the electronic device 1000 can implement the steps in any embodiment of the attitude detection method based on a wireless network according to the embodiments of this application, it can achieve the beneficial effects realized by any embodiment of the attitude detection method based on a wireless network according to the embodiments of this application. For details, please refer to the embodiments described above, and the explanation is omitted here. 【0073】 Figure 11 is another schematic configuration diagram of the electronic device according to an embodiment of the present application. Figure 11 shows a specific structural block diagram of the electronic device according to an embodiment of the present application, and as shown in Figure 11, the electronic device can implement the attitude detection method based on a wireless network according to the above embodiment. The electronic device 1100 may be a mobile terminal, such as a smartphone or a laptop computer. 【0074】 The RF circuit 1110 communicates with a communication network or other devices by receiving and transmitting electromagnetic waves and performing the mutual conversion of electromagnetic waves and electrical signals. The RF circuit 1110 may include various existing circuit elements to perform these functions, such as an antenna, a radio frequency transceiver, a digital signal processor, an encryption / decryption chip, a subscriber identification module (SIM) card, and memory. The RF circuit 1110 may communicate with various networks such as the Internet, a company intranet, and a wireless network, and may communicate with other devices via a wireless network. The wireless network may include a cellular telephone network, a wireless local area network, or a metropolitan area network. The above wireless network may use a variety of communication standards, protocols, and technologies, including Global System for Mobile Communication (GSM), Enhanced Data GSM Environment (EDGE), Wideband Code Division Multiple Access (WCDMA®), Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Wireless Fidelity (Wi-Fi) (e.g., IEEE 802.11a, IEEE 802.11b, IEEE 802.11g and / or IEEE 802.11n standards), Voice over Internet Protocol (VoIP), Worldwide Interoperability for Microwave Access (Wi-Max), other protocols for mail, instant messaging and short messaging, and any other suitable communication protocols, and may also include protocols not currently under development. 【0075】 Memory 1120 can store software programs and modules, for example, program commands / modules corresponding to the attitude detection method based on a wireless network in the above embodiment. The processor 1180 executes various functional applications and the attitude detection method based on a wireless network by executing the software programs and modules stored in memory 1120. 【0076】 Memory 1120 may include high-speed random-access memory and may further include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, memory 1120 may further include memory located remotely from the processor 1180, and these remote memories can be connected to electronic devices 1100 via a network. Examples of the network include, but are not limited to, the Internet, a company intranet, a local area network, a mobile communication network, and combinations thereof. 【0077】 The input unit 1130 may receive input numerical or character information and generate signal inputs from a keyboard, mouse, lever, optical or trackball related to user settings and function control. Specifically, the input unit 1130 may include a touch-sensitive surface 1131 and other input devices 1132. The touch-sensitive surface 1131, also called a touch display or touch panel, can collect user touch operations on or near it (for example, operations on or near the touch-sensitive surface 1131 performed by the user using any suitable object or accessory such as a finger or stylus) and drive corresponding connected devices according to a predetermined program. The touch-sensitive surface 1131 may optionally include two parts: a touch detection device and a touch controller. The touch detection device detects the direction of the user's touch, detects the signal from the touch operation, and transmits the signal to the touch controller. The touch controller receives touch information from the touch detection device, converts it into touch point coordinates, and transmits it to the processor 1180. The touch controller receives and executes instructions transmitted from the processor 1180. Furthermore, the touch-sensitive surface 1131 can be realized using multiple types, such as resistive, capacitive, infrared, and surface acoustic wave types. In addition to the touch-sensitive surface 1131, the input unit 1130 may further include other input devices 1132. Specifically, the other input devices 1132 may include, but are not limited to, one or more of the following: a physical keyboard, function keys (e.g., volume control keys, switch keys, etc.), a trackball, a mouse, and an operating lever. 【0078】 The display unit 1140 may display information entered by the user or provided to the user, and various graphical user interfaces of the electronic device 1100, which may consist of graphics, text, icons, videos, and any combination thereof. The display unit 1140 may include a display panel 1141, which may optionally be arranged using a form such as an LCD (Liquid Crystal Display) or OLED (Organic Light-Emitting Diode). Furthermore, a touch-sensitive surface 1131 may cover the display panel 1141, and after the touch-sensitive surface 1131 detects a touch operation on or near it, it transmits it to the processor 1180 to determine the type of touch event, which then provides the display panel 1141 with a corresponding visual output according to the type of touch event. In the figure, the touch-sensitive surface 1131 and the display panel 1141 realize input and output functions as two independent components, but in some embodiments, the touch-sensitive surface 1131 and the display panel 1141 may be integrated to realize input and output functions. 【0079】 The electronic device 1100 may further include at least one type of sensor 1150, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor may include an ambient light sensor and a proximity sensor, the ambient light sensor can adjust the brightness of the display panel 1141 based on the brightness of the ambient light, and the proximity sensor can be interrupted when the cover is folded or closed. As a type of motion sensor, a gravity acceleration sensor can detect the magnitude of acceleration in each direction (generally three axes), can detect the magnitude and direction of gravity when stationary, and can be used in applications that identify the orientation of a mobile phone (e.g., switching between portrait and landscape screens, related games, magnetometer orientation calibration), functions related to vibration identification (e.g., pedometer, tapping), etc. The electronic device 1100 may further include other sensors such as a gyroscope, barometer, hygrometer, thermometer, and infrared sensor, which are not described here. 【0080】 The audio circuit 1160, speaker 1161, and microphone 1162 can provide an audio interface between the user and the electronic device 1100. The audio circuit 1160 transmits the electrical signal, after converting the received audio data, to the speaker 1161, which converts it back into an audio signal and outputs it. The microphone 1162 converts the collected audio signal into an electrical signal, receives it through the audio circuit 1160, converts it back into audio data, outputs the audio data to the processor 1180 for processing, and then transmits it to another terminal, for example, via the RF circuit 1110, or outputs the audio data to the memory 1120 for further processing. The audio circuit 1160 may further include an earphone jack to provide communication between an external earphone and the electronic device 1100. 【0081】 The electronic device 1100 can assist the user in receiving requests and transmitting information via a transmission module 1170 (e.g., a Wi-Fi module), and provide the user with wireless broadband internet access. Although the transmission module 1170 is shown in the figure, as can be understood, the transmission module 1170 is not an essential component of the electronic device 1100 and may be omitted as necessary, without altering the essence of the invention. 【0082】 The processor 1180 is the control center of the electronic device 1100, connected to each part of the entire mobile phone using various interfaces and lines, and performs various functions and data processing of the electronic device 1100 by operating or executing software programs and / or modules stored in memory 1120 and retrieving data stored in memory 1120, thereby monitoring the electronic device as a whole. The processor 1180 may optionally include one or more processing cores, and in other embodiments, the processor 1180 may integrate an application processor and a modem processor, the application processor mainly handling the operating system, user interface and application programs, etc., and the modem processor mainly handling wireless communications. As can be understood, the above modem processor does not have to be integrated into the processor 1180. 【0083】 The electronic device 1100 further includes a power supply 1190 (e.g., a battery) for supplying power to each component. In other embodiments, the power supply is logically connected to a processor 1180 by a power management system, which enables functions such as charge management, discharge management, and power consumption management. The power supply 1190 may further include one or more DC or AC power supplies, a recharge system, a power fault detection circuit, a power converter or inverter, a power status indicator, and other optional assembled components. 【0084】 Although not shown in the figures, the electronic device 1100 may further include a camera (e.g., a front camera, a rear camera), a Bluetooth® module, etc., which will not be described here. Specifically, in this embodiment, the display unit of the electronic device is a touchscreen display, and the mobile terminal further includes memory and one or more programs, the one or more programs being stored in memory and configured so that one or more programs are executed by one or more processors to realize any of the steps of the attitude detection method based on the wireless network according to the above embodiment. 【0085】 In practical implementation, each of the above modules may be implemented as an independent entity, or they may be combined as desired to be implemented as the same or multiple entities. For specific implementations of each of the above modules, refer to the previous method examples, and the explanation will be omitted here. 【0086】 Those skilled in the art will understand that all or some of the steps of the various methods in the above embodiments can be completed by commands or hardware related to command control, and such commands may be stored in a computer-readable storage medium and loaded and executed by a processor. For this purpose, embodiments of the present application provide a storage medium that stores a plurality of commands which, when executed by a processor, perform any of the steps of the wireless network-based attitude detection method according to the above embodiments. 【0087】 The storage medium may include read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks. 【0088】 Since the commands stored in the storage medium can execute the steps in any embodiment of the attitude detection method based on a wireless network according to the embodiments of the present invention, the beneficial effects realized by any embodiment of the attitude detection method based on a wireless network according to the embodiments of the present invention can be achieved. For details, please refer to the embodiments described above, and the explanation is omitted here. 【0089】 The attitude detection method, apparatus, electronic device, and storage medium based on a wireless network according to embodiments of the present application have been described in detail above. Although the principles and embodiments of the present application have been explained using specific examples in this specification, the above description of embodiments is merely to aid in understanding the method and spirit of the present application, and a person skilled in the art can modify the specific embodiments and scope of application in accordance with the concept of the present application. In short, the contents of this specification should not be understood as limiting the present application. Furthermore, a person skilled in the art can make several improvements and modifications without departing from the principles of the present application, and these improvements and modifications are also considered to be within the scope of protection of the present application.

Claims

[Claim 1] The steps include: acquiring channel status information of a wireless network collected by a signal acquisition device; The steps include: performing noise reduction processing on the channel state information, performing feature extraction on the noise-reduced channel state information to obtain target feature information corresponding to the channel state information; The steps include inputting the target feature information into a pre-trained posture detection model and outputting and acquiring a target posture tag corresponding to the channel state information, The training process for the aforementioned pre-trained posture detection model is as follows: The steps include acquiring training data that includes the history channel status information of the wireless network and posture tags, where each of the history channel status information corresponds to one posture tag. The steps include: performing preprocessing and feature extraction on the historical channel state information to obtain target feature information corresponding to the historical channel state information; The process includes the steps of: training a posture detection model to be trained based on the posture tags and target feature information, and obtaining the trained posture detection model once training is complete; The step of performing preprocessing and feature extraction on the aforementioned historical channel state information to obtain target feature information corresponding to the aforementioned historical channel state information is: The steps include performing a clipping process on the aforementioned historical channel status information and obtaining the clipped historical channel status information, The steps include: performing noise reduction processing on the clipped historical channel state information to obtain the noise-reduced historical channel state information; The step includes performing feature extraction on the denoised historical channel state information to obtain target feature information corresponding to the historical channel state information, The step of performing noise reduction processing on the clipped historical channel state information to obtain the noise-reduced historical channel state information is: The steps include performing a second noise analysis on the clipped historical channel state information, If a second noise analysis is performed and it is determined that noise is present, the second noise obtained through the analysis is removed, and data padding is performed on the historical channel state information from which the second noise has been removed to obtain noise-removed historical channel state information. If a second noise analysis is performed and it is determined that no noise exists, the clipped historical channel state information is replaced with denoised historical channel state information, and the process includes these steps. A method for detecting human posture based on a wireless network, characterized by the following features. [Claim 2] The steps of performing noise reduction processing on the channel state information, and then performing feature extraction on the noise-reduced channel state information to obtain target feature information corresponding to the channel state information are as follows: The steps include: performing a clipping process on the channel state information and obtaining the clipped channel state information; The steps include: determining whether or not noise is present in the clipped channel state information; if noise is present, performing a noise reduction process to obtain the noise-reduced channel state information; The process includes the step of performing feature extraction on denoised channel state information to obtain target feature information corresponding to the channel state information, The method according to feature 1. [Claim 3] The step of performing a clipping process on the aforementioned historical channel status information and obtaining the clipped historical channel status information is: The steps include: performing a first noise analysis on the historical channel state information to obtain the first noise contained in the historical channel state information; The process includes the step of clipping the first noise in the history channel state information to obtain history channel state information that does not contain the first noise. The method according to feature 1. [Claim 4] The step of extracting features from the denoised historical channel state information to obtain target feature information corresponding to the historical channel state information is: The steps include: performing a classification process on the denoised historical channel state information based on the posture tags, obtaining tag data corresponding to each posture tag, and determining the maximum amount of data for the tag data; The steps include: performing a sample synthesis process based on the aforementioned tag data to obtain synthesized data corresponding to each tag data; A step of determining additional data for each tag data from the composite data based on the maximum data amount, and obtaining category data corresponding to each posture tag based on the additional data and the tag data, wherein each category data corresponds to one posture tag. The process includes the step of performing feature extraction on the categorical data to obtain target feature information corresponding to the historical channel state information. The method according to feature 1. [Claim 5] A data acquisition module that acquires channel status information of a wireless network collected by a signal acquisition device, A data processing module that performs noise reduction processing on the channel state information, performs feature extraction on the noise-reduced channel state information, and obtains target feature information corresponding to the channel state information. The module includes a posture determination module that inputs the aforementioned target feature information into a pre-trained posture detection model and outputs and acquires a target posture tag corresponding to the channel state information, The training process for the aforementioned pre-trained posture detection model is as follows: The steps include acquiring training data that includes the history channel status information of the wireless network and posture tags, where each of the history channel status information corresponds to one posture tag. The steps include: performing preprocessing and feature extraction on the historical channel state information to obtain target feature information corresponding to the historical channel state information; The process includes the steps of: training a posture detection model to be trained based on the posture tags and target feature information, and obtaining the trained posture detection model once training is complete; The step of performing preprocessing and feature extraction on the aforementioned historical channel state information to obtain target feature information corresponding to the aforementioned historical channel state information is: The steps include performing a clipping process on the aforementioned historical channel status information and obtaining the clipped historical channel status information, The steps include: performing noise reduction processing on the clipped historical channel state information to obtain the noise-reduced historical channel state information; The step includes performing feature extraction on the denoised historical channel state information to obtain target feature information corresponding to the historical channel state information, The step of performing noise reduction processing on the clipped historical channel state information to obtain the noise-reduced historical channel state information is: The steps include performing a second noise analysis on the clipped historical channel state information, If a second noise analysis is performed and it is determined that noise is present, the second noise obtained through the analysis is removed, and data padding is performed on the historical channel state information from which the second noise has been removed to obtain noise-removed historical channel state information. If a second noise analysis is performed and it is determined that no noise exists, the clipped historical channel state information is replaced with denoised historical channel state information, and the process includes these steps. A wireless network-based human posture detection device characterized by the following features. [Claim 6] A processor, memory, and a computer program stored in the memory and executable on the processor, wherein the processor, upon executing the computer program, realizes a step of the method according to any one of claims 1 to 4. An electronic device characterized by the following features. [Claim 7] When executed by a processor, a computer program is stored which implements a step of the method according to any one of claims 1 to 4. A computer-readable storage medium characterized by the following features.

Images