Action gesture recognition method and device, storage medium and wearable device
By acquiring acceleration and angular velocity data, and combining standard deviation judgment and posture algorithm model, the problems of large data volume and environmental dependence in human posture detection in existing technologies are solved, and efficient action posture recognition is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- 深圳开鸿数字产业发展有限公司
- Filing Date
- 2022-10-18
- Publication Date
- 2026-06-02
AI Technical Summary
Existing human posture detection technologies mainly rely on image recognition, which suffers from problems such as large data collection volume, complex algorithms, and susceptibility to environmental influences, making it difficult to efficiently identify the movements and postures of the elderly and service personnel.
By acquiring acceleration and angular velocity data, the standard deviation is calculated to determine the current state, and an appropriate attitude algorithm model is selected. Combined with feature extraction and preprocessing techniques, the action posture is identified.
It improves the efficiency of human motion and posture recognition, reduces dependence on the environment, and is suitable for motion and posture recognition in wearable devices.
Smart Images
Figure CN115905828B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of Internet technology, and in particular to a method, apparatus, storage medium and wearable device for action and posture recognition. Background Technology
[0002] With the increasing trend of population aging, monitoring the daily lives of the elderly and detecting unexpected emergencies such as postural issues have become important aspects of elderly care services. The daily movements of the elderly are a crucial indicator of their quality of life and health status; therefore, identifying and recording the daily movements of the elderly and monitoring their daily lives is of great significance.
[0003] Furthermore, my country's elderly care services currently rely mainly on manual management and services. During the manual service process, the service actions of service personnel are an important basis for verifying whether they have completed the service items (such as washing hair, cutting hair, changing clothes, turning over, sweeping and mopping the floor, etc.). Therefore, identifying and recording the service actions of service personnel and supervising the service process is of great significance.
[0004] Existing human posture detection technologies are mainly based on image recognition. Image recognition technology uses cameras to collect sequences of human motion images and uses image processing methods to detect and recognize human posture. This method has the disadvantages of large data collection volume, complex algorithms, and susceptibility to the influence of the surrounding environment and limited detection range.
[0005] Therefore, existing technologies have shortcomings and need to be improved and developed. Summary of the Invention
[0006] This application provides a method, apparatus, storage medium, and wearable device for recognizing human body postures, which can improve the efficiency of recognizing human body postures.
[0007] This application provides an action posture recognition method, the method including:
[0008] Acquire multiple acceleration and angular velocity data generated by the target object during the execution of its current action posture;
[0009] From each acceleration data point, obtain acceleration values in at least one direction to obtain multiple acceleration values in at least one direction;
[0010] Calculate the standard deviation of multiple acceleration values, and determine the current state of the target object when performing the current action posture based on the calculated standard deviation;
[0011] Based on the current state, select the target posture algorithm model corresponding to the current state from multiple preset posture algorithm models;
[0012] Multiple angular velocity data and multiple acceleration data are sent to the target posture algorithm model to output the current action posture of the target object.
[0013] In the action posture recognition method described in this application embodiment, before sending the plurality of angular velocity data and the plurality of acceleration data to the target posture algorithm model and outputting the current action posture of the target object, the method further includes:
[0014] Preprocessing is performed on multiple angular velocity data and multiple acceleration data;
[0015] Based on the multiple accelerations obtained after preprocessing, a resultant acceleration sequence is generated, and based on the multiple angular velocities obtained after preprocessing, a resultant angular velocity sequence is generated.
[0016] Feature extraction is performed on the resultant acceleration sequence and the resultant angular velocity sequence, respectively. The extracted features include time-domain features, frequency-domain features, and data change rate features.
[0017] The step of sending multiple angular velocity data and multiple acceleration data to the target posture algorithm model and outputting the current action posture of the target object includes:
[0018] The time-domain features, frequency-domain features, and data change rate features are sent to the target pose algorithm model, and the current action pose of the target object is output.
[0019] In the action posture recognition method described in this application embodiment, the preset posture algorithm model includes a motion posture algorithm model and a static posture algorithm model. When the target posture algorithm model is a motion posture algorithm model, the step of sending the time domain features, frequency domain features, and data change rate features to the target posture algorithm model and outputting the current action posture of the target object includes:
[0020] The motion posture algorithm model compares and analyzes the time domain features, frequency domain features, and data change rate features with the preset time domain features, preset frequency domain features, and preset data change rate features corresponding to each preset motion posture stored in itself, generating multiple matching results, wherein each preset motion posture corresponds to one matching result.
[0021] Obtain the matching value from each matching result, compare the matching values, take the preset action posture corresponding to the matching result with the highest matching value as the current action posture of the target object, and output the current action posture.
[0022] In the action posture recognition method described in this application embodiment, the preset posture algorithm model includes a motion posture algorithm model and a static posture algorithm model. When the target posture algorithm model is a static posture algorithm model, the step of sending the time domain features, frequency domain features, and data change rate features to the target posture algorithm model and outputting the current action posture of the target object includes:
[0023] The time-domain features, frequency-domain features, and data change rate features obtained are analyzed and calculated using a static attitude algorithm model to obtain a static state data value.
[0024] The static state data value is compared with a preset node threshold. Based on the comparison result, the current action posture of the target object is identified and the current action posture is output.
[0025] In the action posture recognition method described in the embodiments of this application, the current state includes a motion state and a stationary state. When the current state of the target object is a motion state, the preprocessing includes data denoising, data zero-bias error compensation, and data interference removal.
[0026] When the current state of the target object is a static state, the preprocessing includes data denoising and data zero-bias error compensation.
[0027] In the action posture recognition method described in this application embodiment, the step of calculating the standard deviation of multiple acceleration values and determining the current state of the target object when performing the current action posture based on the calculated standard deviation includes:
[0028] Calculate the standard deviation of multiple acceleration values and compare the standard deviation with a preset standard deviation threshold;
[0029] If the standard deviation is less than or equal to the preset standard deviation threshold, then the current state of the target object when performing the current action posture is determined to be a stationary state.
[0030] If the comparison result is that the standard deviation is greater than the preset standard deviation threshold, then the current state of the target object when performing the current action posture is determined to be a motion state.
[0031] In the action posture recognition method described in this application embodiment, the preset posture algorithm model includes a motion posture algorithm model and a static posture algorithm model. The step of selecting a target posture algorithm model corresponding to the current state from multiple preset posture algorithm models based on the current state includes:
[0032] If the current state is a stationary state, then select the stationary attitude algorithm model corresponding to the stationary state from multiple preset attitude algorithm models;
[0033] If the current state is a motion state, then select the motion posture algorithm model corresponding to the static state from multiple preset posture algorithm models.
[0034] This application also provides an action posture recognition device, which includes a processor and a memory. The memory stores a computer program, and the processor executes the action posture recognition method described in any of the above embodiments by calling the computer program stored in the memory.
[0035] This application also provides a computer-readable storage medium storing a computer program that, when run on a computer, causes the computer to execute the action posture recognition method described in any of the above embodiments.
[0036] This application also provides a wearable device, which includes the motion and posture recognition device described in the above embodiments.
[0037] The action posture recognition method provided in this application first acquires multiple acceleration data and multiple angular velocity data generated by the target object during the execution of the current action posture. Then, it obtains acceleration values in at least one direction from each acceleration data to obtain multiple acceleration values in at least one direction. Next, it calculates the standard deviation of the multiple acceleration values and determines the current state of the target object when executing the current action posture based on the standard deviation. Then, based on the determined current state, it selects a target posture algorithm model corresponding to the current state from multiple preset posture algorithm models. Then, it sends the previously acquired multiple angular velocity data and multiple acceleration data to the target posture algorithm model. Finally, it identifies the current action posture through the previously selected target posture algorithm model, thereby improving the recognition efficiency of human action postures. Attached Figure Description
[0038] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0039] Figure 1 This is a flowchart illustrating the action posture recognition method provided in an embodiment of this application.
[0040] Figure 2 This is a schematic diagram illustrating the process of determining the current action posture of a target object using a static posture algorithm model provided in an embodiment of this application.
[0041] Figure 3 This is a schematic diagram of the action posture recognition device provided in the embodiments of this application.
[0042] Figure 4 This is a schematic diagram of the structure of a wearable device provided in an embodiment of this application. Detailed Implementation
[0043] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the protection scope of this application.
[0044] This application provides a motion posture recognition method that can be applied to wearable devices. The wearable devices may include smart bracelets, smartwatches, smart gloves, smart wristbands, smart clothing, smart pants, and smart shoes, among others.
[0045] Please see Figure 1 , Figure 1 This is a flowchart illustrating the action posture recognition method provided in an embodiment of this application. The action posture recognition method, applied in a wearable device, may include the following steps:
[0046] Step 101: Obtain multiple acceleration data and multiple angular velocity data generated by the target object during the execution of the current action posture.
[0047] This can be achieved by using sensors including accelerometers and gyroscopes to acquire multiple acceleration and angular velocity data points generated by the target object during its current action posture. Since each time point corresponds to one acceleration and one angular velocity data point, the target object will generate multiple acceleration and angular velocity data points during its current action posture.
[0048] The target population can include elderly people and caregivers in nursing homes. It should be noted that the target population is not limited to the elderly and caregivers; those skilled in the art can choose based on actual needs, and no specific limitations are made here.
[0049] Step 102: Obtain acceleration values in at least one direction from each acceleration data to obtain multiple acceleration values in at least one direction.
[0050] Each acceleration data point includes acceleration values along the X-axis, Y-axis, and Z-axis.
[0051] Step 103: Calculate the standard deviation of multiple acceleration values, and determine the current state of the target object when performing the current action posture based on the calculated standard deviation.
[0052] For example, obtain the acceleration value in the Y-axis direction of each acceleration data point from four acceleration data points to obtain four acceleration values a1, a2, a3, and a4. Then calculate the standard deviation of these four acceleration values a1, a2, a3, and a4. Based on the calculated standard deviation, determine the current state of the target object when performing the current action posture.
[0053] In some embodiments, the current state includes a motion state and a stationary state, and calculating the standard deviation of the plurality of acceleration values, and determining the current state of the target object when performing the current action posture based on the calculated standard deviation, includes:
[0054] Calculate the standard deviation of multiple acceleration values and compare the standard deviation with a preset standard deviation threshold;
[0055] If the standard deviation is less than or equal to the preset standard deviation threshold, then the current state of the target object when performing the current action posture is determined to be a stationary state.
[0056] If the comparison result is that the standard deviation is greater than the preset standard deviation threshold, then the current state of the target object when performing the current action posture is determined to be a motion state.
[0057] Since the standard deviation features of the static state and the moving state do not overlap, a specific value is selected in the non-overlapping region as the preset standard deviation threshold. The preset standard deviation threshold is used to determine whether the target object is in a static state or a moving state when performing the current action posture.
[0058] Step 104: Based on the current state, select the target pose algorithm model corresponding to the current state from multiple preset pose algorithm models.
[0059] In some embodiments, the preset posture algorithm model includes a motion posture algorithm model and a static posture algorithm model, and the step of selecting a target posture algorithm model corresponding to the current state from multiple preset posture algorithm models according to the current state includes:
[0060] If the current state is a stationary state, then select the stationary attitude algorithm model corresponding to the stationary state from multiple preset attitude algorithm models;
[0061] If the current state is a motion state, then select the motion posture algorithm model corresponding to the static state from multiple preset posture algorithm models.
[0062] Step 105: Send multiple angular velocity data and multiple acceleration data to the target posture algorithm model, and output the current action posture of the target object.
[0063] The current action posture can be standing, sitting, lying down, washing hair, changing clothes, sweeping, mopping, and turning over.
[0064] In some embodiments, before sending the plurality of angular velocity data and the plurality of acceleration data to the target posture algorithm model and outputting the current motion posture of the target object, the method further includes:
[0065] Preprocessing is performed on multiple angular velocity data and multiple acceleration data;
[0066] Based on the multiple accelerations obtained after preprocessing, a resultant acceleration sequence is generated, and based on the multiple angular velocities obtained after preprocessing, a resultant angular velocity sequence is generated.
[0067] Feature extraction is performed on the resultant acceleration sequence and the resultant angular velocity sequence, respectively. The extracted features include time-domain features, frequency-domain features, and data change rate features.
[0068] The step of sending multiple angular velocity data and multiple acceleration data to the target posture algorithm model and outputting the current action posture of the target object includes:
[0069] The time-domain features, frequency-domain features, and data change rate features are sent to the target pose algorithm model, and the current action pose of the target object is output.
[0070] Acceleration and angular velocity data obtained from sensors are the raw data of human movement. It is difficult to directly extract the current posture of the target object from this raw data; therefore, feature extraction is necessary. This involves generating resultant acceleration and resultant angular velocity sequences through second-order norm operations, and then extracting features from these sequences. These features include time-domain features, frequency-domain features, and the rate of change of the data. Time-domain features include mean, standard deviation, maximum value, minimum value, and sum of squares. Frequency-domain features include peak concatenation points, sub-peak concatenation points, peak bandwidth, and the number of prominent concatenation points. The rate of change of the data is represented by the difference sequences of resultant acceleration and resultant angular velocity. These difference sequences differ significantly under different movement postures.
[0071] The reason for preprocessing multiple angular velocity and acceleration data is that the original data will have certain interference and loss due to the influence of objective conditions of data acquisition signals. In order to ensure the reliability and universality of the algorithm, a series of preprocessing steps are required to extract clean and useful motion signals from the original data.
[0072] In some embodiments, the current state includes a motion state and a stationary state. When the current state of the target object is a motion state, the preprocessing includes data denoising, data zero-bias error compensation, and data interference removal.
[0073] When the current state of the target object is a static state, the preprocessing includes data denoising and data zero-bias error compensation.
[0074] The raw acceleration and angular velocity data collected by the sensors contain noise, interference, and bias errors, which can affect the algorithm's recognition and therefore require processing. Threshold filtering and mean filtering methods can be used to smooth and denoise the acceleration and angular velocity data.
[0075] The reason why preprocessing does not include de-interference when the target object is currently in a static state is that the training data of the static pose algorithm model is more regular than the training data of the motion pose algorithm model, so there is no need to perform de-interference.
[0076] In some embodiments, the preset posture algorithm model includes a motion posture algorithm model and a static posture algorithm model. When the target posture algorithm model is a motion posture algorithm model, sending the time-domain features, frequency-domain features, and data change rate features to the target posture algorithm model and outputting the current action posture of the target object includes:
[0077] The motion posture algorithm model compares and analyzes the time domain features, frequency domain features, and data change rate features with the preset time domain features, preset frequency domain features, and preset data change rate features corresponding to each preset motion posture stored in itself, generating multiple matching results, wherein each preset motion posture corresponds to one matching result.
[0078] Obtain the matching value from each matching result, compare the matching values, take the preset action posture corresponding to the matching result with the highest matching value as the current action posture of the target object, and output the current action posture.
[0079] In some embodiments, the preset posture algorithm model includes a motion posture algorithm model and a static posture algorithm model. When the target posture algorithm model is a static posture algorithm model, sending the time-domain features, frequency-domain features, and data change rate features to the target posture algorithm model and outputting the current motion posture of the target object includes:
[0080] The time-domain features, frequency-domain features, and data change rate features obtained are analyzed and calculated using a static attitude algorithm model to obtain a static state data value.
[0081] The static state data value is compared with a preset node threshold. Based on the comparison result, the current action posture of the target object is identified and the current action posture is output.
[0082] For example, such as Figure 2 As shown, the preset node thresholds include a first preset node threshold and a second preset node threshold. The static state data value is compared with the first preset node threshold. If the static state data value is less than or equal to the first preset node threshold, the target object's current posture is determined to be a standing posture; if the static state data value is greater than the first preset node threshold, the target object's current posture is determined to be a non-standing posture. If the target object's current posture is a non-standing posture, the static state data value is compared with the second preset node threshold. If the static state data value is less than or equal to the second preset node threshold, the target object's current posture is determined to be a reclining posture; if the static state data value is greater than the second preset node threshold, the target object's current posture is determined to be a lying posture.
[0083] In some embodiments, when sending the plurality of angular velocity data and the plurality of acceleration data to the target posture algorithm model and outputting the current motion posture of the target object, the method further includes:
[0084] Multiple acceleration data and multiple angular velocity data are input into each of the preset attitude algorithm models to optimize and train the preset attitude algorithm models.
[0085] All of the above-mentioned optional technical solutions can be combined in any way to form the optional embodiments of this application, and will not be described in detail here.
[0086] In practice, this application is not limited by the execution order of the described steps. Without causing conflicts, some steps may be performed in other orders or simultaneously.
[0087] As can be seen from the above, the action posture recognition method provided in this application, when recognizing a certain current action posture, first acquires multiple acceleration data and multiple angular velocity data generated by the target object during the execution of the current action posture, then acquires the acceleration value in at least one direction from each acceleration data, and obtains multiple acceleration values in at least one direction. Next, the standard deviation of the obtained multiple acceleration values is calculated, and the current state of the target object when executing the current action posture is determined based on the standard deviation. Then, based on the determined current state, a target posture algorithm model corresponding to the current state is selected from multiple preset posture algorithm models. Next, the previously acquired multiple angular velocity data and multiple acceleration data are sent to the target posture algorithm model. Finally, the current action posture is recognized through the previously selected target posture algorithm model, thereby improving the recognition efficiency of human action posture.
[0088] This application also provides a motion posture recognition device, which can be integrated into a wearable device. The wearable device can be a smart bracelet, smartwatch, smart gloves, smart wristband, smart clothing, smart pants, and smart shoes, etc.
[0089] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of a motion posture recognition device provided in an embodiment of this application. The motion posture recognition device 30 includes a memory 120, one or more processors 180, and one or more application programs, wherein the one or more application programs are stored in the memory 120 and configured to be executed by the processors 180; the memory 120 can be used to store application programs and data. The application programs stored in the memory 120 contain executable code. The application programs can form various functional modules. The processor 180 executes various functional applications and data processing by running the application programs stored in the memory 120. In addition, the memory 120 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 120 may also include a memory controller to provide the processor 180 with access to the memory 120.
[0090] The processor 180 is the control center of the device, connecting various parts of the terminal through various interfaces and lines. It performs various functions and processes data by running or executing applications stored in the memory 120 and calling data stored in the memory 120, thereby providing overall monitoring of the device. Optionally, the processor 180 may include one or more processing cores; preferably, the processor 180 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications.
[0091] Specifically, in this embodiment, the memory 120 stores a computer program, and the processor 180 executes the action posture recognition method described in any of the above embodiments by calling the computer program stored in the memory 120.
[0092] Please see Figure 4 , Figure 4 A schematic diagram of a wearable device provided in an embodiment of this application is shown. This wearable device can be used to implement the action posture recognition method provided in the above embodiments. The wearable device 1200 can be a smart bracelet, smart watch, smart glove, smart wristband, smart clothing, smart pants, and smart shoes, etc.
[0093] like Figure 4 As shown, the wearable device 1200 may include an RF (Radio Frequency) circuit 110, a memory 120 including one or more (only one is shown in the figure) computer-readable storage media, an input unit 130, a display unit 140, a sensor 150, an audio circuit 160, a transmission module 170, a processor 180 including one or more (only one is shown in the figure) processing cores, and a power supply 190, etc. Those skilled in the art will understand that... Figure 4 The structure of the wearable device 1200 shown does not constitute a limitation on the wearable device 1200, and may include more or fewer components than shown, or combine certain components, or have different component arrangements. Wherein:
[0094] RF circuit 110 is used to receive and transmit electromagnetic waves, realizing the mutual conversion between electromagnetic waves and electrical signals, thereby enabling communication with communication networks or other devices. RF circuit 110 may include various existing circuit elements used to perform these functions, such as antennas, radio frequency transceivers, digital signal processors, encryption / decryption chips, Subscriber Identity Module (SIM) cards, memory, etc. RF circuit 110 can communicate with various networks such as the Internet, corporate intranets, and wireless networks, or communicate with other devices via wireless networks.
[0095] The memory 120 can be used to store software programs and modules, such as the program instructions / modules corresponding to the action posture recognition method in the above embodiments. The processor 180 executes various functional applications and data processing by running the software programs and modules stored in the memory 120, thereby improving the efficiency of human action posture recognition. The memory 120 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 120 may further include memory remotely located relative to the processor 180, and these remote memories can be connected to the electronic device 1200 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.
[0096] The input unit 130 can be used to receive input digital or character information, and to generate keyboard, mouse, joystick, optical, or trackball signal inputs related to user settings and function control. Specifically, the input unit 130 may include a touch-sensitive surface 131 and other input devices 132. The touch-sensitive surface 131, also known as a touch display screen or touchpad, can collect touch operations performed by the user on or near it (such as operations performed by the user using a finger, stylus, or any suitable object or accessory on or near the touch-sensitive surface 131), and drive the corresponding connection device according to a pre-set program. Optionally, the touch-sensitive surface 131 may include two parts: a touch detection device and a touch controller. The touch detection device detects the user's touch position and the signal generated by 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, sends it to the processor 180, and can receive and execute commands from the processor 180. In addition, the touch-sensitive surface 131 can be implemented using various types such as resistive, capacitive, infrared, and surface acoustic wave. In addition to the touch-sensitive surface 131, the input unit 130 may also include other input devices 132. Specifically, other input devices 132 may include, but are not limited to, one or more of the following: physical keyboard, function keys (such as volume control buttons, power buttons, etc.), trackball, mouse, joystick, etc.
[0097] Display unit 140 can be used to display information input by the user or information provided to the user, as well as various graphical user interfaces of wearable device 1200. These graphical user interfaces can be composed of graphics, text, icons, video, and any combination thereof. Display unit 140 may include display panel 141, optionally configured as LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), etc. Further, touch-sensitive surface 131 may cover display panel 141. When touch-sensitive surface 131 detects a touch operation on or near it, it transmits the information to processor 180 to determine the type of touch event. Subsequently, processor 180 provides corresponding visual output on display panel 141 according to the type of touch event. Although in Figure 4 In this embodiment, the touch-sensitive surface 131 and the display panel 141 are implemented as two separate components to realize input and output functions. However, in some embodiments, the touch-sensitive surface 131 and the display panel 141 can be integrated to realize input and output functions.
[0098] The wearable device 1200 may also include at least one sensor 150, 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 141 according to the ambient light level, and the proximity sensor can turn off the display panel 141 and / or backlight when the wearable device 1200 is moved to the ear. As a type of motion sensor, a gravity acceleration sensor can detect the magnitude of acceleration in various directions (generally three axes). When stationary, it can detect the magnitude and direction of gravity and can be used for applications that identify the posture of the wearable device (such as landscape / portrait switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, tapping), etc. Other sensors that may be configured in the wearable device 1200, such as gyroscopes, barometers, hygrometers, thermometers, and infrared sensors, will not be described in detail here.
[0099] Audio circuitry 160, speaker 161, and microphone 162 provide an audio interface between the user and wearable device 1200. Audio circuitry 160 converts received audio data into electrical signals, which are then transmitted to speaker 161, where they are converted into sound signals for output. Conversely, microphone 162 converts collected sound signals into electrical signals, which are received by audio circuitry 160, converted back into audio data, and then processed by processor 180 before being transmitted via RF circuitry 110 to, for example, another terminal, or output to memory 120 for further processing. Audio circuitry 160 may also include an earphone jack to facilitate communication between external headphones and wearable device 120.
[0100] Wearable device 1200, through transmission module 170 (e.g., Wi-Fi module), can help users send and receive emails, browse web pages, and access streaming media, providing users with wireless broadband internet access. Although Figure 4 The transmission module 170 is shown, but it is understood that it is not a necessary component of the wearable device 1200 and can be omitted as needed without changing the nature of the invention.
[0101] The processor 180 is the control center of the wearable device 1200. It connects various parts of the wearable device 1200 via various interfaces and lines. By running or executing software programs and / or modules stored in the memory 120, and by calling data stored in the memory 120, it performs various functions and processes data of the wearable device 1200, thereby enabling overall monitoring of the human body. Optionally, the processor 180 may include one or more processing cores; in some embodiments, the processor 180 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 180.
[0102] The wearable device 1200 also includes a power supply 190 (such as a battery) to power the various components. In some embodiments, the power supply may be logically connected to the processor 180 through a power management system, thereby enabling functions such as managing charging, discharging, and power consumption through the power management system. The power supply 190 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.
[0103] Although not shown, the wearable device 1200 may also include a camera (such as a front-facing camera and a rear-facing camera), a Bluetooth module, etc., which will not be described in detail here. Specifically, in this embodiment, the display unit 140 of the wearable device 1200 is a touch screen display, and the wearable device 1200 also includes a memory 120 and one or more programs, one or more of which are stored in the memory 120 and configured to be executed by one or more processors 180. One or more programs contain instructions for performing the following operations:
[0104] Acquire multiple acceleration and angular velocity data generated by the target object during the execution of its current action posture;
[0105] From each acceleration data, obtain the acceleration value in at least one direction to obtain multiple acceleration values in at least one direction;
[0106] Calculate the standard deviation of multiple acceleration values, and determine the current state of the target object when performing the current action posture based on the calculated standard deviation;
[0107] Based on the current state, select the target posture algorithm model corresponding to the current state from multiple preset posture algorithm models;
[0108] Multiple angular velocity data and multiple acceleration data are sent to the target posture algorithm model to output the current action posture of the target object.
[0109] In some embodiments, the processor 180 is configured to preprocess the plurality of angular velocity data and the plurality of acceleration data; generate a resultant acceleration sequence based on the plurality of accelerations obtained after preprocessing, and generate a resultant angular velocity sequence based on the plurality of angular velocities obtained after preprocessing; and extract features from the resultant acceleration sequence and the resultant angular velocity sequence respectively, wherein the extracted features include time domain features, frequency domain features and data change rate features.
[0110] In some embodiments, the processor 180 is configured to send the time-domain features, frequency-domain features, and data change rate features to the target pose algorithm model and output the current action pose of the target object.
[0111] In some embodiments, the processor 180 is configured to compare and analyze the time-domain features, frequency-domain features, and data change rate features with the preset time-domain features, preset frequency-domain features, and preset data change rate features corresponding to each preset action posture stored in its own memory, using a motion posture algorithm model, to generate multiple matching results, wherein each preset action posture corresponds to one matching result; obtain the matching value in each matching result, compare the matching values, take the preset action posture corresponding to the matching result with the highest matching value as the current action posture of the target object, and output the current action posture.
[0112] In some embodiments, the processor 180 is used to analyze and calculate the acquired time-domain features, frequency-domain features, and data change rate features through a static posture algorithm model to obtain a static state data value; compare the static state data value with a preset node threshold, identify the current action posture of the target object based on the comparison result, and output the current action posture.
[0113] In some embodiments, the processor 180 is configured to calculate the standard deviation of a plurality of acceleration values and compare the standard deviation with a preset standard deviation threshold; if the standard deviation is less than or equal to the preset standard deviation threshold, the current state of the target object when performing the current action posture is determined to be a stationary state; if the comparison result is that the standard deviation is greater than the preset standard deviation threshold, the current state of the target object when performing the current action posture is determined to be a moving state.
[0114] In some embodiments, the processor 180 is configured to select a stationary posture algorithm model corresponding to the stationary state from a plurality of preset posture algorithm models if the current state is a stationary state; and to select a motion posture algorithm model corresponding to the stationary state from a plurality of preset posture algorithm models if the current state is a motion state.
[0115] As can be seen from the above, this application embodiment provides a wearable device 1200, which performs the following steps: acquiring multiple acceleration data and multiple angular velocity data generated by a target object during the execution of a current action posture; acquiring acceleration values in at least one direction from each acceleration data to obtain multiple acceleration values in at least one direction; calculating the standard deviation of the multiple acceleration values, and determining the current state of the target object when executing the current action posture based on the calculated standard deviation; selecting a target posture algorithm model corresponding to the current state from multiple preset posture algorithm models according to the current state; sending the multiple angular velocity data and multiple acceleration data to the target posture algorithm model, and outputting the current action posture of the target object. Through the above steps, this application embodiment can improve the recognition efficiency of human action postures.
[0116] This application also provides a storage medium storing a computer program. When the computer program is run on a computer, the computer executes the action posture recognition method described in any of the above embodiments.
[0117] It should be noted that, for the action posture recognition method described in this application, those skilled in the art will understand that all or part of the process of the action posture recognition method described in the embodiments of this application can be implemented by a computer program controlling related hardware. The computer program can be stored in a computer-readable storage medium, such as the memory of a wearable device, and executed by at least one processor within the wearable device. During execution, it can include the process of the embodiments of the action posture recognition method described. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), etc.
[0118] The above provides a detailed description of the action posture recognition method, apparatus, storage medium, and wearable device provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and its core ideas. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.
Claims
1. A method for recognizing action postures, characterized in that, The method includes: Acquire multiple acceleration and angular velocity data generated by the target object during the execution of its current action posture; From each acceleration data, obtain the acceleration value in at least one direction to obtain multiple acceleration values in at least one direction; Calculate the standard deviation of multiple acceleration values, and determine the current state of the target object when performing the current action posture based on the calculated standard deviation; Based on the current state, a target attitude algorithm model corresponding to the current state is selected from multiple preset attitude algorithm models, wherein the preset attitude algorithm models include a static attitude algorithm model; The multiple angular velocity data and multiple acceleration data are preprocessed, and a resultant acceleration sequence is generated based on the multiple accelerations obtained after preprocessing, and a resultant angular velocity sequence is generated based on the multiple angular velocities obtained after preprocessing. Feature extraction is performed on the resultant acceleration sequence and the resultant angular velocity sequence, respectively. The extracted features include time-domain features, frequency-domain features, and data change rate features. When the target attitude algorithm model is a static attitude algorithm model, the time domain features, frequency domain features and data change rate features are sent to the static attitude algorithm model. The static attitude algorithm model analyzes and calculates the obtained time domain features, frequency domain features and data change rate features to obtain a static state data value. The static state data value is compared with at least one preset node threshold. Based on the comparison result, the current action posture of the target object is identified and the current action posture is output.
2. The action posture recognition method as described in claim 1, characterized in that, The preset posture algorithm model includes a motion posture algorithm model, and the method includes: The motion posture algorithm model compares and analyzes the time domain features, frequency domain features, and data change rate features with the preset time domain features, preset frequency domain features, and preset data change rate features corresponding to each preset motion posture stored in itself, generating multiple matching results, wherein each preset motion posture corresponds to one matching result. Obtain the matching value from each matching result, compare the matching values, take the preset action posture corresponding to the matching result with the highest matching value as the current action posture of the target object, and output the current action posture.
3. The action posture recognition method as described in claim 1, characterized in that, The current state includes a moving state and a stationary state. When the current state of the target object is a moving state, the preprocessing includes data denoising, data zero-bias error compensation, and data interference removal. When the current state of the target object is a static state, the preprocessing includes data denoising and data zero-bias error compensation.
4. The action posture recognition method as described in claim 1, characterized in that, The current state includes a moving state and a stationary state. The step of calculating the standard deviation of multiple acceleration values and determining the current state of the target object when performing the current action posture based on the calculated standard deviation includes: Calculate the standard deviation of multiple acceleration values and compare the standard deviation with a preset standard deviation threshold; If the standard deviation is less than or equal to the preset standard deviation threshold, then the current state of the target object when performing the current action posture is determined to be a stationary state. If the comparison result is that the standard deviation is greater than the preset standard deviation threshold, then the current state of the target object when performing the current action posture is determined to be a motion state.
5. The action posture recognition method as described in claim 1, characterized in that, The preset posture algorithm model includes a motion posture algorithm model and a static posture algorithm model. The step of selecting a target posture algorithm model corresponding to the current state from multiple preset posture algorithm models based on the current state includes: If the current state is a stationary state, then select the stationary attitude algorithm model corresponding to the stationary state from multiple preset attitude algorithm models; If the current state is a motion state, then select the motion posture algorithm model corresponding to the static state from multiple preset posture algorithm models.
6. A motion posture recognition device, characterized in that, The device includes a processor and a memory, the memory storing a computer program, and the processor executes the action posture recognition method according to any one of claims 1 to 5 by calling the computer program stored in the memory.
7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when run on a computer, causes the computer to perform the action posture recognition method according to any one of claims 1 to 5.
8. A wearable device, characterized in that, The wearable device includes the motion and posture recognition device as described in claim 6.