A cross-scenario sports physical fitness pattern recognition monitoring method and system

Abstract

The application relates to the technical field of sports training monitoring, and provides a cross-scene sports physical fitness mode recognition monitoring method and system. The motion posture data, physiological index data and environment situation data of a target object are monitored and acquired in real time, so that the motion state, physiological load and environment of the athlete can be comprehensively perceived; the motion base element is recognized according to the motion posture data and a motion base element sequence is generated; the predetermined motion type of the target object is judged by using the motion base element sequence, the physiological index data and the environment situation data; the motion mode is closely associated with the environment characteristics and the physiological state, so that accurate judgment can be made according to the internal consistency of the motion and the physiology even when the environment signal is fuzzy. According to the judged predetermined motion type, the motion base element sequence and the physiological index data, the physical fitness consumption of the target object is evaluated, and the accurate quantification of the physical fitness load of the athlete is realized.

Description

Technical Field

[0001] This application relates to the field of sports training monitoring technology, and more specifically, to a cross-scenario sports fitness pattern recognition and monitoring method and system. Background Technology

[0002] In modern sports training, real-time monitoring of athletes' physical condition is crucial for optimizing training and preventing injuries. Sports fitness pattern recognition and monitoring systems, by integrating inertial measurement units, heart rate sensors, and other devices, can identify movement patterns and comprehensively assess workload. However, in complex and ever-changing training environments, especially during transitions between different scenarios, this system faces significant technical challenges.

[0003] Take a comprehensive training facility including a multi-purpose training corridor as an example. This corridor connects indoor and outdoor training areas, employing a semi-open design with partial roofs and walls, and containing metal structures. This environment causes GPS signals to become severely unstable due to obstruction, attenuation, and multipath effects, while indoor Wi-Fi signals are also weakly diffused, resulting in a persistently "fuzzy" state of environmental characteristics. Under these circumstances, the "scene method selection unit" within the system struggles to clearly determine whether the current scene is indoor or outdoor, potentially switching frequently between two analysis methods. This frequent switching severely consumes the device's computing and memory resources, leading to high processor load and increased latency in processing real-time data. Simultaneously, this area is intended for complex combination training such as sprinting, changing direction around cones, and jump cushioning, which inherently generate variable and non-standard posture and physiological data. The system, already under resource constraints and response delays, must process this complex data, making it unable to consistently call appropriate analysis methods. Consequently, the identification of athlete energy expenditure and movement standards suffers from significant deviations and delays, rendering real-time feedback ineffective and potentially misleading the training process.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] This application discloses a cross-scene sports physical fitness pattern recognition and monitoring method and system, aiming to solve the problems of existing sports physical fitness pattern recognition and monitoring systems in complex and ever-changing training scenarios, especially in transitional areas between scenarios, where unstable and ambiguous environmental signals lead to difficulties in scene recognition, excessive system resource consumption, response delays, and inaccurate recognition of athletes' physical exertion and movement standardization. The technical solution of this application is as follows:

[0006] In a first aspect, this application discloses a cross-scenario sports physical fitness pattern recognition and monitoring method, which includes:

[0007] Real-time monitoring and acquisition of motion posture data, physiological index data, and environmental context data of the target object;

[0008] Based on motion posture data, identify motion primitives in the motion process of the target object and generate a sequence of motion primitives; the motion primitives include at least one of the following: take-off primitive, air posture primitive, single-foot landing primitive, change-of-direction step primitive, and acceleration step primitive.

[0009] Based on the action primitive sequence, physiological index data, and environmental context data, determine the predetermined action type of the target object;

[0010] Assess the physical exertion of the target subject based on the predetermined movement type, movement primitive sequence, and physiological index data.

[0011] Furthermore, based on the above methods, the predetermined action types include ramp acceleration sprint, continuous change of direction around cones, and jump and landing cushioning actions; physiological indicator data includes heart rate; and environmental context data includes GPS signal and Wi-Fi signal.

[0012] More specifically, in some implementation schemes, the predetermined action type of the target object is determined based on the action primitive sequence, physiological index data, and environmental context data. This includes: if a predetermined number of consecutive acceleration step primitives are identified within a predetermined first time window, and the heart rate shows a rapid upward trend while the dynamic characteristics of the GPS signal in the environmental context data match a predetermined fluctuation pattern, then it is determined to be a ramp acceleration sprint action; if a predetermined number of consecutive and alternating direction change step primitives are identified within a predetermined second time window, and the dynamic characteristics of the Wi-Fi signal in the environmental context data show that it is in a weak diffusion range, then it is determined to be a continuous direction change slalom action; if a subsequence of take-off primitives, air posture primitives, and single-leg landing primitives is identified in the action primitive sequence, and the subsequence is completed within a predetermined third time window, while the heart rate is higher than a predetermined heart rate threshold, and the dynamic characteristics of the GPS signal and the Wi-Fi signal in the environmental context data meet a predetermined signal characteristic pattern, then it is determined to be a jump and landing buffer action.

[0013] Based on this, the dynamic characteristics of GPS signals in the environmental context data are consistent with the preset fluctuation pattern, including: the signal strength variance of the GPS signal is greater than the preset variance threshold, and the positioning loss frequency is higher than the preset frequency threshold; the dynamic characteristics of Wi-Fi signals in the environmental context data show that they are in a weak diffusion range, including: the received signal strength indication values ​​of Wi-Fi signals from multiple known access points are all lower than the preset strength threshold, and their variance is higher than the preset variance threshold; the dynamic characteristics of GPS signals and Wi-Fi signals in the environmental context data satisfy the preset signal characteristic pattern, including: the signal strength variance of the GPS signal is greater than the preset variance threshold, and the positioning loss frequency is higher than the preset frequency threshold; at the same time, the received signal strength indication values ​​of Wi-Fi signals from multiple known access points are all lower than the preset strength threshold, and their variance is higher than the preset variance threshold.

[0014] Furthermore, based on the predetermined movement type, movement primitive sequence, and physiological index data, the physical exertion of the target subject is assessed, including: if the predetermined movement type is ramp acceleration sprint, the physical exertion of the target subject is determined based on the average acceleration amplitude, duration, and the difference between the average heart rate and the resting heart rate during the ramp acceleration sprint; if the predetermined movement type is continuous slalom, the physical exertion of the target subject is determined based on the number of directional step primitives, the amplitude of angular velocity change, and the slope of heart rate change included in the continuous slalom; if the predetermined movement type is jump and landing cushioning, the physical exertion of the target subject is determined based on the peak vertical acceleration of the landing impact during the jump and landing cushioning movement, the peak heart rate during the jump, and the resting heart rate.

[0015] Based on the above, according to the motion posture data, the motion primitives in the motion process of the target object are identified, and a motion primitive sequence is generated, including: in response to the identification of the jump primitive, a landing expectation state is activated and a landing expectation time window is started; within the landing expectation time window, the vertical acceleration of the inertial measurement unit is monitored; when the instantaneous value of the vertical acceleration is lower than a preset negative impact threshold, or the instantaneous rate of change of the vertical acceleration is lower than a preset negative rate of change threshold, a potential landing impact signal is determined to have occurred.

[0016] As a technological improvement, based on motion posture data, the motion primitives of the target object during its motion process are identified and generated. This includes: if multiple potential landing impact signals are detected within the expected landing time window, the earliest potential landing impact signal with the largest impact amplitude is identified as the starting signal of the single-foot landing primitive; after identifying the starting signal, a landing stabilization confirmation window is initiated; if the vertical acceleration recovers to the static range within the landing stabilization confirmation window, and the angular velocities of each axis are all lower than the preset angular velocity stabilization threshold, then the single-foot landing primitive is finally confirmed.

[0017] As a further improvement, in response to the identification of the jump primitive, the following steps are taken: monitoring the vertical acceleration Az_t at the current moment and the vertical acceleration Az_t-1 at the previous moment; when Az_t is greater than a first positive threshold and Az_t-1 is less than a second positive threshold, and the average value of the vertical acceleration is between a first negative threshold and a third positive threshold within the subsequent first preset time period, it is determined that the jump primitive has been identified.

[0018] In one embodiment, the target object includes a wearable monitoring device; the wearable monitoring device includes a sensor array for real-time monitoring and acquisition of the target object's motion posture data, physiological index data, and environmental context data; the sensor array includes at least an inertial measurement sensor, a heart rate sensor, a GPS module, and a Wi-Fi module.

[0019] Secondly, this application also discloses a cross-scenario sports physical fitness pattern recognition and monitoring system, which includes: a monitoring module for real-time monitoring and acquisition of motion posture data, physiological index data, and environmental context data of a target object; a recognition and generation module for recognizing motion primitives in the motion process of the target object based on the motion posture data and generating a sequence of motion primitives; the motion primitives include at least one of take-off primitives, aerial posture primitives, single-leg landing primitives, change-of-direction step primitives, and acceleration step primitives; a judgment module for judging the predetermined motion type of the target object based on the motion primitive sequence, physiological index data, and environmental context data; and a physical fitness assessment module for assessing the physical fitness consumption of the target object based on the predetermined motion type, motion primitive sequence, and physiological index data.

[0020] Beneficial effects

[0021] This application provides a cross-scenario sports fitness pattern recognition and monitoring method. By real-time monitoring and acquisition of the target object's motion posture data, physiological index data, and environmental context data, it can comprehensively perceive the athlete's motion state, physiological load, and surrounding environment. Based on this, this application identifies motion primitives from the motion posture data and generates motion primitive sequences. These motion primitives (such as take-off primitives, air posture primitives, single-leg landing primitives, change-of-direction step primitives, and acceleration step primitives) can finely describe the athlete's motion details, overcoming the limitations of traditional methods in recognizing complex movements. Furthermore, this application comprehensively utilizes motion primitive sequences, physiological index data, and environmental context data to determine the target object's predetermined movement type. This multi-dimensional fusion judgment mechanism effectively solves the problems of scene recognition difficulties and frequent method switching caused by the instability of a single environmental signal in ambiguous scenarios such as "compound training corridors" in existing technologies. By closely linking motion patterns with environmental features and physiological states, even when environmental signals are ambiguous, accurate judgments can be made based on the inherent consistency between motion and physiology, avoiding excessive consumption of system resources and response delays. Finally, based on the determined predetermined movement type, movement primitive sequence, and physiological indicator data, this application assesses the physical exertion of the target subject, achieving precise quantification of the athlete's physical load. In summary, this application, through innovative multi-source data fusion and intelligent judgment mechanisms, effectively solves the technical challenges faced by existing sports fitness monitoring systems in cross-scenario applications, such as inaccurate identification, high resource consumption, and slow response. It significantly improves the accuracy, real-time performance, and robustness of fitness monitoring, providing more reliable guidance for optimizing training programs and preventing sports injuries. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating the steps of the cross-scenario sports physical fitness pattern recognition and monitoring method disclosed in an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of the cross-scenario sports physical fitness pattern recognition and monitoring system disclosed in an embodiment of the present invention. Detailed Implementation

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which these embodiments belong; the terminology used herein and in the specification of the application is for the purpose of describing particular embodiments only and is not intended to limit these embodiments; the terms "comprising" and "having," and any variations thereof, in the specification of these embodiments and the foregoing drawings, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification of these embodiments and the foregoing drawings are used to distinguish different objects, not to describe a particular order.

[0026] The implementation details of the technical solution in this embodiment are described in detail below:

[0027] Traditional sports fitness pattern recognition and monitoring systems face significant technical challenges when applied across different scenarios, especially in complex and variable training environments and areas with unstable signals during transitions. For example, in semi-open "composite training corridors," the continuous instability and ambiguity of GPS and Wi-Fi signals make it difficult for the system to accurately determine the current training scenario, leading to frequent switching of scenario analysis methods. This not only puts enormous pressure on the computing and memory resources of the monitoring equipment but also significantly increases the system's response latency in processing real-time sensor data. Furthermore, combined movements performed in such complex scenarios, such as ramp acceleration sprints, continuous changes of direction around cones, or jumps and landing cushioning, generate complex and variable posture data and physiological indicators that are difficult to standardize. This prevents the system from promptly and stably calling upon and running the most suitable analysis method, ultimately resulting in serious deviations and significant delays in recognizing athletes' energy expenditure and movement accuracy, rendering real-time feedback ineffective.

[0028] In response to this, firstly, this application proposes a cross-scenario sports fitness pattern recognition and monitoring method, such as... Figure 1 As shown, the method includes:

[0029] S101, real-time monitoring and acquisition of motion posture data, physiological index data and environmental context data of the target object;

[0030] S102, Based on the motion posture data, identify the motion primitives in the motion process of the target object and generate a sequence of motion primitives; the motion primitives include at least one of the following: jump primitive, air posture primitive, single-leg landing primitive, change of direction primitive, and acceleration primitive.

[0031] S103, Based on the action primitive sequence, physiological index data, and environmental context data, determine the predetermined action type of the target object;

[0032] S104, assess the physical exertion of the target object based on the predetermined action type, action primitive sequence, and physiological indicator data.

[0033] This application effectively solves the problems of inaccurate identification and evaluation delay in traditional systems during complex cross-scene training by comprehensively analyzing motion posture data, physiological index data, and environmental context data, and by introducing motion primitive recognition and predetermined motion type judgment mechanisms. This enables more accurate and stable monitoring and evaluation of the physical exertion of the target object.

[0034] To better understand the cross-scenario sports fitness pattern recognition and monitoring method proposed in this application, some key terms are explained first. "Target object" generally refers to an athlete undergoing sports training or any individual requiring fitness monitoring. "Motion posture data" refers to data collected by sensors reflecting the target object's body motion state, such as acceleration, angular velocity, and posture angle. "Physiological index data" refers to data reflecting the target object's internal physiological state, such as heart rate, respiratory rate, and blood oxygen saturation. "Environmental context data" refers to data reflecting the characteristics of the training environment in which the target object is located, such as GPS signal strength, Wi-Fi signal strength, air pressure, and temperature. "Motion primitives" refer to the basic, indivisible motion units that constitute complex movements, such as take-off, landing, change of direction, and acceleration. These primitives are the foundation for recognizing higher-level motion patterns. "Motion primitive sequence" refers to a set of motion primitives arranged in chronological order, describing the target object's motion trajectory and behavioral patterns over a period of time. "Predetermined movement type" refers to a predefined, complex movement with a specific training purpose or technical requirement, such as ramp acceleration sprint, continuous change of direction around cones, and jumps and landing cushioning. "Energy expenditure" refers to the energy consumed by the target subject during a specific movement, and is an important indicator for measuring training load and effectiveness.

[0035] The implementation environment of this application can be a wearable monitoring device integrating multiple sensors. This device can collect the aforementioned data in real time and execute subsequent pattern recognition and physical fitness assessment algorithms through a built-in processing unit or by communicating with external computing devices. The core of the cross-scene sports physical fitness pattern recognition and monitoring method of this application lies in the multi-dimensional and refined analysis of the movement state of the target object.

[0036] Firstly, various technologies can be employed to monitor and acquire real-time motion posture data, physiological index data, and environmental context data of the target object. For example, motion posture data can be acquired using inertial measurement unit (IMU) sensors worn on key parts of the target object's body (such as the wrist, ankle, and torso), which can provide information such as three-axis acceleration and three-axis angular velocity. Physiological index data can be monitored in real-time using photoplethysmography (PPG) sensors or electrocardiogram (ECG) sensors to detect heart rate and heart rate variability. Environmental context data can be obtained through GPS modules integrated into wearable devices to acquire location and speed information, Wi-Fi modules to acquire surrounding Wi-Fi signal strength and access point information, and barometers to acquire altitude change data. These sensors can operate independently or, through data fusion technology, provide more comprehensive and accurate raw data.

[0037] Secondly, based on the motion posture data, the motion primitives of the target object during its movement are identified, and a sequence of motion primitives is generated. Motion primitives are the basic units constituting complex motions. The motion primitives defined in this application include at least one of the following: take-off primitive, aerial posture primitive, single-leg landing primitive, change-of-direction step primitive, and acceleration step primitive. For example, the identification of the take-off primitive can be determined by analyzing the instantaneous rate of change and peak value of vertical acceleration; when the vertical acceleration rapidly increases and exceeds a certain threshold, it may indicate that a take-off has occurred. For the single-leg landing primitive, it can be identified by monitoring the negative impact peak value or instantaneous rate of change of vertical acceleration. The change-of-direction step primitive can be identified by analyzing drastic changes in horizontal acceleration and angular velocity; for example, when the horizontal acceleration undergoes a directional reversal in a short period of time, accompanied by a large yaw angular velocity, it may indicate that a change of direction has occurred. The acceleration step primitive can be identified by a continuous increase in horizontal acceleration. The identification of these motion primitives can be based on preset threshold rules or can be achieved by training a large amount of labeled data using machine learning models (such as support vector machines or neural networks). Once a single action primitive is identified, it can be organized into a sequence of action primitives according to the order in which they occur. This sequence can reflect the specific movement pattern of the target object over a period of time.

[0038] Next, based on the action primitive sequence, physiological indicator data, and environmental context data, the predetermined action type of the target object is determined. This step is key to solving the cross-scene recognition problem in this application. For example, when the system recognizes a continuous acceleration step primitive sequence, and the physiological indicator data shows a rapid increase in heart rate, and the dynamic characteristics of the GPS signal in the environmental context data (such as large signal strength variance and high positioning loss frequency) match the preset fluctuation pattern, it can be determined that the target object is performing a "ramp acceleration sprint action". This GPS signal fluctuation pattern usually appears in complex environments where the signal is blocked or affected by multipath effects, such as the transition area of ​​a "composite training corridor". As another example, when a continuous and alternating direction change step primitive sequence is recognized, and the dynamic characteristics of the WLAN signal in the environmental context data show that it is in a weak diffusion range (such as the received signal strength indicators of multiple known access points are all below the preset threshold and have high variance), it can be determined as a "continuous direction change around slalom action". This WLAN signal characteristic usually appears in indoor-outdoor transition areas, with signal strength between strong indoor signals and no outdoor signals. When a sequence of action primitives contains subsequences of take-off, mid-air posture, and single-leg landing primitives appearing sequentially, and these subsequences are completed within a preset time, while the heart rate is higher than a preset heart rate threshold, and the dynamic characteristics of the GPS signal and WLAN signal in the environmental context data both meet preset signal characteristic patterns, it can be identified as a "jump and landing buffer action." Through this multi-source data fusion and pattern matching method, even in scenarios with unclear environmental signals, the predetermined action type of the target object can be determined more accurately.

[0039] Finally, the physical exertion of the target subject is assessed based on the predetermined movement type, movement primitive sequence, and physiological indicator data. The assessment method for physical exertion is customized according to different predetermined movement types. For example, if the predetermined movement type is "hill acceleration sprint," physical exertion can be determined based on the average acceleration amplitude, duration, and the difference between the average heart rate and resting heart rate during the sprint. The acceleration amplitude reflects the intensity of the work done, the duration reflects the total amount of work done, and the heart rate difference reflects the load on the cardiovascular system. If the predetermined movement type is "continuous slalom," physical exertion can be determined based on the number of directional step primitives included in the movement, the amplitude of angular velocity change, and the slope of heart rate change. The number of directional steps and the amplitude of angular velocity change reflect the load and energy consumption of the neuromuscular system, while the slope of heart rate change reflects the instantaneous response of the cardiovascular system. If the predetermined movement type is "jump and landing cushioning," physical exertion can be determined based on the peak vertical acceleration of the landing impact during the movement, the peak heart rate during the jump, and the resting heart rate. The peak vertical acceleration upon landing reflects the impact load on the lower limb muscles, while the peak heart rate during a jump reflects the maximum load on the cardiovascular system from explosive exercise. This targeted assessment model allows for more precise quantification of energy expenditure for different movement types, providing a reliable basis for evaluating training effectiveness.

[0040] The cross-scenario sports fitness pattern recognition and monitoring method proposed in this application effectively solves the challenges faced by existing technologies in complex cross-scenario training by real-time monitoring and comprehensive analysis of the target object's motion posture data, physiological index data, and environmental context data, and on this basis, identifying motion primitives, determining the predetermined motion type, and finally assessing physical exertion.

[0041] Specifically, the core innovation of this application lies in its approach: instead of relying solely on a single environmental signal to determine the training scenario, it incorporates environmental context data as a crucial input for multi-dimensional judgment of the intended action type. In traditional systems, when environmental signals (such as GPS or Wi-Fi signals) become blurred in transitional areas, the system struggles with scenario identification, leading to frequent switching of analysis methods and impacting the accuracy and real-time performance of monitoring. This application, by introducing action primitive sequences and physiological indicator data, integrates them with environmental context data for judgment. Even with unstable environmental signals, it can utilize the characteristics of the movement itself and the body's physiological responses to help determine the intended action type being performed by the target object.

[0042] For example, in scenarios like a "composite training corridor" where both GPS and Wi-Fi signals are unstable, traditional systems may struggle to determine whether training is indoors or outdoors, leading to frequent switching between analytical methods. However, this application, by identifying continuous acceleration step primitive sequences and rapid heart rate increases, combined with specific fluctuation patterns in GPS signals, can accurately determine that the target is performing a "ramp acceleration sprint." This judgment is no longer based on a fuzzy binary scenario division but on a deep understanding of movement behavior and physiological responses, thus avoiding wasted system resources and monitoring delays caused by inaccurate scenario identification.

[0043] Compared to the closest existing technology, the advantages of this application lie in its robustness and adaptability. Existing technologies often fail to identify the scene accurately when facing transitional areas with ambiguous environmental signals, leading to monitoring failure or decreased accuracy. This application constructs a more intelligent and flexible pattern recognition framework by deeply integrating motion primitive recognition, physiological indicator analysis, and environmental context data. This collaborative analysis of multi-source heterogeneous data allows the system to compensate for uncertainties in some data sources (such as environmental signals) through features from other data sources (such as movement posture and physiological indicators), thereby achieving accurate judgment of predetermined motion types and precise assessment of physical exertion. Therefore, this application significantly improves the practicality and reliability of sports fitness monitoring systems in complex and variable training scenarios, providing more precise guidance for athletes' scientific training.

[0044] Specifically, in some embodiments of this application, the predetermined action types, physiological indicator data, and environmental context data in the aforementioned cross-scenario sports fitness pattern recognition and monitoring method are further defined and explained. In the aforementioned cross-scenario sports fitness pattern recognition and monitoring method, the predetermined action types include ramp acceleration sprints, continuous change-of-direction slalom maneuvers, and jumps and landing cushioning maneuvers; the physiological indicator data includes heart rate; and the environmental context data includes GPS signals and Wi-Fi signals.

[0045] Specifically, predetermined movement types refer to representative movement patterns commonly seen in sports training and competition, and their identification is of great significance for physical fitness assessment. In this application, these predetermined movement types are specifically defined as ramp acceleration sprint, continuous change-of-direction slalom, and jump and landing cushioning. Ramp acceleration sprint typically involves high-intensity short-distance acceleration running on sloping terrain; continuous change-of-direction slalom refers to movement with frequent changes of direction in confined spaces, such as weaving between obstacles; and jump and landing cushioning covers the entire process from takeoff to air posture and then to safe landing. Physiological index data can be understood as biological parameters reflecting the internal state of the target object's body. In this application, physiological index data is specifically defined as heart rate, which can be monitored in real time by a heart rate sensor in a wearable device. Heart rate is a direct and important indicator for measuring exercise intensity and physical exertion.

[0046] In practical applications, environmental context data refers to relevant information about the external environment in which the target object is located. In this application, environmental context data is specifically defined as GPS signals and Wi-Fi signals. GPS signals are mainly used to provide information such as the target object's position and speed in the outdoor environment, reflecting its macroscopic movement trajectory and environmental characteristics; Wi-Fi signals are mainly used to provide the target object's position indoors or in a local area, as well as the characteristics of the surrounding wireless network environment, to help determine its specific context.

[0047] This application's solution makes pattern recognition and physical fitness assessment more targeted and accurate by clearly defining predetermined action types, physiological indicator data, and environmental context data. Specifically, the predetermined action types are limited to ramp acceleration sprints, continuous change-of-direction slalom maneuvers, and jumps and landing cushioning maneuvers. This allows the system to focus on these typical high-energy-consuming and high-tech sports movements, thereby optimizing the efficiency and accuracy of the recognition algorithm. Simultaneously, using heart rate as the primary physiological indicator data directly reflects the physiological load of the target subject when performing these specific actions, providing a reliable physiological basis for assessing physical exertion. Furthermore, the introduction of GPS and Wi-Fi signals as environmental context data allows the system to adjust recognition strategies and parameters according to the specific environment of the target subject (e.g., an open outdoor field or an indoor training hall), effectively addressing data differences and challenges in different scenarios and improving the method's cross-scenario adaptability.

[0048] Through the aforementioned technical solutions, this application enables more precise identification of specific sports movement patterns and, combined with key physiological indicators and environmental context data, provides a more detailed and accurate assessment of physical exertion. Specifically, clearly defined movement types help the system to train and apply the recognition model in a targeted manner, improving its ability to distinguish complex movement patterns. The introduction of heart rate as a core physiological indicator allows the assessment of physical exertion to be based not only on the movement itself but also on the real-time physiological responses of the target individual, thereby improving the scientific rigor and personalization of the assessment. Furthermore, by integrating GPS and Wi-Fi signals, the method of this application can intelligently adapt to different sports scenarios. Whether it is outdoor training or indoor activities, it can effectively acquire and utilize environmental information, ensuring the robustness and accuracy of pattern recognition and physical exertion assessment, significantly enhancing the system's practical value and application scope.

[0049] In some embodiments described above, this application proposes a method for determining the predetermined action type of a target object based on action primitive sequences, physiological index data, and environmental context data. However, in practical applications, achieving accurate and robust pattern recognition of complex physical fitness movements based on this multi-source heterogeneous data remains a challenge. Relying solely on a single data source or simple rules may lead to misjudgments or insufficient recognition accuracy, especially in complex sports environments across different scenarios. To address this, this application further proposes a method for determining the predetermined action type of the target object, including the following specific steps:

[0050] If a preset first number of acceleration step primitives are detected within a preset first time window, and the heart rate shows a rapid upward trend, and the dynamic characteristics of the GPS signal in the environmental context data match the preset fluctuation pattern, then it is determined to be a ramp acceleration sprint action.

[0051] If a preset second number of directional change step primitives with continuous and alternating directions are identified within a preset second time window, and the dynamic characteristics of the Wi-Fi signal in the environmental context data show that it is in a weak diffusion range, then it is determined to be a continuous directional change around the slalom action.

[0052] If a subsequence is identified in the action primitive sequence where the take-off primitive, the air posture primitive, and the single-leg landing primitive appear in sequence, and the subsequence is completed within a preset third time window, and the heart rate is higher than a preset heart rate threshold, and the dynamic characteristics of the GPS signal and the Wi-Fi signal in the environmental context data meet a preset signal characteristic pattern, then it is determined to be a jump and landing buffer action.

[0053] Specifically, the aforementioned preset first time window, preset second time window, and preset third time window refer to the time lengths set for specific action types, used to limit the occurrence time of the action primitive sequence. Their values ​​can be adjusted according to the characteristics of different sports and empirical data. The preset first quantity and preset second quantity refer to the minimum number of specific action primitives that need to be identified within the corresponding time window, to ensure the continuity and integrity of the action. A rapid increase in heart rate can be understood as a significant and continuous increase in the target object's heart rate value within a short period, usually determined by calculating the rate of change of heart rate or the difference from the baseline heart rate. A heart rate higher than the preset heart rate threshold means that the target object's heart rate reaches or exceeds a pre-set physiological intensity limit, indicating that they are in a high-intensity exercise state.

[0054] In practical applications, the dynamic characteristics of GPS signals in environmental context data that match a preset fluctuation pattern typically refer to the specific fluctuation patterns of parameters such as GPS signal strength, positioning accuracy, or loss frequency during movement, related to outdoor conditions, high-speed movement, or terrain changes. For example, significant changes in signal strength or positioning accuracy may be observed. The dynamic characteristics of Wi-Fi signals in environmental context data showing a weak diffusion range can be interpreted as generally low Wi-Fi signal strength, possibly originating from multiple unstable access points. This usually indicates that the target object is in an indoor environment, with signal obstruction, or with limited mobility; for example, the received signal strength indication value is generally low and fluctuates significantly. When the dynamic characteristics of both GPS and Wi-Fi signals in environmental context data satisfy a preset signal characteristic pattern, it means that the dynamic characteristics of both GPS and Wi-Fi signals simultaneously meet a specific preset combination pattern to more accurately identify complex motion scenarios, such as simultaneously exhibiting certain combinations of characteristics of high-speed outdoor movement and limited indoor signal coverage.

[0055] This application's solution effectively solves the problem of insufficient recognition accuracy from a single data source by fusing action primitive sequences, physiological indicator data, and environmental context data from multiple dimensions. Specifically, for ramp acceleration sprints, the key features are high-intensity acceleration and an outdoor environment. By identifying continuous acceleration step primitives, the action characteristics of high-intensity exercise can be accurately captured. Simultaneously, combining this with the rapid rise in heart rate reflects the physiological response to high-intensity exercise. Furthermore, the specific dynamic characteristics of GPS signals can effectively indicate that the target is outdoors and may be in a terrain-changing environment, thus distinguishing ramp acceleration sprints from other flat-ground acceleration actions. For continuous slalom maneuvers, the core is rapid directional changes and is usually performed in a confined space. By identifying continuous and alternating directional change step primitives, its agility and directional change characteristics can be accurately captured. Combined with Wi-Fi signal indication of a weak diffusion range, this effectively indicates that the target is indoors or in a semi-enclosed training area, thus distinguishing this action from outdoor wide-range directional running.

[0056] The characteristics of a jump and landing cushioning motion lie in the explosive takeoff, air posture, and subsequent landing impact. By identifying the sequential subsequences of takeoff primitives, air posture primitives, and single-leg landing primitives, the entire process of a jump can be captured completely. Simultaneously, a heart rate higher than a preset heart rate threshold reflects the high physical demands of the jump. Combining the dynamic characteristics of GPS and Wi-Fi signals that meet preset signal characteristic patterns allows for further confirmation of the specific environment in which the action occurs, such as jump training conducted within a specific training area, thereby improving the accuracy and robustness of the recognition.

[0057] In some preferred embodiments, this application is implemented as follows: Assume an athlete is undergoing physical training. When the athlete is sprinting uphill outdoors, the monitoring system continuously identifies a preset first number (e.g., 10) of acceleration step primitives within a preset first time window (e.g., 5 seconds). Simultaneously, the athlete's heart rate rapidly increases from 70 beats / minute at rest to 150 beats / minute, and the variance of the GPS signal strength significantly increases, leading to a higher frequency of location loss. At this time, the system will accurately determine that the athlete is performing an uphill acceleration sprint. When the athlete is performing agility training in an indoor basketball court, performing continuous directional changes around cones, the system identifies a preset second number (e.g., 15) of continuous and alternating directional change step primitives within a preset second time window (e.g., 8 seconds). Simultaneously, the Wi-Fi signal strength is generally below -70dBm, and the signal strength from multiple access points fluctuates significantly. The system then determines that the athlete is performing continuous directional changes around cones. When an athlete is training with a box vault in a gym, if the system recognizes that the sequence of motion primitives includes a take-off primitive, an aerial posture primitive, and a single-leg landing primitive, and this subsequence is completed within a preset third time window (e.g., 2 seconds), and the athlete's heart rate reaches 170 beats per minute, exceeding a preset heart rate threshold (e.g., 140 beats per minute), and the dynamic characteristics of both the GPS and Wi-Fi signals meet preset indoor complex environment signal patterns, the system determines that the athlete is performing a jump and landing cushioning action. These specific judgment conditions and thresholds can be calibrated and optimized based on the actual training scenario and individual athlete differences.

[0058] Specifically, in the process of determining the predetermined action type of the target object, the dynamic characteristics of the environmental context data can be further defined. The dynamic characteristics of the GPS signal in the aforementioned environmental context data conform to a preset fluctuation pattern, including: the signal strength variance of the GPS signal is greater than a preset variance threshold, and the positioning loss frequency is higher than a preset frequency threshold; the dynamic characteristics of the Wi-Fi signal in the aforementioned environmental context data show that it is in a weak diffusion range, including: the received signal strength indication values ​​from multiple known access points in the Wi-Fi signal are all lower than a preset strength threshold, and their variance is higher than a preset variance threshold; the dynamic characteristics of the GPS signal and the Wi-Fi signal in the aforementioned environmental context data satisfy a preset signal characteristic pattern, including: the signal strength variance of the GPS signal is greater than a preset variance threshold, and the positioning loss frequency is higher than a preset frequency threshold; simultaneously, the received signal strength indication values ​​from multiple known access points in the Wi-Fi signal are all lower than a preset strength threshold, and their variance is higher than a preset variance threshold.

[0059] The variance of GPS signal strength refers to the dispersion of the received signal strength value relative to its average value within a certain time window. When a target object moves in complex terrain such as a ramp or in a concealed environment, the reception quality of the GPS signal is often unstable, resulting in large fluctuations in signal strength, i.e., high variance. The location loss frequency refers to the number or proportion of times the GPS module fails to provide effective location information within a specific time period. This typically occurs in scenarios with severe signal obstruction or where rapid movement makes signal acquisition difficult. Preset variance thresholds and preset frequency thresholds can be set based on actual application scenarios and empirical data to distinguish between normal movement and movement under specific circumstances.

[0060] The dynamic characteristics of the Wi-Fi signal indicate a weak diffusion range, which can be interpreted as poor Wi-Fi coverage or sparse signal sources in the target's environment. Specifically, if the Received Signal Strength Indication (RSSI) values ​​from multiple known access points are all below a preset strength threshold, it indicates that the target is far from these access points or that there are many obstacles. Simultaneously, if the variance is higher than a preset variance threshold, it further indicates large signal strength fluctuations, possibly due to rapid movement of the target in an indoor or semi-indoor environment, leading to unstable signal reception. The preset strength threshold and preset variance threshold can also be calibrated according to the actual environment.

[0061] When the dynamic characteristics of both GPS and Wi-Fi signals need to simultaneously meet a preset signal characteristic pattern, it means that the target object may be in a complex environment simultaneously affected by GPS signal obstruction and weak Wi-Fi signal diffusion, such as performing jumping and landing cushioning actions in an urban canyon, on the edge of an indoor stadium, or in a multi-story building area. This combined pattern provides a more accurate basis for judging the environmental situation.

[0062] This application's solution quantifies the dynamic characteristics of GPS and Wi-Fi signals in environmental context data, thereby enabling more accurate judgment of predetermined action types. Specifically, by setting thresholds for GPS signal strength variance, positioning loss frequency, and Wi-Fi signal strength indication and variance, abstract concepts such as "matching preset fluctuation patterns," "weak diffusion range," and "preset signal characteristic patterns" can be transformed into calculable and determinable numerical conditions. This quantification process allows the system to more accurately identify the specific environmental context in which the target object is located, such as distinguishing between movement in open areas and sheltered environments, or the difference between indoor and outdoor movement. This provides a more reliable input for subsequent action type judgment, avoiding misjudgments caused by ambiguity in environmental context assessment.

[0063] Through the above technical solutions, this application can significantly improve the accuracy and robustness of cross-scenario sports fitness pattern recognition and monitoring methods. By refining and quantifying the dynamic characteristics of environmental context data, the system can more accurately determine the predetermined action type of the target object when facing complex and ever-changing sports scenarios. For example, in ramp acceleration sprinting, accurately identifying the fluctuation characteristics of GPS signals can effectively eliminate interference from flat ground sprinting; in continuous change-of-direction slalom maneuvers, identifying the weak diffusion range of Wi-Fi signals can more accurately determine whether the target object is in an indoor or semi-indoor slalom training scenario; in jumping and landing cushioning actions, combining the composite characteristics of GPS and Wi-Fi signals can more reliably identify jumping activities performed in complex environments. This refined environmental context judgment provides a more solid foundation for subsequent physical exertion assessment, thereby enhancing the practical value and accuracy of the entire monitoring method.

[0064] This application further proposes the above-mentioned method of assessing the physical exertion of the target object based on the predetermined action type, action primitive sequence, and physiological index data, including: if the predetermined action type is ramp acceleration sprint, then the physical exertion of the target object is determined based on the average acceleration amplitude, duration, and the difference between the average heart rate and the resting heart rate during the ramp acceleration sprint action; if the predetermined action type is continuous slalom, then the physical exertion of the target object is determined based on the number of directional step primitives, the amplitude of angular velocity change, and the slope of heart rate change included in the continuous slalom action; if the predetermined action type is jump and landing cushioning, then the physical exertion of the target object is determined based on the peak vertical acceleration of the landing impact during the jump and landing cushioning action, the peak heart rate during the jump, and the resting heart rate.

[0065] Specifically, when the intended movement type is determined to be a hill start sprint, the determination of physical exertion is based on the average acceleration amplitude, duration, and the difference between the average heart rate and resting heart rate during the duration of the movement. The average acceleration amplitude reflects the explosive power and intensity during the sprint, the duration quantifies the duration of high-intensity exercise, and the difference between the average heart rate and resting heart rate directly indicates the degree of cardiovascular load during exercise. These parameters comprehensively consider the physiological and mechanical characteristics of hill start sprints, a high-intensity, short-duration explosive movement.

[0066] When the predetermined movement type is determined to be continuous slalom, the physical exertion is determined based on the number of directional step units, the amplitude of angular velocity change, and the slope of heart rate change included in the continuous slalom movement. The number of directional step units directly reflects the frequency and number of times the athlete changes direction, the amplitude of angular velocity change quantifies the abruptness and intensity of the directional change, and the slope of heart rate change characterizes the speed and amplitude of the heart rate's response to intermittent high-intensity directional changes. These parameters effectively capture the comprehensive requirements of agility, explosive power, and cardiorespiratory endurance in continuous slalom.

[0067] When the intended movement type is determined to be a jump and landing cushioning, physical exertion is determined based on the peak vertical acceleration of the landing impact, the peak heart rate during the jump, and the resting heart rate. The peak vertical acceleration of the landing impact is a key indicator of the impact force experienced by the body upon landing and is closely related to the load on the musculoskeletal system. The peak heart rate during the jump reflects the maximum load on the cardiovascular system at the moment of the jump. The resting heart rate serves as a baseline to assess the relative level of physiological stress. These parameters collectively characterize the explosive power, cushioning capacity, and cardiopulmonary load during the jumping movement.

[0068] The proposed solution employs customized physical exertion assessment models for different predetermined movement types, enabling more accurate capture of the unique physiological and biomechanical characteristics of each movement pattern. For example, ramp acceleration sprints primarily involve explosive power and cardiorespiratory endurance, thus its assessment focuses on acceleration and heart rate changes; continuous slalom maneuvers emphasize agility and rapid reaction, therefore its assessment focuses on the number of swerves and angular velocity changes; while jumps and landing cushioning involve explosive power, impact absorption, and cardiorespiratory load, thus its assessment focuses on landing impact and peak heart rate. It is precisely this differentiated assessment strategy that makes the quantification of physical exertion more aligned with the needs of actual sports scenarios, avoiding the assessment bias that may arise from a single model.

[0069] In some preferred embodiments, this application is implemented as follows: Assume the target object is undergoing physical training. When the system detects that the target object is performing a hill-climb sprint, the monitoring module continuously collects the average acceleration amplitude during the duration of the movement, which is 5.2 m / s², lasting for 25 seconds. Simultaneously, it records the target object's average heart rate as 175 beats / minute and its resting heart rate as 60 beats / minute. Based on this data, the physical fitness assessment module calculates a heart rate difference of 115 beats / minute and, combining the average acceleration amplitude and duration, determines the physical exertion of this hill-climb sprint to a specific value, such as 250 joules / kg, using a preset algorithm model.

[0070] When the system detects that the target is performing a continuous slalom maneuver, the monitoring module will count the number of directional step primitives involved in the maneuver as 12, the average angular velocity change as 180 degrees / second, and the heart rate change slope as 3 beats / minute / second. The physical fitness assessment module will use these parameters, through an assessment model specifically designed for continuous slalom maneuvers, to determine the physical exertion of this training session, for example, 300 joules / kg.

[0071] When the system detects that the target is performing a jump and landing cushioning motion, the monitoring module collects data such as the peak vertical acceleration of the landing impact (-8g), the peak heart rate during the jump (190 beats / minute), and the target's resting heart rate (60 beats / minute). The fitness assessment module integrates this data and, using a fitness consumption model for jump and landing cushioning motions, determines the fitness expenditure of this jump, for example, 280 joules / kg. In this way, the system can provide more accurate and personalized fitness consumption assessment results based on the specific type of movement.

[0072] Specifically, in some embodiments of the aforementioned cross-scenario sports fitness pattern recognition and monitoring method, the step of identifying motion primitives in the target object's movement process based on motion posture data and generating a sequence of motion primitives can be further refined. The aforementioned step of identifying motion primitives in the target object's movement process based on motion posture data and generating a sequence of motion primitives includes: in response to the identification of a jump primitive, activating a landing expectation state and initiating a landing expectation time window; within the landing expectation time window, monitoring the vertical acceleration of the inertial measurement unit; and determining that a potential landing impact signal has occurred when the instantaneous value of the vertical acceleration is lower than a preset negative impact threshold, or when the instantaneous rate of change of the vertical acceleration is lower than a preset negative rate of change threshold.

[0073] The "response to jump primitive detection" mechanism refers to the system triggering a landing detection mechanism when it detects a motion pattern matching jump characteristics by analyzing motion attitude data, such as data from an inertial measurement unit (IMU). Jump primitive detection can be based on a rapid increase in vertical acceleration followed by a brief period of weightlessness. Activating a landing expectation state and initiating a landing expectation time window means that after confirming the target object has completed its jump, the system enters a anticipated landing monitoring mode and sets a limited time period to focus on the appearance of landing impact signals. This landing expectation time window is designed to avoid misjudgments and improve detection efficiency. Monitoring the vertical acceleration of the inertial measurement unit within the landing expectation time window means continuously collecting and analyzing vertical acceleration data from the inertial measurement unit within the set time window. Vertical acceleration is a key indicator reflecting the force experienced by the human body upon contact with the ground. When the instantaneous value of the vertical acceleration is lower than a preset negative impact threshold, or the instantaneous rate of change of the vertical acceleration is lower than a preset negative rate of change threshold, a potential landing impact signal is determined to have occurred. This means identifying the characteristics of a landing impact by setting specific thresholds. Upon landing, the human body's contact with the ground generates an upward reaction force, resulting in a significant negative peak in the vertical acceleration (i.e., the acceleration direction is opposite to the direction of gravity), or a very rapid rate of decrease in acceleration (rate of change). By monitoring these characteristics, a preliminary judgment can be made as to whether a landing impact has occurred. The preset negative impact threshold and the preset negative rate of change threshold are empirical values ​​pre-set based on human kinematics and dynamics characteristics, or parameters obtained through training models.

[0074] This application's solution, upon identifying the takeoff element, immediately activates the landing anticipation state and initiates a landing anticipation time window, thereby limiting the detection range of landing impact to a reasonable time window. Within this time window, the system continuously monitors the vertical acceleration of the inertial measurement unit. When the instantaneous value of the vertical acceleration is lower than a preset negative impact threshold, or its instantaneous rate of change is lower than a preset negative rate of change threshold, these signal characteristics are considered potential landing impact signals. This mechanism can effectively capture the typical impact characteristics generated when the human body contacts the ground after a jump, laying the foundation for subsequent accurate identification of single-foot landing elements.

[0075] The above technical solution enables more accurate and timely identification of potential landing impact signals after a jump. By limiting the expected landing time window, interference from non-landing impact signals can be effectively reduced, improving the accuracy of landing detection. Furthermore, combining the instantaneous value and rate of change of vertical acceleration for judgment makes landing impact identification more sensitive and reliable, providing more accurate input data for subsequent motion element sequence generation and physical exertion assessment.

[0076] This application further proposes to identify and generate motion primitives based on the motion posture data of the target object during its motion process, including: if multiple potential landing impact signals are detected within the expected landing time window, the earliest potential landing impact signal with the largest impact amplitude is identified as the starting signal of the single-foot landing primitive; after identifying the starting signal, a landing stability confirmation window is initiated; if, within the landing stability confirmation window, the vertical acceleration recovers to the static range and the angular velocities of each axis are all lower than a preset angular velocity stability threshold, the single-foot landing primitive is finally confirmed.

[0077] Specifically, "multiple potential landing impact signals" refer to all acceleration signals that meet a preset negative impact threshold or a preset negative rate of change threshold that the system may identify within the expected landing time window during the target object's jump from takeoff to landing, due to reasons such as body contact with the ground or brief secondary contact. To accurately identify the true single-foot landing element from these signals, this application proposes a screening mechanism. The "earliest temporally occurring and largest impact amplitude potential landing impact signal" is selected as the "starting signal of the single-foot landing element." "Earliest temporally occurring" aims to capture the moment of the first effective contact with the ground, while "largest impact amplitude" is used to distinguish between the main landing impact and any possible secondary or minor contact.

[0078] Furthermore, after confirming the initial signal of the single-leg landing primitive, the system will "initiate a landing stability confirmation window" to ensure the integrity and stability of the landing action. This window is a preset time period used to observe the posture changes of the target object after landing. Within this window, if "the vertical acceleration returns to the static range," that is, the acceleration value approaches zero or is within a very small fluctuation range, it indicates that the target object has stopped falling or rebounding, and the body is in a relatively static state. At the same time, if "the angular velocity of each axis is lower than the preset angular velocity stability threshold," it means that the rotation or swaying of the target object's body has significantly decreased and tended to stabilize after landing. When both conditions are met simultaneously, the system "finally confirms the single-leg landing primitive," thereby ensuring that the identified landing action is complete and stable.

[0079] This application's solution effectively addresses the problem of accurately identifying single-foot landing elements in complex motion environments by introducing a screening mechanism for multiple potential landing impact signals and a post-landing stability confirmation mechanism. Specifically, within the expected landing time window, by prioritizing the earliest-appearing potential landing impact signal with the largest impact amplitude as the starting signal for the single-foot landing element, secondary impacts caused by body swaying or uneven ground can be effectively eliminated, thus capturing the actual landing moment more accurately. Furthermore, after identifying the starting signal, by activating the post-landing stability confirmation window and monitoring whether the vertical acceleration and angular velocities of each axis have returned to the static range and are below the preset stability threshold, the integrity and stability of the landing action can be further verified. This two-stage confirmation mechanism enables the system not only to identify landing impacts but also to ensure the stability of the target object's posture after landing, thereby avoiding misjudging unstable landings or false impacts as valid single-foot landing elements.

[0080] As a specific implementation, suppose an athlete is training with a vaulting box. When the athlete jumps off the box, the inertial measurement unit in their wearable monitoring device records vertical acceleration data. Within the athlete's expected landing time window, the sensor may detect multiple negative impact signals: the first is the primary impact of the foot contacting the ground, followed by one or two smaller secondary impacts due to adjustments in the body's center of gravity or slight rebound. According to the scheme of this application, the system first identifies all potential landing impact signals. Then, by comparing the time and impact amplitude of these signals, the system selects the earliest occurring signal with the largest impact amplitude as the starting signal of the single-foot landing primitive, which typically corresponds to the moment the athlete's foot first contacts the ground. After confirming the starting signal, the system immediately initiates a post-landing stability confirmation window. Within this window, if the athlete's vertical acceleration is detected to rapidly recover to near-zero static range, and the angular velocities of their body in all axes (e.g., pitch, roll, yaw angular velocities) rapidly decrease and remain below a preset stability threshold, the system finally confirms this single-foot landing primitive. Conversely, if the athlete's body continues to sway after landing and the angular velocity fails to stabilize quickly, the system may not ultimately confirm the single-foot landing primitive, thus avoiding misjudging an unstable landing action as a valid primitive.

[0081] Specifically, the above-mentioned operation in response to the identification of the jump primitive can be further refined into the following steps. The response to the identification of the jump primitive includes:

[0082] Monitor the vertical acceleration Az_t at the current moment and the vertical acceleration Az_t-1 at the previous moment;

[0083] When Az_t is greater than the first positive threshold and Az_t-1 is less than the second positive threshold, and the average value of the vertical acceleration is between the first negative threshold and the third positive threshold within the subsequent first preset time period, the jump primitive is identified.

[0084] Specifically, the vertical acceleration Az_t refers to the acceleration value of the target object along the vertical direction acquired by the inertial measurement unit (IMU) sensor at the current time t. The vertical acceleration Az_t-1 refers to the acceleration value of the target object along the vertical direction acquired by the IMU sensor at the time immediately preceding the current time t (t-1). These acceleration values ​​are usually measured in units of gravitational acceleration g and undergo appropriate filtering to reduce noise interference. The first positive threshold and the second positive threshold are pre-set positive acceleration thresholds used to define the intensity of the upward push during takeoff. The first positive threshold is usually set to a higher positive value to capture the strong upward acceleration at the moment of takeoff; the second positive threshold is usually set to a lower positive value to confirm that the target object is not in a continuous upward motion state before the strong upward acceleration, thereby more accurately identifying the starting point of the takeoff. The first preset duration refers to a short time window whose length can be adjusted according to the actual motion type and sampling frequency. For example, it can be set to 0.1 to 0.5 seconds to assess whether the target object enters a state of airborne or low gravity after takeoff. The first negative threshold and the third positive threshold are used to define the effective range of the average vertical acceleration within the first preset duration. The first negative threshold can be a value close to zero or slightly less than zero, indicating a slight downward acceleration that may occur during the airborne phase; the third positive threshold is a small positive value to ensure that there is no significant upward acceleration during this period, thus excluding other non-takeoff motion patterns.

[0085] The proposed solution captures the acceleration changes of a target object at the moment of takeoff by monitoring its vertical acceleration Az_t and its previous vertical acceleration Az_t-1. Specifically, when Az_t is greater than a first positive threshold and Az_t-1 is less than a second positive threshold, it indicates that the target object is undergoing a rapid upward acceleration process, which is usually the initial stage of the takeoff action. Furthermore, by monitoring the average value of the vertical acceleration over a subsequent first preset time period and determining whether it falls between a first negative threshold and a third positive threshold, it is possible to effectively confirm whether the target object has entered a state of flight or low gravity. This combined judgment mechanism can effectively distinguish between a true takeoff action and other body swaying or non-takeoff-related upward movements, thereby improving the accuracy and robustness of takeoff primitive identification.

[0086] The above technical solution enables more accurate and reliable identification of the target object's takeoff primitives. Traditional takeoff identification methods may rely solely on a single acceleration peak, making them susceptible to interference from noise or atypical motion. This application combines the acceleration change trends before and after takeoff with the average acceleration state over a period after takeoff to form a multi-dimensional judgment criterion, significantly reducing the false judgment rate and ensuring the accuracy of subsequent landing expectation state activation. This lays a solid foundation for the effective operation of the entire physical fitness pattern recognition and monitoring method.

[0087] Specifically, in this application, in order to achieve real-time monitoring and collection of motion posture data, physiological index data and environmental context data of the target object, the target object is configured as a wearable monitoring device.

[0088] In some embodiments of this application, the target object includes a wearable monitoring device; the wearable monitoring device includes a sensor group for real-time monitoring and collection of the target object's motion posture data, physiological index data, and environmental context data; the sensor group includes at least an inertial measurement sensor, a heart rate sensor, a GPS module, and a Wi-Fi module.

[0089] The wearable monitoring device can be understood as a smart device that can be worn on the body of a target object, such as a smartwatch, smart bracelet, smart clothing, or athletic shoes with integrated sensors. Its main function is to closely follow the movement of the target object to continuously and uninterruptedly acquire relevant data. Furthermore, the sensor array is the core component of this wearable monitoring device, integrating various types of sensors to achieve comprehensive data acquisition across different dimensions. Specifically, the inertial measurement sensors may include accelerometers, gyroscopes, and magnetometers, used to accurately capture the target object's motion posture, velocity, and direction changes in three-dimensional space, thereby generating motion posture data. For example, an accelerometer can measure linear acceleration, and a gyroscope can measure angular velocity; these data together constitute a detailed description of the motion posture. The heart rate sensor is used to monitor the target object's heart rate in real time, as an important physiological indicator. This sensor typically employs techniques such as photoplethysmography (PPG) to calculate heart rate by detecting changes in blood volume, aiming to reflect the target object's physiological load and energy consumption during movement. In addition, the GPS and Wi-Fi modules are used to collect environmental context data. The GPS module receives satellite signals to provide dynamic characteristics such as the target object's geographical location and movement speed, making it particularly suitable for outdoor sports scenarios. The Wi-Fi module scans surrounding wireless network signals to provide location information and environmental characteristics within indoor or specific areas. For example, it uses received signal strength indication (RSSI) values ​​to determine the environmental range of the target object, aiming to assist in determining the type and characteristics of the sports scene.

[0090] This application's solution concretizes the target object into a wearable monitoring device and integrates an inertial measurement sensor, heart rate sensor, GPS module, and Wi-Fi module into a sensor array, achieving comprehensive, real-time, and continuous acquisition of motion posture data, physiological index data, and environmental context data. Specifically, the inertial measurement sensor can capture the fine motion trajectories and posture changes of various body parts at high frequency, providing basic data for subsequent motion element recognition. The heart rate sensor continuously monitors the target object's heart rate fluctuations, directly reflecting its exercise intensity and physiological response. Simultaneously, the GPS and Wi-Fi modules work together to provide accurate or relatively accurate location information and environmental characteristics in outdoor and indoor environments, respectively, enabling the system to recognize and adapt to different exercise scenarios. Therefore, the synergistic effect of the aforementioned sensor array ensures that the required multi-dimensional data can be accurately and effectively acquired in various exercise scenarios, providing reliable data support for subsequent physical fitness pattern recognition and physical exertion assessment.

[0091] Secondly, this application also discloses a cross-scenario sports fitness pattern recognition and monitoring system, such as... Figure 2 As shown, the system includes:

[0092] Monitoring module 201 is used to monitor and acquire motion posture data, physiological index data and environmental context data of the target object in real time;

[0093] The identification and generation module 202 is used to identify the motion primitives of the target object during its movement based on the motion posture data, and generate a sequence of motion primitives; the motion primitives include at least one of the following: jump primitive, air posture primitive, single-leg landing primitive, change-of-direction step primitive, and acceleration step primitive;

[0094] The judgment module 203 is used to determine the predetermined action type of the target object based on the action primitive sequence, physiological indicator data, and environmental context data.

[0095] The physical fitness assessment module 204 is used to assess the physical fitness consumption of the target object based on the predetermined movement type, movement primitive sequence and physiological index data.

[0096] The first aspect of the implementation described above already includes a method for real-time monitoring and acquisition of the target object's motion posture data, physiological index data, and environmental context data; identification of motion primitives during the target object's movement based on the motion posture data and generation of motion primitive sequences; determination of the target object's predetermined action type based on the motion primitive sequences, physiological index data, and environmental context data; and assessment of the target object's physical exertion based on the predetermined action type, motion primitive sequences, and physiological index data. These details will not be elaborated upon here. It should be emphasized that the cross-scenario sports physical fitness pattern recognition and monitoring system proposed in this application implements the above functions in a modular manner.

[0097] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for cross-scenario sports physical fitness pattern recognition and monitoring, characterized in that, The method includes: Real-time monitoring and acquisition of motion posture data, physiological index data, and environmental context data of the target object; Based on the motion posture data, the motion primitives in the motion process of the target object are identified, and a sequence of motion primitives is generated; the motion primitives include at least one of the following: jump primitive, air posture primitive, single-leg landing primitive, change of direction primitive, and acceleration primitive. Based on the action primitive sequence, physiological index data, and environmental context data, determine the predetermined action type of the target object; The physical exertion of the target subject is assessed based on the predetermined action type, action primitive sequence, and physiological indicator data.

2. The cross-scenario sports physical fitness pattern recognition and monitoring method according to claim 1, characterized in that, The predetermined action types include ramp acceleration sprint, continuous change of direction around cones, and jump and landing cushioning actions; the physiological indicator data includes heart rate; and the environmental context data includes GPS signal and Wi-Fi signal.

3. The cross-scenario sports physical fitness pattern recognition and monitoring method according to claim 2, characterized in that, The step of determining the predetermined action type of the target object based on the action primitive sequence, physiological indicator data, and environmental context data includes: If a preset first number of acceleration step primitives are detected within a preset first time window, and the heart rate shows a rapid upward trend, and the dynamic characteristics of the GPS signal in the environmental context data match the preset fluctuation pattern, then it is determined to be a ramp acceleration sprint action. If a preset second number of directional change step primitives with continuous and alternating directions are identified within a preset second time window, and the dynamic characteristics of the Wi-Fi signal in the environmental context data show that it is in a weak diffusion range, then it is determined to be a continuous directional change around the slalom action. If a subsequence is identified in the action primitive sequence where the take-off primitive, the air posture primitive, and the single-leg landing primitive appear in sequence, and the subsequence is completed within a preset third time window, and the heart rate is higher than a preset heart rate threshold, and the dynamic characteristics of the GPS signal and the Wi-Fi signal in the environmental context data meet a preset signal characteristic pattern, then it is determined to be a jump and landing buffer action.

4. The cross-scenario sports physical fitness pattern recognition and monitoring method according to claim 3, characterized in that, The dynamic characteristics of the GPS signal in the environmental context data are consistent with the preset fluctuation pattern, including: the signal strength variance of the GPS signal is greater than the preset variance threshold, and the positioning loss frequency is higher than the preset frequency threshold. The dynamic characteristics of the Wi-Fi signal in the environmental context data show that it is in a weak diffusion range, including: the received signal strength indication values ​​from multiple known access points in the Wi-Fi signal are all lower than a preset strength threshold, and their variance is higher than a preset variance threshold. The dynamic characteristics of the GPS signal and the Wi-Fi signal in the environmental context data meet the preset signal characteristic pattern, including: the signal strength variance of the GPS signal is greater than the preset variance threshold, and the positioning loss frequency is higher than the preset frequency threshold; at the same time, the received signal strength indication values ​​of the Wi-Fi signal from multiple known access points are all lower than the preset strength threshold, and their variance is higher than the preset variance threshold.

5. The cross-scenario sports physical fitness pattern recognition and monitoring method according to claim 4, characterized in that, The step of assessing the physical exertion of the target subject based on the predetermined action type, action primitive sequence, and physiological indicator data includes: If the predetermined action type is ramp acceleration sprint, the physical exertion of the target object is determined based on the average acceleration amplitude, duration, and the difference between the average heart rate and the resting heart rate during the ramp acceleration sprint action. If the predetermined action type is continuous slalom, then the physical exertion of the target object is determined based on the number of directional step primitives, the amplitude of angular velocity change, and the slope of heart rate change included in the continuous slalom action. If the predetermined action type is a jump and landing cushioning, the physical exertion of the target object is determined based on the peak vertical acceleration of the landing impact during the jump and landing cushioning action, the peak heart rate during the jump, and the resting heart rate.

6. The cross-scenario sports physical fitness pattern recognition and monitoring method according to claim 1, characterized in that, Based on the motion posture data, identify the motion primitives during the movement of the target object, and generate a sequence of motion primitives, including: In response to the detection of the take-off primitive, a landing expectation state is activated and a landing expectation time window is initiated. During the expected landing time window, the vertical acceleration of the inertial measurement unit is monitored; When the instantaneous value of the vertical acceleration is lower than a preset negative impact threshold, or the instantaneous rate of change of the vertical acceleration is lower than a preset negative rate of change threshold, a potential impact signal is determined to have occurred.

7. The cross-scenario sports physical fitness pattern recognition and monitoring method according to claim 6, characterized in that, Based on the motion posture data, identify the motion primitives of the target object during its motion process, including: If multiple potential landing impact signals are detected within the expected landing time window, the earliest potential landing impact signal with the largest impact amplitude is identified as the starting signal of the single-foot landing element. After the initial signal is detected, a landing stability confirmation window is initiated. If, within the landing stability confirmation window, the vertical acceleration returns to the static range and the angular velocities of each axis are all below the preset angular velocity stability threshold, then the single-foot landing element is finally confirmed.

8. The cross-scenario sports physical fitness pattern recognition and monitoring method according to claim 6, characterized in that, The response to the identification of the jump primitive includes: Monitor the vertical acceleration Az_t at the current moment and the vertical acceleration Az_t-1 at the previous moment; When Az_t is greater than the first positive threshold and Az_t-1 is less than the second positive threshold, and the average value of the vertical acceleration is between the first negative threshold and the third positive threshold within the subsequent first preset time period, the jump primitive is identified.

9. The cross-scenario sports physical fitness pattern recognition and monitoring method according to claim 1, characterized in that, The target object includes a wearable monitoring device; the wearable monitoring device includes a sensor group for real-time monitoring and collection of the target object's motion posture data, physiological index data, and environmental context data; the sensor group includes at least an inertial measurement sensor, a heart rate sensor, a GPS module, and a Wi-Fi module.

10. A cross-scenario sports physical fitness pattern recognition and monitoring system, characterized in that, The system includes: The monitoring module is used to monitor and acquire the target object's motion posture data, physiological index data, and environmental context data in real time. The identification and generation module is used to identify the motion primitives of the target object during its movement based on the motion posture data, and generate a sequence of motion primitives; the motion primitives include at least one of the following: jump primitive, air posture primitive, single-leg landing primitive, change-of-direction step primitive, and acceleration step primitive; The judgment module is used to determine the predetermined action type of the target object based on the action primitive sequence, physiological indicator data, and environmental context data. The physical fitness assessment module is used to assess the physical fitness consumption of the target object based on the predetermined movement type, movement primitive sequence, and physiological indicator data.

Images