Motion Sensing Wearable Apparatus For Evaluating Action Quality

Description

CROSS-REFERENCE TO RELATED APPLICATION(S)

This application claims priority to Chinese Patent Application 202411248459.3, filed Sep. 5, 2024, the entire disclosure of which is hereby incorporated by reference.

TECHNICAL FIELD

The present disclosure relates to motion sensing wearable technology, and more particularly to methods, apparatuses, storage media, and electronic devices for evaluating the quality of physical actions.

BACKGROUND

As portable electronic devices become more intelligent, more and more users are using portable electronic devices as exercise aids. For example, users utilize a portable electronic device to record exercise information during fitness activities such as running, swimming, or strength training, and one or more performance parameters of the fitness activities are measured based on the exercise information recorded by the portable electronic device. However, the performance parameters of the fitness activities are usually abstract and cannot meet the needs of the users.

SUMMARY

In view of this, the embodiments of the present disclosure provide methods and apparatuses for action quality evaluation, a storage medium and an electronic device.

According to a first aspect of the present disclosure, a method for action quality evaluation is provided, including:

• • obtaining motion data collected during a fitness activity of a user;
  • determining a value of at least one action evaluation metric of the user according to the motion data;
  • determining an action quality evaluation result of the user according to the value of the at least one action evaluation metric;
  • the at least one action evaluation metric includes at least one of:
  • an action consistency metric, for evaluating a consistency of amplitudes and completion time duration of multiple fitness actions performed by the user;
  • an action stability metric, for evaluating a degree of jitter or wobble of one or more fitness actions performed by the user;
  • an action continuity metric, for evaluating a continuity degree of one or more fitness actions performed by the user;
  • an action variability metric, for evaluating an intensity variability of multiple fitness actions performed by the user; or
  • a force control metric, for evaluating force exertion condition during one or more fitness actions performed by the user.

In conjunction with any implementation provided in the present disclosure, determining the value of at least one action evaluation metric of the user according to the motion data includes:

• • extracting metric feature data corresponding to each of the at least one fitness action performed by the user according to the motion data; and
  • obtaining the value of the at least one action evaluation metric for the user according to the metric feature data corresponding to each of the at least one fitness action.

In conjunction with any implementation provided in the present disclosure, determination of the metric feature data corresponding to each fitness action is independent of an action type of the each fitness action.

In conjunction with any implementation provided in the present disclosure, determination of the metric feature data corresponding to each of the at least one fitness action includes performing data reconstruction processing on the motion data, so as to reduce or eliminate an influence of the action type.

In conjunction with any implementation provided in the present disclosure, the metric feature data corresponding to each of the at least one fitness action is obtained by performing singular value decomposition processing on the motion data.

In conjunction with any implementation provided in the present disclosure, determining the value of at least one action evaluation metric of the user according to the motion data includes:

• • obtaining at least one motion data segment according to the motion data, each motion data segment includes motion data corresponding to at least one fitness action;
  • performing data reconstruction processing on each of the at least one motion data segment to obtain at least one reconstructed data segment; and
  • obtaining the value of the at least one action evaluation metric of the user according to the at least one reconstructed data segment.

In conjunction with any implementation provided in the present disclosure, the reconstructed data segment includes at least one of virtual principal axis data or virtual auxiliary axis data, where the virtual principal axis data is used for representing a signal response to primary motion of the user, and the virtual auxiliary axis data is used for representing a signal response to sources other than the primary motion.

In conjunction with any implementation provided in the present disclosure, the action quality evaluation result of the user includes at least one of:

• • a statistical result of the value of the at least one action evaluation metric of the user;
  • evaluative description of the at least one action evaluation metric of the user; or
  • an action improvement suggestion for the user.

According to a second aspect of the present disclosure, another method for action quality evaluation is provided, including:

• • obtaining motion data collected during a fitness activity of a user;
  • obtaining at least one motion data segment according to the motion data, each motion data segment includes motion data corresponding to at least one fitness action;
  • performing data reconstruction processing on each of the at least one motion data segment to obtain at least one reconstructed data segment; and
  • obtaining an action quality evaluation result of the user according to the at least one reconstructed data segment.

In conjunction with any implementation provided in the present disclosure, the data reconstruction processing is used to reduce or eliminate an influence of an action type of the fitness action.

In conjunction with any implementation provided in the present disclosure, the data reconstruction processing includes singular value decomposition processing.

In conjunction with any implementation provided in the present disclosure, the reconstructed data segment includes at least one of virtual principal axis data or virtual auxiliary axis data, where the virtual principal axis data is used for representing a signal response to primary motion of the user, and the virtual auxiliary axis data is used for representing a signal response to sources other than the primary motion.

In conjunction with any implementation provided in the present disclosure, obtaining the action quality evaluation result of the user according to the at least one reconstructed data segment includes:

• • determining the value of a first action evaluation metric of one or more fitness actions performed by the user according to virtual principal axis data in the reconstructed data segment, and the first action evaluation metric includes at least one of an action consistency metric, action stability metric, an action continuity metric, an action variability metric or an action regularity metric; or
  • determining the value of a second action evaluation metric of one or more fitness actions performed by the user according to virtual auxiliary axis data in the reconstructed data segment, and the second action evaluation metric includes at least one of an action stability metric or a force control metric.

According to a third aspect of the present disclosure, an apparatus for action quality evaluation is provided, including: at least one unit or module, configured to execute the method for action quality evaluation in any possible implementation of any aspect of the present disclosure.

In some embodiments, the at least one unit or module includes:

• • a motion data acquisition module, configured to acquire the motion data collected during the user's fitness activity;
  • a metric value determination module, configured to determine the value of at least one action evaluation metric of the user based on the motion data; and
  • an assessment result determination module, configured to determine the action quality evaluation result of the user according to the value of the at least one action evaluation metric.

In some other embodiments, the at least one unit or module includes:

• • a motion data acquisition module, configured to acquire the motion data collected during the user's fitness process.
  • a motion data segment determination module, configured to obtain at least one motion data segment based on the motion data, where each motion data segment includes motion data corresponding to at least one fitness action;
  • a reconstructed data segment determination module, configured to perform data reconstruction processing on each of the at least one motion data segment to obtain at least one reconstructed data segment; and
  • a second assessment result determination module, configured to obtain the action quality evaluation result of the user based on the at least one reconstructed data segment.

According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided, with machine-readable instructions stored thereon, and when the machine-readable instructions are executed by a processor, the processor is caused to implement the method for action quality evaluation in any implementation of any aspect of the present disclosure.

According to a fifth aspect of the present disclosure, an electronic device is provided, including:

• • a processor; and
  • a memory for storing instructions executable by the processor;
  • where the processor is configured to execute the method for action quality evaluation in any implementation of any aspect of the present disclosure.

The technical solutions provided in the embodiments of the present disclosure can achieve the following beneficial effects:

In the method and apparatus for action quality evaluation, the storage medium and the electronic device provided in the embodiments of the present disclosure, the motion data collected during the user's fitness process is firstly obtained, and then, the value of at least one action evaluation metric of the user is determined according to the motion data. Finally, the action quality evaluation result of the user is determined according to the value of the at least one action evaluation metric. With the scheme for the action quality evaluation provided in the embodiments of the present disclosure, the action quality evaluation result determined based on the values of one or more action evaluation metrics are used to characterize the user's action quality, which is easier for the user to understand, thereby meeting the user's needs.

It should be understood that the above mentioned general description and the following detailed description are examples and explanatory only, and do not limit the present disclosure.

BRIEF DESCRIPTION OF THE DRAWINGS

Drawings are provided to better understand the present disclosure and are not intended to limit the present disclosure.

In order to more clearly illustrate the technical solutions in one or more embodiments of the present disclosure or related technologies, the following briefly introduces the drawings that need to be used in the description of the embodiments or related technologies:

FIG. 1 is a structural schematic diagram of a system for action quality evaluation according to an example embodiment of the present disclosure;

FIG. 2 is a flowchart of a method for action quality evaluation according to an example embodiment of the present disclosure;

FIG. 3 is an example interface of an action quality evaluation result according to an example embodiment of the present disclosure;

FIG. 4 is another flowchart of the method for action quality evaluation according to an example embodiment of the present disclosure;

FIG. 5 is a structural schematic diagram of an apparatus for action quality evaluation shown according to an example embodiment of the present disclosure;

FIG. 6 is another structural schematic diagram of the apparatus for action quality evaluation according to an example embodiment of the present disclosure;

FIG. 7 is a structural schematic diagram of an electronic device according to an example embodiment of the present disclosure.

DETAILED DESCRIPTION

Example embodiments will be described herein in detail, and examples thereof are shown in the accompanying drawings. In the following description, if the drawings are referred to, unless otherwise indicated, a same reference number in different drawings represents a same or similar elements.

In order to enhance the effect of exercise, users tend to know their performance during the exercise or fitness activities. Therefore, users can wear, hold or have portable electronic devices being attached to during the exercise process, and utilize at least one sensor of the portable electronic devices to measure physiological and/or motion data of the users during exercise. For example, parameters such as speed, acceleration and/or the like can be measured. The portable electronic devices may be wearable devices for example, which are worn on the wrist, ear, eye, head, neck, waist, foot and/or other body parts of users, or may be portable devices attached to clothing, jewelry or other electronic devices. The portable electronic devices may collect one or more types of physiological data of the users during the exercise or fitness activities, such as heart rate, blood oxygen, blood pressure, body temperature and/or the like, or collect the motion data of the users, or collect the environmental data of the environment where the users are performing fitness activities. Then, the collected one or more types of data can be analyzed and processed to obtain one or more motion performance indicators of the users and provide the motion performance indicators to the users, which is beneficial for the users to understand their own motion performance and improve or strengthen one or some aspects, thereby improving the effect of exercise.

The embodiments of the present disclosure provide a system for action quality evaluation, which may obtain the motion data collected during the workout or fitness activities of the user, and determine a value of at least one action evaluation metric of the user according to the motion data. Then, the action quality evaluation result of the user is determined according to the value of the at least one action evaluation metric.

Referring to FIG. 1, which shows a structural schematic diagram of the system for action quality evaluation applicable for the embodiments of the present disclosure. The system may include a wearable device 102, a server 104 and an intermediate device 106.

The wearable device 102 is a computing device configured to be worn by an individual during operation. The wearable device 102 may be implemented as a wristwatch, wristband, bracelet, armband, leg-band, leg-ring, etc., or in the form of other types of wearable devices. The wearable device 102 includes one or more sensors 108 for detecting the motion data of the user wearing the wearable device 102. In some examples, the sensors 108 may include a motion sensor, which is configured to measure the motion data of the user. For example, the sensors 108 may include one or more of an accelerometer, a gyroscope, a magnetometer, a speed sensor, a displacement sensor, or other types of sensors, or a combination thereof. The motion parameters represent measurable parameters associated with a movement process of the user wearing the wearable device 102. In some examples, the motion data may include one or more of acceleration, velocity, displacement, Euler angles, or other types of motion data as the user performs various fitness activities. The sensors 108 may continuously, intermittently, or otherwise periodically collect the motion data of the user. In some other examples, the sensor 108 may include one or more physiological sensors for measuring one or more physiological parameters of the user, such as one or more physiological parameters of the user, such as heart rate, body temperature, emotion, stress, blood oxygen, blood pressure, blood glucose, or other types of blood analyte metrics or the like. In some other examples, the sensors 108 may include one or more environmental sensors for measuring the environmental data of the environment where the user is currently located. The environmental data may include, for example, one or more of: positioning, indoor or outdoor, ambient temperature, ambient humidity, air quality, barometric pressure, altitude, etc.

The wearable device 102 further includes a processor 110 and a memory 111. The memory 111 stores application programs or other executable instructions. The processor 110 is configured to run the application programs or other executable instructions to process the motion data generated based on the motion parameters collected by the sensors 108.

The wearable device 102 may further include input and/or output components, such as one or more of a display, an audio output component, a tactile output component, and/or other types of output components. The display may include one or more of a touch sensor, a force sensor, a pressure sensor, etc., which is not limited herein.

The server 104 may include a hardware server (e.g., a server), a software server (e.g., a web server), and/or a virtual server. The server program 112 is software for detecting movement conditions of the user wearing the wearable device 102.

The server program 112 may access the database 114 on the server 104 to perform at least some functions of the server program 112. The database 114 is a database or other types of data storage for storing, managing, or otherwise providing data for delivering the functions of the server program 112. For example, the database 114 may be a relational database management system, an object database, an XML database, a configuration management database, a management information repository, one or more flat files, other suitable non-transient storage mechanisms, or a combination thereof.

The intermediate device 106 is a device for facilitating communication between the wearable device 102 and the server 104. The intermediate device 106 may be a computing device, such as a mobile terminal (e.g., a smartphone, a tablet, a laptop, or other mobile devices) or other computers (e.g., a desktop computer or other non-mobile computers). Alternatively, the intermediate device 106 may be a network hardware, such as a router, a switch, a load balancer, other network devices, or a combination thereof. Alternatively, the intermediate device 106 may be other network connection devices. For example, the intermediate device 106 may be a networked power charger for the wearable device 102.

For example, depending on the specific implementations of the intermediate device 106, the intermediate device 106 may run the application program 118. The application program 118 configures the intermediate device 106 to send data to the wearable device 102 or receive data from the wearable device 102, and/or to send data to the server 104 or receive data from the server 104. Additionally, the application program 118 be configured to receive commands from the user of the intermediate device 106 in response to the operations of the user. For example, in the case that the intermediate device 106 is a computing device with a touch screen display, the user of the intermediate device 106 may issue commands by touching a part of the display corresponding to a user-interface element of the application program.

In some implementations, the client device is granted permission to access the server program 112. In some examples, the client device may be a mobile device, such as a smartphone, a tablet, a laptop, etc. In some other examples, the client device may be a desktop computer or other non-mobile computers. The client device may run a client application program to communicate with the server program 112. For example, the client application program may be a mobile application capable of accessing some or all of the functions and/or data of the server program 112. In some examples, the client device may communicate with the server 104 over the network 116. In some of such implementations, the client device may be the intermediate device 106.

In some implementations, the intermediate device 106 receives data from the wearable device 102 with a short-range communication protocol. The short-range communication protocol for example, may be Bluetooth®, Bluetooth® Low Energy, infrared, Z-Wave, ZigBee, other protocols, or a combination thereof. The intermediate device 106 sends the data received from the wearable device 102 to the server 104 via the network 116. The network 116 for example, may be a local area network, a wide area network, a machine-to-machine network, a virtual private network, or other public or private networks. The network 116 may use a remote communication protocol, such as Ethernet, TCP, IP, power-line communication, Wi-Fi, GPRS, GSM, CDMA, other protocols, or a combination thereof.

In some implementations, the intermediate device 106 may be omitted. For example, the wearable device 102 is configured to communicate directly with the server 104 via the network 116. The direct communication between the wearable device 102 and the server 104 via the network 116 may include for example the use of a remote, low-power system or other communication mechanisms.

In the embodiments of the present disclosure, the aforementioned method for action quality evaluation may be performed by any one of the wearable device, the intermediate device (such as a mobile terminal), or the server. Alternatively, the aforementioned method for action quality evaluation may be performed cooperatively by at least two of the wearable device, the intermediate device, and the server.

In some examples, the wearable device obtains the motion data of the user collected by at least one sensor during the fitness activities and sends the motion data to the intermediate device. The intermediate device analyzes and processes the motion data to obtain the value of at least one action evaluation metric of the user and sends the value of the at least one action evaluation metric to the server. The server determines the action quality evaluation result of the user based on the value of at least one action evaluation metric, and then sends the action quality evaluation result to at least one of the intermediate device or the wearable device for output to the user. In some other examples, the wearable device obtains the motion data of the user collected by at least one sensor during the fitness process and then sends the data to the intermediate device or the server. The intermediate device or the server analyzes and processes the motion data to obtain the value of at least one action evaluation metric of the user and determines the action quality evaluation result of the user based on the value of the action evaluation metric. Then, the intermediate device or the server sends the action quality evaluation result to the wearable device for output to the user. In some other examples, the processor of the wearable device obtains the motion data of the user collected by at least one sensor during the fitness process, analyzes and processes the motion data to obtain the value of at least one action evaluation metric of the user, determines the action quality evaluation result of the user based on the value of at least one action evaluation metric, and outputs the action quality evaluation result to the user, or sends the action quality evaluation result to the intermediate device so that the intermediate device outputs it to the user.

It should be noted that the foregoing description is merely illustrative and intended to enable those skilled in the art to better understand the technical solutions of the embodiments of the present disclosure. The present disclosure does not impose any specific limitation on the object executing the procedures of the aforementioned method for action quality evaluation.

A detailed description of the method for action quality evaluation in some example embodiments of the present disclosure is provided below in conjunction with the accompanying drawings.

FIG. 2 is a flowchart of the method for action quality evaluation according to an example embodiment of the present disclosure. As shown in FIG. 2, the method in the example embodiment may include 201 to 203. The method may be executed by the system for action quality evaluation to evaluate the action quality of the user during fitness activities such as strength training or the like.

At 201, motion data collected during a fitness activity is obtained.

The motion data may include at least one of acceleration data, angular velocity data, velocity data, displacement data, Euler angle data, or geomagnetic sensor data.

In some examples, the system for action quality evaluation may include at least one of an accelerometer, a gyroscope, or a geomagnetic sensor. Among other things, the accelerometer may be configured to collect acceleration data, velocity data, or displacement data, the gyroscope may be configured to collect angular velocity data or Euler angle data, and the geomagnetic sensor may be configured to collect data about the earth's magnetic field. Alternatively, the system for action quality evaluation may further include other types of motion sensors, and the type of motion sensors is not limited herein.

In some examples, after obtaining the raw data collected by the motion sensor, at least a portion of the raw data is processed to obtain at least one of velocity data, displacement data, Euler angle data or the like. The types of data included in the motion data are not limited herein.

The processor may obtain the motion data for a period of time, or obtain the motion data within a window of a fixed length.

At 202, a value of at least one action evaluation metric of the user is determined based on the motion data.

The at least one action evaluation metric includes at least one of an action consistency metric, an action stability metric, an action continuity metric, an action variability metric, or a force control metric.

The action consistency metric is used for evaluating a consistency of at least one of amplitudes or completion times of a plurality of fitness actions performed by the user. Generally, the more consistent the actions are, the higher the action quality is.

The action stability metric is used for evaluating a degree of jitter or wobble of one or more fitness actions performed by the user, and may be used to measure a deviation of the user in the non-force-exertion direction. It may refer to a degree of jitter or wobble of at least a part of the user's body in terms of displacement or angle in one or more directions. For example, the action stability metric refers to the shifting, swaying, jittering, tilting or the like of at least a portion of the user's body in the non-force-exertion direction. Generally, the more stable the action is, the higher the action quality is.

The action continuity metric is used for evaluating an action continuity of one or more fitness actions performed by the user, and may be used to measure whether there are sudden or unnecessary pauses during the execution of one or more fitness actions. Generally, the more coherent the actions are, the better the action quality is.

The action variability metric is used for evaluating an intensity variation of a plurality of fitness actions performed by the user, and may be used to measure whether the user reaches exhaustion or fatigue.

The force control metric is used for evaluating the force exertion condition during one or more fitness actions performed by the user, and may be used to measure the force application condition of the user during the execution of fitness actions, such as, for example, determine whether there is sudden force application or a sharp change in force, or whether the user relies on inertia to complete the actions, or the like.

In some embodiments, if the fitness actions performed by the user are symmetrical actions, or, if the user wears sensors on different parts of the body, the at least one action evaluation metric may further include an action balance metric, which is used for evaluating a left-right balance degree of the fitness actions performed by the user. In some other embodiments, the at least one action evaluation metric may further include an action regularity metric, which is used for evaluating whether one or more fitness actions performed by the user are standard. Optionally, the at least one action evaluation metric may include other types of metrics, which are not limited herein.

The method for action quality evaluation in the embodiments of the present disclosure is applicable to various types of exercises, such as strength training (e.g., squats, sit-ups, jumping jacks, burpees, push-ups, pull-ups, presses, rows, lateral raises, deadlifts, etc.), running, swimming, etc.

In some embodiments, the motion data is segmented to obtain a plurality of motion data segments, where each motion data segment may include one or more fitness actions. For example, the motion data of two fitness actions performed by the user may be segmented into a motion data segment corresponding to fitness action/and a motion data segment corresponding to fitness action 2. Then, by processing each of the multiple motion data segments, the values of the action evaluation metrics of a plurality of fitness actions corresponding to the multiple motion data segments are obtained.

In some embodiments, cycle detection may be performed on the obtained motion data to segment the motion data into a plurality of motion data segments, where each motion data segment may include one or more cycles. For example, based on the obtained motion data, one or more of the acceleration data, velocity data, displacement data or the like is subjected to waveform analysis, such as at least one of peak detection, valley detection, zero-point detection or the like, to determine the action cycles and identify different fitness actions.

In some other embodiments, one or more types of pre-processing may be performed on the obtained motion data, such as one or more of filtering, denoising, and data conversion processing. In an optional example, a coordinate system conversion processing is performed on the obtained motion data to convert the motion data from the device coordinate system to the world coordinate system. The coordinate-system conversion processing may be implemented with any suitable attitude analysis algorithm or model, which is not limited herein.

In some embodiments, in order to reduce or eliminate the influence of the type of exercise (interchangeably used with the terms “action type”, “motion type” and “exercise type”) on the determination of the action evaluation metrics of fitness actions, so that the processing of the motion data is independent of a specific type of exercise and be universally applicable to a variety of types of exercises, after segmenting the motion data to obtain a plurality of motion data segments, data reconstruction processing may be performed on each of the plurality of motion data segments to obtain a plurality of reconstructed data segments. Then, based on the obtained plurality of reconstructed data segments, the values of the action evaluation metrics of a plurality of fitness actions corresponding to the plurality of motion data segments are determined.

In some embodiments, the motion data may include data from more than two axes, such as acceleration data in three axes, acceleration data and angular velocity data, or accelerometer data, gyroscope data and magnetometer data, or the like. The data reconstruction processing may be performed on each of the plurality of motion data segments in the following manner: for each motion data segment, an algorithm, data decomposition, data analysis, or coordinate transformation processing may be performed on the motion data segment by using an algorithm such as SVD (Singular Value Decomposition) algorithm, IDA (Independent Component Analysis) algorithm, or PCA (Principal Component Analysis) algorithm or the like, to obtain multi-axis processed data, and then, data reconstruction is performed on the obtained multi-axis processed data to obtain a reconstructed data segment. The data reconstruction processing separates the primary signal response and the secondary signal response from the motion data segment, or extract the primary data features and the secondary data features from the motion data segment, thereby facilitating subsequent determination of the values of the motion evaluation metrics.

In some optional embodiments, the reconstructed data segment may contain a primary signal response or a signal response to primary motion (i.e., movement of the user, such as movement caused by performing the fitness action), and/or, contain a secondary signal response or a signal response to other sources other than the primary motion such as noise. For example, the reconstructed data segment may include at least one of virtual primary-axis data or virtual secondary-axis data. The virtual primary-axis data and the virtual secondary-axis data are representations of data on the virtual coordinate axes constructed during the data reconstruction processing, and correspond to the signal response to the primary motion of the user and the signal response to other sources, respectively. In this way, the value of at least one motion evaluation metric of the user is determined based on at least one of the virtual primary-axis data or the virtual secondary-axis data.

In some embodiments, singular value decomposition processing may be performed on the motion data segment to obtain multi-axis processed data, and then, data of the first two or more axes associated with a stronger signal response in the multi-axis processed data is reconstructed to obtain the virtual primary-axis data, and data of the last one or more axes associated with a weaker signal response in the multi-axis processed data is reconstructed to obtain the virtual secondary-axis data. In this way, the obtained virtual primary-axis data is associated with a main movement of the user, and is reconstructed to include a component corresponding to larger singular values. The virtual secondary-axis data is reconstructed to include a component corresponding to smaller singular values. The virtual secondary-axis data is used to represent the signal response to other sources other than the primary motion of the user, such as, for example, representing a relatively small variation in the motion data caused by noise, minor distractions, or other components of movement that are less significant compared to the main movement of the user.

In some embodiments, each of the multiple motion data segments may include at least one of acceleration data, velocity data, displacement data, or Euler angle data, or may further or alternatively include other types of data. Correspondingly, the data reconstruction processing may be performed on each type of motion data included in the motion data segment separately to obtain the reconstructed data segment corresponding to each type of motion data, or, the data reconstruction processing may be performed on only a portion of the motion data with a particular type to obtain the reconstructed data segment corresponding to the particular type of motion data, and the reconstructed data segment corresponding to other types of motion data may be obtained based on the reconstructed data segment corresponding to the particular type of motion data, but the present disclosure does not limit hereto.

In an example, in response to the motion data segment including at least one of acceleration data or angular velocity data, the data reconstruction processing is performed on at least one of the acceleration data or the angular velocity data included in the motion data segment to obtain at least one of a reconstructed acceleration data segment or a reconstructed angular velocity data segment. The reconstructed acceleration data segment may include at least one of virtual primary-axis acceleration data or virtual secondary-axis acceleration data, and the reconstructed angular velocity data segment may include at least one of virtual primary-axis angular velocity data or virtual secondary-axis angular velocity data.

In another example, in response to the motion data segment including at least one of velocity data or Euler angle data, the data reconstruction processing is performed on at least one of the velocity data or the Euler angle data included in the motion data segment to obtain at least one of a reconstructed velocity data segment or a reconstructed Euler angle data segment. The reconstructed velocity data may include at least one of virtual primary-axis velocity data or virtual secondary-axis velocity data, and the reconstructed Euler angle data segment may include at least one of virtual primary-axis Euler angle data or virtual secondary-axis Euler angle data.

In another example, in response to the displacement data being included in the motion data segment, the data reconstruction processing is performed on the displacement data included in the motion data segment to obtain a reconstructed displacement data segment. The reconstructed displacement data segment may include at least one of virtual primary-axis displacement data or virtual secondary-axis displacement data.

In another example, after performing data reconstruction processing on the motion data segment to obtain at least one of the reconstructed acceleration data segment or the reconstructed angular velocity data segment, at least one of the reconstructed velocity data segment or the reconstructed displacement data segment may be further obtained based on the reconstructed acceleration data segment, and alternatively, a reconstructed Euler angle data segment may be further obtained based on the reconstructed angular velocity data segment. The embodiments of the present disclosure do not limit the manner of obtaining the respective reconstructed data segments.

In addition, the foregoing embodiments are described in terms of performing data reconstruction processing on each of the plurality of motion data segments separately as an example, where the object of the data reconstruction processing is a single motion data segment. In some other embodiments, the object of the data reconstruction processing is two or more motion data segments. e.g., the data reconstruction processing may be performed jointly on two or more adjacent motion data segments, or the data reconstruction processing on a motion data segment is performed with reference to at least one adjacent motion data segment.

In some embodiments, after the data reconstruction processing is performed on each of the plurality of motion data segments to obtain multiple reconstructed data segments, for each reconstructed data segment, metric feature data of at least one action evaluation metric for one or more fitness actions corresponding to the reconstructed data segment is calculated, and then, the value corresponding to the at least one action evaluation metric is determined respectively based on the calculated metric feature data corresponding to the at least one action evaluation metric.

In some embodiments, some or each of the plurality of motion data segments may correspond to one fitness action. Correspondingly, the value of at least one action evaluation metric corresponding to the fitness action is obtained based on the reconstructed data segment corresponding to the motion data segment.

In some other embodiments, some or each of the plurality of motion data segments may correspond to two or more fitness actions. In this case, the value of the at least one action evaluation metric corresponding to the two or more fitness actions may be obtained, where the at least one action evaluation metric is used for evaluating the two or more fitness actions. Alternatively, the value of the at least one action evaluation metric corresponding to each of the two or more fitness actions may be obtained, or both of the above may be obtained.

In some other embodiments, during the data segmentation processing, the motion data corresponding to each fitness action may be obtained, and then at least two fitness actions that meet a specific condition may be divided into a group. For example, based on the motion data corresponding to at least two neighboring fitness actions, the similarity between these at least two neighboring fitness actions may be calculated based on the motion data corresponding to the at least two neighboring fitness actions. If the resulted similarity is greater than a preset similarity threshold or meets other conditions, which indicates that these two neighboring fitness actions are likely to be of the same type of fitness actions, the at least two neighboring fitness actions may be divided into a group. In another example, user input may be received, such as input for selecting, starting, or stopping a certain type of exercise, input for indicating a change in the type of fitness actions or exercises, input for indicating a switch of the groups or the like, and then at least two neighboring fitness actions may be divided into a group based on the user input. The embodiments of the present disclosure do not limit the implementations of grouping of the fitness actions.

In this case, the action quality of a fitness action group may be evaluated. For example, a value of one or more action evaluation metrics corresponding to a fitness action group is determined. In some examples, the motion data corresponding to at least two fitness actions that are divided into a group is taken as a motion data segment, that is, the motion data segment includes the motion data corresponding to at least two fitness actions belonging to the same group. In some other examples, the motion data corresponding to at least two fitness actions that are divided into a group belong to different motion data segments. For instance, a motion data segment contains the motion data of one fitness action. In this case, based on the value of the one or more action evaluation metrics corresponding to each of the at least two motion data segments, or corresponding metric feature data of each of the at least two motion data segments, the value of the one or more action evaluation metrics corresponding to at least two fitness actions included in the group may be obtained. In an example, the metric feature data corresponding to the fitness action group may be determined based on the metric feature data corresponding to each of the fitness actions in a fitness action group, and then, a value of the one or more action evaluation metrics corresponding to the fitness action group is determined based on the calculated metric feature data. In another example, the value of the one or more action evaluation metrics corresponding to each fitness action in a fitness action group may be determined, and then, the value of the one or more action evaluation metrics corresponding to the fitness action group are obtained based on the value of the one or more action evaluation metrics corresponding to each of the fitness actions.

It should be understood that, the above-mentioned embodiments are described by taking the fitness action group including two or more fitness actions as an example, and in some other embodiments, the fitness action group may include a single fitness action, the quality evaluation of which is assessed similarly to the determination of the values of the action evaluation metrics for the single fitness action, which will not be repeated herein.

Optionally, an overall quality evaluation of the user's current fitness or exercise session may further be performed. For example, the action quality of all fitness actions of the user during the current exercise or fitness activity may be determined based on a plurality of motion data segments. e.g., the value of at least one action evaluation metric corresponding to the current fitness activity is determined.

In the foregoing embodiments, the values of the action evaluation metrics may be determined in various manners based on the metric feature data. In some embodiments, mapping processing may be performed on the metric feature data to obtain the values of the action evaluation metrics. In an example, the metric feature data may be mapped based on a pre-determined mapping function or mapping rule to obtain the values of the action evaluation metrics. In another example, at least one of the mapping function, mapping parameters, or the mapping rule is determined based on at least one of the data related to the current fitness activity, user attribute data, or the historical motion data of the user. The data related to the current fitness activity may include one or more of, for example, user input, the above-mentioned motion data, exercise time duration, exercise time point or period, physiological data collected during the exercise process or the like. The user attribute data may include one or more of, for example, user grouping, age, gender, BMI, fitness goals, fitness habits or the like.

In an example, a mapping function between the metric feature data of the action evaluation metrics and the values (e.g., scores) of the action evaluation metrics may be obtained in advance in the following manner.

First, during the fitness activities of one or more users, at least one sensor is utilized to collect the motion data of the user to obtain motion data. At the same time, a camera is utilized to shoot videos of the fitness activities to obtain a plurality of fitness videos.

Then, for the plurality of pieces of collected motion data, the metric feature data of an action evaluation metric corresponding to each piece of motion data is calculated respectively, and professional fitness coaches and/or ordinary users are requested to score the plurality of fitness videos for the action evaluation metric, so as to obtain the metric feature data and subjective scores corresponding to the same action evaluation metric for the same one or more fitness actions respectively, that is, the correspondence between the metric feature data of the action evaluation metric and the subjective score is obtained.

Then, based on the correspondence, function fitting is performed to obtain a mapping function between the metric feature data of the action evaluation metric and the value of the action evaluation metric.

In some embodiments, prior knowledge may be further taken into account in the determination of the mapping function or rule. In an example, the mapping function obtained in the above example may be adjusted based on prior knowledge, or function fitting may be performed based on prior knowledge. e.g., fine tuning of key points may be performed based on prior knowledge, and re-fitting is performed after some key points are raised or lowered. In another example, a score range is set to [0, 100], and a portion exceeding the score range is truncated to the boundary value. In another example, a piece-wise function is utilized as the mapping function, or a piece-wise function is utilized for fine-tuning, etc.

In some embodiments, feedback from users, coaches, or other subjects may be received, such as feedback received after the action quality evaluation result is output. Alternatively, subjective evaluation input from users, coaches, or other subjects may be received. In this case, the current mapping function or rule may be adjusted for iterative optimization based on at least one of the received feedback or subjective evaluation input.

After obtaining the mapping relationship between the metric feature data of the target action evaluation metric and its value based on the aforementioned method, the value of the target action evaluation metric may be determined based on the calculated metric feature data of the target action evaluation metric.

It should be noted that the foregoing description of determination of the mapping relationship between the metric feature data and the value of the action evaluation metric is merely illustrative. In practical applications, the mapping relationship between the metric feature data and the value of the target action evaluation metric may be obtained based on other fitting mechanisms, and the present disclosure is not limited thereto.

It should be understood that, in the embodiments of the present disclosure, the value of the action evaluation metric may be realized as a score or an indicator, such as “high”, “medium”, “low”, or “excellent”, “good”, “fair”, “poor”, or the like. The present application does not impose any limitation on the specific implementations of the value of the action evaluation metric.

In the embodiments of the present disclosure, through data reconstruction processing, a same algorithm or model may be used to determine the values of the action evaluation metrics without considering the specific exercise type to which the fitness actions belong, which has a high degree of universality, not only reduce the complexity of the overall procedure, but also avoid the influence of the accuracy of the algorithm for exercise type recognition on the action quality evaluation result, thereby contributing to improvement of the accuracy of the action quality evaluation result.

At 203, an action quality evaluation result of the user is determined according to the value of the at least one action evaluation metric.

As mentioned above, for a certain action evaluation metric, at least one of the following may be obtained: one or more values of the action evaluation metric for one or more fitness actions, one or more values of the action evaluation metric for a fitness action group, or one or more values of the action evaluation metric for the entire fitness activity.

In some embodiments, the action quality evaluation result may include at least one of: a value of each of the at least one action evaluation metric, a statistical result of the value of the at least one action evaluation metric, evaluative description of each of the at least one action evaluation metric, or a suggestion for the user to improve the actions. In an example, a value of the at least one action evaluation metric corresponding to the fitness activity may be obtained based on the value of the at least one action evaluation metric obtained at 202. In another example, the values of a certain action evaluation metric obtained at 202 may be statistically processed to obtain an average of the action evaluation metric corresponding to the fitness actions performed by the user during the fitness activity, or obtain a maximum or minimum value of the action evaluation metric and a position of the corresponding fitness action, or obtain a number, a proportion, or positions of fitness actions whose values of the action evaluation metric reach a preset threshold, or obtain a number, a proportion, or positions of fitness actions whose values of the action evaluation metric do not reach the preset threshold, etc. In addition, at least one of an evaluative description or an action improvement suggestion for one or more fitness actions performed by the user may be obtained based on the value of the at least one action evaluation metric. The action improvement suggestion may be, for example, the user performs one or more fitness actions with poor stability, or the user has relatively good strength control in performing one or more fitness actions, etc. In some other examples, the value of the action evaluation metric is compared with a preset threshold to determine whether a desired fitness effect is achieved or whether there is a risk of injury. In an example, if the values of the intensity variability of a plurality of fitness actions indicate that an attenuation degree of the plurality of fitness actions is less than a preset threshold, such as 10%, it is determined that the execution of the plurality of fitness actions does not achieve the desired fitness effect. In another example, if the values of the intensity variability of the plurality of fitness actions are higher than a preset threshold, such as 50%, it is determined that the user has a relatively high risk of injury. In this case, optionally, at least one of the evaluative description or action improvement suggestions may be output during the fitness session of the user, so as to facilitate the user to make adjustments during subsequent execution of fitness actions.

In some embodiments, the action quality evaluation result may be output. In some examples, the action quality evaluation result may be output via an intermediate device, or via a wearable device, or via other devices. In an example, the action quality evaluation result may be displayed on a visualization interface, such as on the screen of the wearable device or the intermediate device. In another example, the action quality evaluation result may be output via an audio output device. The specific implementations of the output device and the specific output manners are not limited herein.

In some embodiments, the exercise type data input by the user may be received. In this case, while outputting the action quality evaluation result, its corresponding exercise type data may also be output.

In some embodiments, the value of the at least one action evaluation metric may be processed based on an LLM (Large Language Model) to obtain the action quality evaluation result. In this case, the action quality evaluation result may include comprehensive information about the action quality of the user in terms of the at least one action evaluation metric.

In some examples, the value of the at least one action evaluation metric obtained above may be converted into a description in a preset data structure, optionally conforming to a preset data format such as a json schema format. Then, the converted description in the json schema format is input into the LLM, and the action quality evaluation result output by the LLM is obtained.

In some examples, the action quality evaluation result output by the LLM is similar to the action quality evaluation of various fitness actions performed by the user during the exercise process by a professional fitness coach (i.e., a human coach). For example, the action quality evaluation result may be a structured report such as: Your overall performance in the exercise activity is excellent and has an overall score of 85 (e.g., “Overall Performance: 85, Excellent”); for the barbell bench press, your overall action consistency score is 80, but the 6^th, 7^th, and 9^th actions are poorly performed with an action consistency, and the scores of these actions are lower than the standard (e.g., Action Consistency: 80, Note: the actions #6, #7 and #9 are poorly performed with a score fell below the standard); your overall action stability score is 90 with only the 7^th action is slightly worse (e.g., Action Stability: 90, Note: action #7 is slightly worse); your overall action continuity score is 70, and the 7th and 9th actions need to be improved (e.g., Action Continuity: 70, Note: actions #7 and #9 need to be improved); your action variability is well controlled and has a score of 90, but there is a slight decline in performing the 4th action (e.g., Action Variability: 90, Note: there is a slight decline in performing the action #7); for the barbell squat, your action consistency and action stability are well controlled, both of which have a score above 90 (e.g., Action Consistency: 92; Action Stability: 94); and for the bench press, attention should be given to improve the action continuity and continue to maintain good action stability.

It should be understood that the foregoing description of the action quality evaluation result output by the LLM is merely illustrative, and in intended to enable those skilled in the art to better understand the technical solutions of the embodiments of the present disclosure. In practical applications, the action quality evaluation result output by the LLM may be realized in other forms, such as in the form of a table or in the form of a radar chart as shown in FIG. 3, which is not limited in the present disclosure.

In the scheme for action quality evaluation provided in the embodiments of the present disclosure, the action quality evaluation result determined based on the value of one or more action evaluation metrics is utilized to characterize the action quality of the user, which is easier for the user to understand, and thus satisfy the user's needs.

The determination of the metric feature data or the value of each action evaluation metric for one or more fitness actions will be described in detail below with specific examples.

(1) The action consistency metric metric, is used for evaluating a consistency of at least one of the amplitudes or the completion times of a plurality of fitness actions performed by the user. The plurality of fitness actions may be multiple fitness actions included in a single motion data segment, multiple fitness actions included in a fitness action group, or multiple fitness actions included in the entire fitness activity, etc.

In some examples, the consistency of the amplitudes, i.e., an amplitude consistency metric, of the plurality of fitness actions may be determined in the following manner. Reconstructed displacement data corresponding to each of the plurality of fitness actions, such as the virtual primary-axis displacement data, is determined based on the reconstructed data corresponding to the each fitness action, and then the value of the amplitude consistency metric of the plurality of fitness actions is determined based on the reconstructed displacement data corresponding to each of the plurality of fitness actions.

In an example, statistical processing is performed on the reconstructed displacement data corresponding to each fitness action to obtain statistical data, and the metric feature data corresponding to the each fitness action may be obtained based on the statistical data.

For instance, at least one of the peak region data (e.g., the first few largest values) or the valley region data (e.g., the first few smallest values) in the virtual primary-axis displacement data corresponding to each fitness action may be determined, and at least one of the peak region data or the valley region data, such as 90% of the displacement peak, the average of the data above 90% of the displacement peak or the like, is taken as the metric feature data. In this case, the smaller the metric feature data is, the more consistent the action amplitude is when the user performs the plurality of fitness actions, and the higher the action quality is. In some other examples, at least one of the peak region data or the valley region data in the virtual primary-axis displacement data corresponding to each fitness action may be normalized, e.g., the normalization is performed by using the peak value (i.e., the maximum value) of the virtual primary-axis displacement data corresponding to the fitness action to obtain the metric feature data.

Then, the consistency of the amplitudes of the plurality of fitness actions may be obtained based on the metric feature data corresponding to each of the plurality of fitness actions. In some embodiments, the metric feature data corresponding to each of the plurality of fitness actions is normalized to obtain the normalized metric feature data, and then the consistency of the amplitudes of the plurality of fitness actions is obtained based on the normalized metric feature data. For example, a statistical value (e.g. the maximum value) of the metric feature data of the multiple fitness actions is utilized to normalize the metric feature data corresponding to each of the plurality of fitness actions.

In some other embodiments, after obtaining the metric feature data corresponding to each of the plurality of fitness actions, the metric feature data corresponding to each of the plurality of fitness actions is averaged to obtain the metric feature data corresponding to the plurality of fitness actions, and the consistency of the amplitudes of the plurality of fitness actions is obtained based on the metric feature data corresponding to the plurality of fitness actions.

In some examples, the consistency of the completion times of the plurality of fitness actions may be determined in the following manner. The action completion time corresponding to each of the plurality of fitness actions is determined based on a time duration of the motion data corresponding to the each fitness action, and then the consistency of the completion times (also referred to as rhythm) of the plurality of fitness actions is determined based on the action completion time corresponding to each of the plurality of fitness actions.

In an example, a variance, a standard deviation, or similar statistical quantities of the action completion times corresponding to the plurality of fitness actions is taken as the metric feature data of the consistency of the completion times. For instance, a difference between the action completion times corresponding to adjacent fitness actions of the plurality of fitness actions is calculated, and a standard deviation of at least two obtained differences is calculated as the metric feature data of the consistency of the completion times of the plurality of fitness actions.

In this case, the smaller the metric feature data is, the smaller the difference in the completion times of the plurality of fitness actions performed by the user is, that is, the better the consistency of the completion times is, and the higher the action quality is.

(2) The action stability metric, also referred to herein as decay, is used for evaluating the degree of jitter or wobble of one or more fitness actions performed by the user. The action stability metric may include at least one of displacement stability or angle stability. The displacement stability may be used for evaluating the degree of jitter of the displacement of at least a part of the user's body in one or more directions during the execution of a single fitness action, and angle stability is used for evaluating the degree of wobble or rotation of the angle of at least a part of the user's body in one or more directions during the execution of the fitness action.

In some embodiments, the virtual secondary-axis displacement data of the fitness action may be used as the basis for the metric feature data of displacement stability. In some examples, at least one of the peak region data or the valley region data of the virtual secondary-axis displacement data is determined, and the metric feature data of displacement stability is determined according to at least one of the peak region data or the valley region data of the virtual secondary-axis displacement data. For instance, a ratio of a displacement value corresponding to the second quantile to a displacement value corresponding to the third quantile of the virtual secondary-axis displacement data is calculated, and the ratio is taken as the metric feature data of displacement stability, where the second quantile may be 90% or a higher or lower value close to the peak, and the third quantile may be 10% or a higher or lower value close to the valley. In this case, the smaller the ratio is, the smaller the displacement jitter amplitude of the fitness action performed by the user is, the better the displacement stability is, and the higher the action quality is.

In some embodiments, the virtual primary-axis Euler angle data of the fitness action is taken as the basis for the metric feature data of angle stability. In some examples, at least one of the peak region data or the valley region data of the virtual primary-axis Euler angle data is determined, and the metric feature data of angle stability is determined according to at least one of the peak region data or the valley region data of the virtual primary-axis Euler angle data. For instance, a ratio of the Euler angles corresponding to the fourth quantile and the fifth quantile of the virtual primary-axis Euler angle data after being cosine converted is calculated, and the ratio is taken as the metric feature data of angle stability, where the fourth quantile is 85% or a higher or lower value close to the peak, and the fifth quantile is 15% or a higher or lower value close to the valley. In this case, the smaller the ratio is, the smaller the amplitude of wobble or rotation of the fitness action performed by the user is, the better the action stability metric is, and the higher the action quality is.

In some embodiments, fused metric feature data is further obtained based on the metric feature data of angle stability and the metric feature data of displacement stability. For example, the metric feature data of angle stability and the metric feature data of displacement stability are summed with weights to obtain the fused metric feature data. In this case, the metric feature data of action stability metric may include at least one of the metric feature data of angle stability, the metric feature data of displacement stability, or the fused metric feature data.

In some embodiments, after obtaining the metric feature data of action stability metric corresponding to each of the plurality of fitness actions, the metric feature data of action stability metric corresponding to the plurality of fitness actions is statistically processed, such as weighted averaging, averaging, or normalization or the like, to obtain the metric feature data of action stability metric corresponding to the plurality of fitness actions, and then the action stability metric corresponding to the plurality of fitness actions is obtained. In some other embodiments, the action stability metric corresponding to each of the plurality of fitness actions is obtained based on the metric feature data of action stability metric corresponding to the each fitness action, and the action stability metric corresponding to the plurality of fitness actions is obtained based on the action stability metric corresponding to each of the plurality of fitness actions, which is not limited in the present disclosure.

(3) The action continuity metric, is used for evaluating the continuity of one or more fitness actions performed by the user.

In some embodiments, the metric feature data of the action continuity metric corresponding to a fitness action is obtained based on the virtual primary-axis acceleration data and/or velocity corresponding to the fitness action. In some examples, a percentage of a time duration, during which the velocity value in the virtual primary-axis velocity data of the fitness action is less than or equal to a preset velocity threshold, to a total time duration is calculated, or a percentage of a time duration during which the acceleration value in the virtual primary-axis acceleration data is less than or equal to a preset acceleration threshold to the total duration is calculated, and the obtained percentage is taken as the aforementioned metric feature data. In this case, the smaller the percentage is, the fewer sudden or unnecessary pauses occur during the fitness action performed by the user, the more coherent the action is, and the higher the action quality is.

In some embodiments, after obtaining the metric feature data of the action continuity metric corresponding to each of the plurality of fitness actions, statistical processing such as weighted averaging, averaging, or normalization may be performed on the metric feature data of the action continuity metric corresponding to each of the plurality of fitness actions to obtain the metric feature data of the action continuity metric corresponding to the plurality of fitness actions, and then the action continuity metric corresponding to the plurality of fitness actions is obtained. In some other embodiments, the action continuity metric corresponding to each of the plurality of fitness actions is obtained based on the metric feature data of the action continuity metric corresponding to the each fitness action, and the action continuity metric corresponding to the plurality of fitness actions is obtained based on the action continuity metric corresponding to each of the plurality of fitness actions, which is not limited in the present disclosure.

(4) The action variability metric, is used for evaluating the intensity change of a plurality of fitness actions performed by the user, and is used to determine whether the user reaches a certain degree of exhaustion and/or whether there is a risk of injury during the execution of the plurality of fitness actions.

In strength training, the training effect of multiple fitness actions is usually measured by determining whether the trainer reaches a certain degree of exhaustion. Therefore, the training effect of the trainer is measured by evaluating the attenuation of the velocity or acceleration corresponding to the multiple fitness actions.

In some embodiments, the metric feature data of the action variability metric corresponding to a fitness action is obtained based on the virtual primary-axis velocity data and/or acceleration. In some examples, statistical processing such as averaging is performed on the virtual primary-axis velocity data and/or acceleration corresponding to the fitness action to obtain the statistical value of the virtual primary-axis of velocity and/or acceleration corresponding to the fitness action. Then, the statistical values of the virtual primary-axis velocity data and/or acceleration of the first M fitness actions and the last N fitness actions of the multiple fitness actions are calculated to obtain the metric feature data of the action variability metric corresponding to the multiple fitness actions, where both M and N are integers greater than or equal to 1, which may be preset or determined according to the number of the multiple fitness actions, e.g., M and N are 20% or more or less of the number of the multiple fitness actions. In an example, a difference or a proportion of change between the statistical values of the virtual primary-axis of velocity and/or acceleration of the first M fitness actions and the last N fitness actions is taken as the metric feature data of the action variability metric corresponding to the multiple fitness actions. In another example, after obtaining the acceleration-based metric feature data and the velocity-based metric feature data respectively, the acceleration-based metric feature data and the velocity-based metric feature data are fused, such as by weighted summation or weighted averaging, to obtain the fused metric feature data, and the fused metric feature data is taken as the metric feature data of the action variability metric corresponding to the multiple fitness actions.

(5) The force control metric, is used for evaluating the strength control condition of one or more fitness actions performed by the user.

The force control metric metric is used for evaluating whether there is a situation in which the user completes the action with the help of inertia or elasticity in a non-essential way, or a situation in which there is an unnecessary sudden exertion of force during the execution of the fitness action, which may be determined by evaluating whether there is a sharp change in the force applied by the user during the execution of the fitness action.

In some embodiments, the metric feature data of the force control metric corresponding to a fitness action is obtained based on the virtual secondary-axis acceleration data corresponding to the fitness action. In some examples, one or more types of statistical processing such as averaging and normalization is performed on the virtual secondary-axis acceleration data corresponding to the fitness action to obtain the metric feature data of the force control metric corresponding to the fitness action. It can be understood that, the smaller the metric feature data is, the fewer the user completes the action with the help of inertia or elasticity when performing the fitness action, and the fewer the force changes drastically when the user performs the fitness action, the better the force control metric is, and the higher the action quality is.

In some embodiments, after obtaining the metric feature data of the force control metric corresponding to each of the plurality of fitness actions, statistical processing such as weighted averaging, averaging, or normalization or the like is performed on the metric feature data of the force control metric corresponding to each of the multiple fitness actions to obtain the metric feature data of the force control metric corresponding to the multiple fitness actions, and then the force control metric corresponding to the multiple fitness actions is obtained. In some other embodiments, the force control metric corresponding to each of the plurality of fitness actions is obtained based on the metric feature data of the force control metric corresponding to the each fitness action, and the force control metric corresponding to the multiple fitness actions is obtained based on the force control metric corresponding to each of the plurality of fitness actions, which is not limited in the present disclosure.

(6) The action normality metric, is used for evaluating the normality or standardization degree of one or more fitness actions performed by the user.

In some embodiments, the metric feature data of the action normality metric corresponding to a fitness action is obtained based on the virtual primary-axis acceleration data of the fitness action. In some examples, the virtual primary-axis acceleration data of the fitness action is compared with preset virtual primary-axis acceleration reference data. e.g., a similarity between two pieces of data is calculated using the DTW (dynamic time warping) algorithm or the like, and the similarity is taken as the metric feature data of the action normality metric. The DTW algorithm does not require the two pieces of data to have the same length and can avoid the influence of a difference in action period on the similarity. In some examples, a similarity or difference between the virtual primary-axis acceleration data of a certain fitness action and the virtual primary-axis acceleration data of at least one adjacent fitness action may be determined, and the metric feature data of the action normality metric corresponding to the fitness action is obtained according to the similarity or difference.

In this case, the larger the metric feature data is, the higher the similarity between the fitness action performed by the user and the reference action is, the better the action normality is, and the higher the action quality is.

In some embodiments, after obtaining the metric feature data of the action normality metric corresponding to each of the plurality of fitness actions, statistical processing such as weighted averaging, averaging, or normalization is performed on the metric feature data of the action normality metric corresponding to each of the multiple fitness actions to obtain the metric feature data of the action normality metric corresponding to the multiple fitness actions, and then the value of the action normality metric corresponding to the multiple fitness actions is obtained. In some other embodiments, the value of the action normality metric corresponding to each of the plurality of fitness actions is obtained based on the metric feature data of the action normality metric corresponding to the each fitness action, and the value of the action normality metric corresponding to the multiple fitness actions is obtained based on the value of the action normality metric corresponding to each of the plurality of fitness actions, which is not limited in the present disclosure.

(7) The action balance metric, is used for evaluating the left-right balance degree of one or more fitness actions performed by the user. The metric is applicable to symmetric fitness actions performed by the user. In this case, the above-mentioned motion data is obtained from the sensor data collected by a set of sensors located at symmetric or nearly symmetric parts of the user's body, such as the left and right wrists, left and right ankles, left and right legs, left and right ears, left and right eyes of the user, etc.

In some embodiments, the value of the action balance metric is obtained by comparing a similarity or difference between motion data corresponding to the fitness action that is obtained from the sensors on the left side and right side respectively. In some examples, the metric feature data of the action balance metric is obtained based on the virtual primary-axis velocity data. For instance, the difference between the virtual primary-axis velocity data based on the left-side sensor and the virtual primary-axis velocity data based on the right-side sensor corresponding to the fitness action is determined. The metric feature data of the action balance metric corresponding to this fitness action is obtained based on an average of the differences between the virtual primary-axis velocity data based on the left-side sensor and the virtual primary-axis velocity data based on the right-side sensor, or based on a difference between the average of the virtual primary-axis velocity data based on the left-side sensor and the average of the virtual primary-axis velocity data based on the right-side sensor, or a ratio of change in the above-mentioned data from the sensors on the left and right sides.

It can be understood that, the smaller the metric feature data is, the better the left-right balance is when the user performs the fitness action, and further, the better the action balance metric is, and the higher the action quality is.

In some embodiments, after obtaining the metric feature data of the action balance metric corresponding to each of the plurality of fitness actions, statistical processing such as weighted averaging, averaging, or normalization or the like is performed on the metric feature data of the action balance metric corresponding to the each of the multiple fitness actions to obtain the metric feature data of the action balance metric corresponding to the multiple fitness actions, and then the value of the action balance metric corresponding to the multiple fitness actions is obtained. In some other embodiments, the value of the action balance metric corresponding to each of the plurality of fitness actions is obtained based on the metric feature data of the action balance metric corresponding to the each fitness action, and the value of the action balance metric corresponding to the plurality of fitness actions is obtained based on the value of the action balance metric corresponding to each of the plurality of fitness actions, such as by averaging, weighted averaging or the like, which is not limited in the present disclosure.

The foregoing is described by taking the virtual primary-axis velocity data as an example. In some embodiments, the metric feature data of the action balance metric corresponding to the fitness action may be determined based on at least one of the virtual primary-axis acceleration data, the virtual primary-axis displacement data, or the virtual primary-axis Euler angle data, which will not be repeated herein.

Optionally, during extraction of the metric feature data of each of the above-mentioned action evaluation metrics, relative values, statistical values, or normalized values may be used, which can effectively avoid the influence caused by the differences between different types of fitness actions and/or the differences between different individuals in performing the same fitness actions, thereby improving the accuracy and applicability of the action quality evaluation result.

FIG. 4 is another flowchart of the method for action quality evaluation in some example embodiments of the present disclosure. As shown in FIG. 4, the method in the example embodiments may include 401 to 403.

At 401, motion data collected during the fitness activity of the user is acquired.

The motion data may include at least one of acceleration data, velocity data, displacement data, or Euler angle data.

At 402, at least one motion data segment is obtained based on the motion data, where each motion data segment includes motion data corresponding to at least one fitness action.

In some embodiments, the motion data is segmented to obtain multiple motion data segments, where each motion data segment may correspond to one or more fitness actions. For example, the motion data of two fitness actions performed by the user is segmented into a motion data segment corresponding to fitness action/and a motion data segment corresponding to fitness action 2.

At 403, data reconstruction processing is performed on each of the at least one motion data segment to obtain at least one reconstructed data segment.

In some embodiments, data reconstruction processing is performed on each of the multiple motion data segments in the following manner: for each motion data segment, the data in the motion data segment is decomposed using an SVD algorithm to obtain multi-axis processed data. For example, singular value decomposition is performed on a three-axis motion signal A(t×3) to obtain U(t×3), Σ(3), and V(3×3), where U(t×3) is an orthogonal matrix, Σ(3) is a diagonal matrix with elements on the diagonal being singular values and optionally ordered from largest to smallest, and V(3×3) is an orthogonal matrix. Then, data reconstruction processing is performed on the obtained multi-axis processed data to obtain the reconstructed data segment. For example, the obtained singular values are processed to obtain the virtual principal axis signal and the virtual principal axis signal, where at least one larger singular value is used to construct the virtual principal axis signal, and at least one smaller singular value is used to construct the virtual secondary axis signal. Continuing the above example, the first two features in Σ (i.e., the signals of the first two orthogonal axes) are reconstructed to obtain a virtual primary-axis signal, and the last feature is reconstructed to obtain a virtual secondary-axis signal. It is assumed that the virtual primary-axis signal is Axis_m, which is obtained by the following formula:

Axis m = U 2 ⁢ ∑ 2 ⁢ V 2 t .

It is assumed that the virtual auxiliary-axis signal is Axis_i, it is obtained by the following formula:

Axis i = U 1 ⁢ ∑ 1 ⁢ V 1 t .

Optionally, the data reconstruction processing may be performed in other ways,
which is not limited herein.

In some embodiments, the obtained reconstructed data segment may include at least one of virtual primary-axis data or virtual secondary-axis (also referred to as virtual auxiliary axis) data. In practical applications, data reconstruction is performed on data of at least two axes in the multi-axis data obtained from data decomposition which have strong signal responses, to obtain virtual primary-axis data, and data reconstruction is performed on data of at least one axis in the multi-axis data with weak signal responses to obtain virtual secondary-axis data. In this case, the obtained virtual primary-axis data is used to characterize the signal response to the primary motion, and the virtual secondary-axis data is used to characterize the signal response to sources other than the primary motion.

At 404, an action quality evaluation result of the user is obtained based on the at least one reconstructed data segment.

It can be seen from the calculation process of the metric feature data of each action evaluation metric described above that, the value of at least one action evaluation metric for one or more fitness actions performed by the user may be determined based on the virtual primary-axis data in the reconstructed data segment, where the at least one action evaluation metric includes at least one of the action consistency metric, the action stability metric, the action continuity metric, the action variability metric, the action normality metric, or the action balance metric.

Further, the value of at least one action evaluation metric for one or more fitness actions performed by the user may be determined based on the virtual secondary-axis data in the reconstructed data segment, where the at least one action evaluation metric includes at least one of the action stability metric or the force control metric.

For the specific implementations of the method for action quality evaluation, reference may be made to any of the foregoing embodiments, which will not be repeated herein.

In the method for action quality evaluation provided by the embodiments of the present disclosure, after the motion data collected during the user's fitness process is acquired, the motion data is segmented to obtain at least one motion data segment, and then, data reconstruction processing is performed on each motion data segment to obtain at least one reconstructed data segment. Finally, the action quality evaluation result of the user is obtained based on the at least one reconstructed data segment. In the present solution, by performing reconstruction processing on the motion data segment to obtain the reconstructed data segment, the influence of different motion types can be reduced or eliminated, so that the determination of the value of each action evaluation metric does not depend on the action type. That is, for different types of fitness actions, the action quality evaluation result may be determined based on the same algorithm or model, which has a high degree of universality.

FIG. 5 is a schematic structural diagram of an apparatus for action quality evaluation shown in an example embodiment of the present disclosure. As shown in FIG. 5, the action quality evaluation device may include the following modules.

A motion data acquisition module 51, configured to acquire the motion data collected during the user's fitness activity.

A metric value determination module 52, configured to determine the value of at least one action evaluation metric of the user based on the motion data.

A first assessment result determination module 53, configured to determine the action quality evaluation result of the user according to the value of the at least one action evaluation metric.

The at least one action evaluation metric includes at least one of:

• • an action stability metric, for evaluating a degree of jitter or wobble of one or more fitness actions performed by the user;
  • an action continuity metric, for evaluating a continuity degree of one or more fitness actions performed by the user;
  • an action variability metric, for evaluating an intensity variability of multiple fitness actions performed by the user; or
  • a force control metric, for evaluating a force exertion condition during one or more fitness actions performed by the user.

Optionally, when the metric value determination module 52 is configured to determine the values of at least one action evaluation metric of the user based on the motion data, it includes: determining the metric feature data corresponding to each fitness action among at least one fitness action performed by the user based on the motion data, and obtaining the values of at least one action evaluation metric of the user according to the metric feature data corresponding to each fitness action among the at least one fitness action.

Optionally, the determination of the metric feature data corresponding to each of the at least one fitness action does not depend on the action type of the each fitness action.

Optionally, the determination of the metric feature data corresponding to each of the at least one fitness action includes data reconstruction processing of the motion data, and the data reconstruction processing is used to reduce or eliminate the influence of the action type.

Optionally, the metric feature data corresponding to each of the at least one fitness action is obtained by performing singular value decomposition processing on the motion data.

Optionally, when the metric value determination module 52 is configured to determine the value of at least one action evaluation metric of the user based on the motion data, it includes: obtaining at least one motion data segment according to the motion data, each motion data segment includes motion data corresponding to at least one fitness action; performing data reconstruction processing on each of the at least one motion data segment to obtain at least one reconstructed data segment; and obtaining the value of the at least one action evaluation metric of the user according to the at least one reconstructed data segment.

Optionally, the reconstructed data segment includes at least one of virtual primary-axis data or virtual secondary-axis data, where the virtual primary-axis data is used to characterize the signal response to the primary motion, and the virtual secondary-axis data is used to characterize the signal response to sources other than the primary motion.

Optionally, the action quality evaluation result of the user includes at least one of: a statistical result of the value of the at least one action evaluation metric of the user; evaluative description of the at least one action evaluation metric of the user; or an action improvement suggestion for the user.

FIG. 6 is a schematic structural diagram of another apparatus for action quality evaluation in an example embodiment of the present disclosure. As shown in FIG. 6, the apparatus for action quality evaluation may include the following modules.

A motion data acquisition module 51, is configured to acquire the motion data collected during the user's fitness process.

A motion data segment determination module 62, is configured to obtain at least one motion data segment based on the motion data, where each motion data segment includes motion data corresponding to at least one fitness action.

A reconstructed data segment determination module 63, is configured to perform data reconstruction processing on each of the at least one motion data segment to obtain at least one reconstructed data segment.

A second assessment result determination module 64, is configured to obtain the action quality evaluation result of the user based on the at least one reconstructed data segment.

Optionally, the data reconstruction processing is used to reduce or eliminate the influence of the action type.

Optionally, the data reconstruction processing includes singular value decomposition processing.

Optionally, the reconstructed data segment includes at least one of virtual primary-axis data or virtual secondary-axis data, where the virtual primary-axis data is used to characterize the signal response to the primary motion, and the virtual secondary-axis data is used to characterize the signal response to sources other than the primary motion.

Optionally, the second assessment result determination module 64 is configured to determine the value of at least one action evaluation metric of one or more fitness actions performed by the user according to the virtual primary-axis data in the reconstructed data segment, where the at least one action evaluation metric includes at least one of the action consistency metric, the action stability metric, the action continuity metric, the action variability metric, or the action normality metric.

Optionally, the second assessment result determination module 64 is configured to determine the value of at least one action evaluation metric of one or more fitness actions performed by the user according to the virtual secondary-axis data in the reconstructed data segment, where the at least one action evaluation metric includes at least one of the action stability metric or the force control metric.

The realization of the functions and roles of each module in the above-mentioned apparatus is described in detail in the implementations of corresponding processes in the above-mentioned method, which will not be repeated herein.

For the embodiments of the apparatus, since they basically correspond to the embodiments of the method, reference can be made to the descriptions in the method embodiments.

In some embodiments, the present disclosure further provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, any of the above-mentioned methods are implemented.

In some embodiments, the present disclosure further provides a computer program product, including a computer program or instructions. When the computer program or instructions is executed by a processor, the methods described in any of the above-mentioned embodiments are implemented.

As shown in FIG. 7, FIG. 7 is a structural diagram of an electronic device in an example embodiment of the present disclosure. The device 700 may be a smart phone or mobile phone, a tablet, a laptop, a desktop computer, a wearable device (e.g., a watch, glasses, gloves, headgear (e.g., a hat, a helmet, a virtual reality headset, an augmented reality headset, a head-mounted device (HMD), a headband), a pendant, an armband, a leg ring, shoes, a vest), a server, etc.

Referring to FIG. 7, the device 700 may include one or more of the following components: a processing component 702, a memory 704, a power supply component 706, a multimedia component 708, an audio component 710, an input/output (I/O) interface 712, a sensor component 714, and a communication component 716.

The processing component 702 generally controls the overall operation of the device 700, such as operations associated with display, phone calls, data communication, camera operations, and recording operations. The processing component 702 may include one or more processors 720 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 702 may include one or more modules to facilitate the interaction between the processing component 702 and other components. For example, the processing component 702 may include a multimedia module to facilitate the interaction between the multimedia component 708 and the processing component 702.

The memory 704 is configured to store various types of data to support the operation at the device 700. Examples of such data include instructions for any application or method operating on the device 700, contact data, phone book data, messages, pictures, videos, etc. The memory 704 may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random-access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, a magnetic disk, or an optical disk.

The power supply component 706 provides power to various components of the device 700. The power supply component 706 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the device 700.

The multimedia component 708 includes a screen that provides an output interface between the device 700 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of touch or swipe actions, but also the duration and pressure associated with the touch or swipe operations. In some embodiments, the multimedia component 708 includes a front camera and/or a rear camera. When the device 700 is in an operating mode, such as a shooting mode or a video mode, the front camera and/or the rear camera may receive external multimedia data. Each of the front camera and the rear camera may be a fixed optical lens system or have a focal length and optical zoom capabilities.

The audio component 710 is configured to output and/or input audio signals. For example, the audio component 710 includes a microphone (MIC). When the device 700 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is configured to receive external audio signals. The received audio signals may be further stored in the memory 704 or sent via the communication component 715. In some embodiments, the audio component 710 further includes a speaker for outputting audio signals.

The I/O interface 712 provides an interface between the processing component 702 and peripheral interface modules, which may be a keyboard, a click wheel, buttons, etc. These buttons may include, but are not limited to, a home button, a volume button, a start button, and a lock button.

The sensor component 714 includes one or more sensors for providing various aspects of state assessment for the device 700. For example, the sensor component 714 may detect the on/off state of the device 700, the relative positioning of components, such as the display and the keypad of the device 700. The sensor component 714 may also detect the change in position of the device 700 or a component in the device 700, the presence or absence of user contact with the device 700, the orientation or acceleration/deceleration of the device 700, and the change in temperature of the device 700. The sensor component 714 may include a proximity sensor, configured to detect the presence of nearby objects without any physical contact. The sensor component 714 may further include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor component 714 may further include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

The communication component 716 is configured to facilitate wired or wireless communication between the device 700 and other devices. The device 700 may access a wireless network based on a communication standard, such as WiFi, 2G, 5G, or 4G, or a combination thereof. In an example embodiment, the communication component 716 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In an example embodiment, the communication component 716 further includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio-frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, or other technologies.

In an example embodiment, the device 700 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components for performing the above-mentioned method.

In an example embodiment, a non-transitory computer-readable storage medium including instructions is further provided, such as the memory 704 including instructions. The above-mentioned instructions may be executed by the processor 720 of the device 700 to complete the above-mentioned method. For example, the non-temporary computer-readable storage medium may be a ROM, a random-access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, or an optical data storage device.

Those skilled in the art will readily conceive other embodiments of the present disclosure after considering the specification and practicing the embodiments disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or conventional technical means in the technical field not disclosed in the present disclosure. The specification and examples are to be regarded as example only, and the scope and spirit of the present disclosure are indicated by the following claims.

It should be understood that the present disclosure is not limited to the precise structures that are described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present disclosure is limited by the appended claims.

The above are merely example embodiments of the present disclosure and are not intended to limit the present disclosure. Any modifications, equivalent replacements, improvements or the like made within the spirit and principle of the present disclosure shall be included in the scope of protection of the present disclosure.