A method and system for sports and health data processing
By converting sports and health data into a two-dimensional flow field representation and encoding streamline trajectories, and combining dynamic properties and trusted motion fingerprint generation, the problems of low encoding efficiency and insufficient privacy protection in sports and health data processing are solved, achieving efficient data encoding and privacy protection.
Patent Information
- Application Number
- CN202611253473.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-18
- Publication Date
- 2026-11-10
Smart Images

Figure CN123069423A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, and in particular to a method and system for processing sports and health data. Background Technology
[0002] With the widespread adoption of smart wearable devices and mobile internet, the collection of sports and health data has exploded. This data includes heart rate, step count, GPS tracking, and Inertial Measurement Unit (IMU) sensor data. This data has significant commercial value in areas such as personal health management, sports training optimization, insurance pricing, and sports betting, and the trading of sports and health data is gradually becoming an emerging industry.
[0003] In processing sports and health data, the main technologies employed are deep learning models for encoding and analyzing time-series data. Convolutional neural networks lack the ability to handle long-range dependencies when processing time-series tasks. Long Short-Term Memory (LSTM) networks possess some long-range modeling capabilities, but their computational efficiency is low. Transformer networks have strong modeling capabilities, but their computational efficiency is extremely low, failing to meet the needs of real-time transaction verification. Furthermore, sports and health data is directly linked to economic value, and data providers may engage in dishonest practices such as tampering with step counts, modifying heart rates, and fabricating training records, while trading platforms lack effective means of verifying authenticity. Buyers want to verify the authenticity and value of the data before a transaction, but sellers are unwilling to provide the original data to avoid privacy leaks.
[0004] While the aforementioned existing technologies can achieve the collection and basic analysis of sports and health data, there is still room for improvement in terms of encoding efficiency for long-term time-series data, data authenticity verification, and transaction privacy protection. Existing encoding methods treat sports time-series data as unstructured sequences for uniform encoding, failing to utilize the spatiotemporal directionality and dynamic laws inherent in sports data, resulting in low encoding efficiency and an inability to preserve the intrinsic physical structure of sports data. Summary of the Invention
[0005] To address the aforementioned issues, this application provides a method and system for processing sports and health data. By converting sports and health data into a two-dimensional flow field representation with physical meaning and performing sequence encoding along the directional path of the flow field, the efficiency of long-term processing of sports and health data is improved, while providing anti-counterfeiting verification and privacy protection transaction methods.
[0006] This application discloses a method for processing sports and health data, the method comprising: A velocity field is constructed based on multi-source sports and health data; the multi-source sports and health data includes at least two of the following: heart rate data from smart wearable devices, GPS trajectory data, and IMU sensor data; The motion dynamics tensor is calculated based on the dynamic properties of the velocity field to generate a two-dimensional discrete flow field; Calculate streamline trajectories based on the aforementioned two-dimensional discrete flow field; A sequence scan is performed along the streamline trajectory to generate a flow field scanning feature sequence.
[0007] Optionally, constructing the velocity field based on multi-source sports and health data includes: Multimodal alignment is performed on the multi-source sports and health data to obtain unified time-series data; The velocity field is constructed based on the unified time-series data.
[0008] Optionally, constructing the velocity field based on the unified time-series data includes: The unified time-series data is mapped to a two-dimensional spatial grid, so that each grid point corresponds to a velocity vector; the velocity vector contains motion direction and speed information. The velocity field is generated based on the two-dimensional spatial grid containing the velocity vector.
[0009] Optionally, calculating the motion dynamics tensor based on the dynamic properties of the velocity field includes: The divergence and curl are calculated based on the velocity field to obtain the dynamic properties; The kinematic tensor is calculated based on the divergence and the curl.
[0010] Optionally, the calculation of streamline trajectories based on the two-dimensional discrete flow field includes: The streamline trajectory is obtained by solving the streamline equation of the two-dimensional discrete flow field through numerical integration.
[0011] Optionally, after generating the flow field scanning feature sequence, the method further includes: Encoding is performed based on the flow field scanning feature sequence; the update rate of the state during the encoding process is related to the trajectory curvature of the flow field scanning feature sequence; the state includes long-term physical state and short-term action state; During the encoding process, the long-term physical state and the short-term action state are fused to generate fused state features.
[0012] Optionally, the encoding based on the flow field scanning feature sequence includes: The trajectory curvature is calculated based on the velocity and acceleration components of the flow field scanning feature sequence. The update rate is adaptively adjusted based on the trajectory curvature; the trajectory curvature is positively correlated with the update rate.
[0013] Optionally, after generating the fusion state features, the method further includes: The fused state features are randomly projected and hashed to generate a trusted motion fingerprint.
[0014] Optionally, after generating a trusted motion fingerprint, the method further includes: During the transaction phase, the video encoding characteristics of the video data from the buyer are obtained; Calculate the cosine similarity between the video coding features and the trusted motion fingerprint; The authenticity of the video data is determined based on the cosine similarity.
[0015] Based on the above-mentioned method for processing sports and health data, this application also discloses a system for processing sports and health data, used to execute the above-mentioned method. The system includes: a multimodal alignment module, a two-dimensional flow field construction module, a flow field scanning module, a curvature-sensing dual-state encoder, a trusted fingerprint generation module, and a trusted verification module. The multimodal alignment module is used to acquire multi-source sports and health data and to perform multimodal alignment on the multi-source sports and health data. The two-dimensional flow field construction module is used to construct a velocity field and introduce dynamic properties to construct a two-dimensional discrete flow field. The flow field scanning module is used to calculate streamline trajectories and perform sequential scanning along the streamline trajectories to generate a flow field scanning feature sequence. The curvature-aware dual-state encoder is used to adaptively adjust the state update rate according to the trajectory curvature and fuse long-term physical state and short-term action state to generate fused state features. The trusted fingerprint generation module is used to perform random projection and hash mapping on the fused state features to generate a trusted motion fingerprint; The trusted verification module is used to extract video encoding features from the video data from the buyer, and to calculate the cosine similarity between the video encoding features and the trusted motion fingerprint, and to determine the authenticity of the video data based on the cosine similarity.
[0016] Based on the method for processing sports and health data disclosed in the above embodiments, this embodiment correspondingly discloses an apparatus for processing sports and health data, including: a construction unit, a tensor calculation unit, a trajectory calculation unit, and a scanning unit; The construction unit is used to construct a velocity field based on multi-source sports and health data; the multi-source sports and health data includes at least two of the following: heart rate data from smart wearable devices, GPS trajectory data, and IMU sensor data; The tensor calculation unit is used to calculate the motion dynamics tensor based on the dynamic properties of the velocity field, and generate a two-dimensional discrete flow field. The trajectory calculation unit is used to calculate streamline trajectories based on the two-dimensional discrete flow field; The scanning unit is used to perform sequential scanning along the streamline trajectory to generate a flow field scanning feature sequence.
[0017] Optionally, the building unit includes: The alignment subunit is used to perform multimodal alignment on the multi-source motion health data to obtain unified time-series data; Construct sub-units to build the velocity field based on the unified time-series data.
[0018] Optionally, the building subunit includes: The mapping subunit is used to map the unified time-series data to a two-dimensional spatial grid, so that each grid point corresponds to a velocity vector; the velocity vector contains motion direction and speed information. Generate sub-units for generating the velocity field based on the two-dimensional spatial grid with the velocity vector.
[0019] Optionally, the tensor computation unit includes: The attribute calculation subunit is used to calculate the divergence and curl based on the velocity field to obtain the dynamic attributes; Tensor computation subunit is used to calculate the kinematics tensor based on the divergence and the curl.
[0020] Optionally, the trajectory calculation unit includes: The solution sub-element is used to solve the streamline equations of the two-dimensional discrete flow field through numerical integration, thereby obtaining the streamline trajectory.
[0021] Optionally, the device further includes: An encoding unit is used to encode based on the flow field scanning feature sequence; the update rate of the state during the encoding process is related to the trajectory curvature of the flow field scanning feature sequence; the state includes long-term physical state and short-term action state; The fusion unit is used to fuse the long-term physical state and the short-term action state during the encoding process to generate fused state features.
[0022] Optionally, the encoding unit includes: A curvature calculation subunit is used to calculate the trajectory curvature based on the velocity and acceleration components of the flow field scanning feature sequence. An update subunit is used to adaptively adjust the update rate according to the trajectory curvature; the trajectory curvature is positively correlated with the update rate.
[0023] Optionally, the device further includes: The mapping unit is used to perform random projection and hash mapping on the fused state features to generate a reliable motion fingerprint.
[0024] Optionally, the device further includes: The feature extraction unit is used to obtain the video encoding features of the video data from the buyer during the transaction phase. A consistency calculation unit is used to calculate the cosine similarity between the video coding features and the trusted motion fingerprint; A determining unit is used to determine the authenticity of the video data based on the cosine similarity.
[0025] This application also provides a sports and health data processing system, including functional modules configured to perform the steps of the above methods.
[0026] This application also provides an electronic device, including a processor and a memory, wherein the processor implements the above-described method when executing a program in the memory.
[0027] This application also provides a computer-readable storage medium storing a program that causes a processor to perform the above-described methods.
[0028] This application discloses a method and system for processing sports and health data. By unifying multi-source heterogeneous sports and health data into a two-dimensional flow field, standardized encoding of data from different devices and in different formats is achieved, solving the problem of severe data heterogeneity. Sequential scanning along streamline trajectories ensures that the information propagation direction conforms to the laws of real motion, avoiding spatial neighborhood destruction caused by traditional fixed-direction scanning. By introducing a curvature sensing mechanism, the encoder can adaptively adjust the state update rate according to the intensity of motion, responding quickly at high curvature and retaining data for long periods at low curvature. Simultaneously, a dual-state system captures long-term physical trends and short-term movement details, significantly improving the encoding efficiency and representational ability of long-term time-series data. By generating a unique, stable, and irreversible reliable motion fingerprint, both data verifiability and the seller's original data privacy are guaranteed. By introducing cross-modal cosine similarity verification during the transaction stage, data authenticity can be determined without disclosing the original data, effectively resolving the contradiction between privacy protection and authenticity verification in transactions. Attached Figure Description
[0029] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0030] Figure 1 This is a flowchart illustrating a method for processing sports and health data disclosed in an embodiment of this application; Figure 2 This is a flowchart illustrating another method for processing sports and health data disclosed in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a sports and health data processing system disclosed in an embodiment of this application. Detailed Implementation
[0031] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0032] Example 1: This application discloses a method for processing sports and health data.
[0033] For details, please refer to Figure 1 The method for processing sports and health data disclosed in this embodiment includes the following steps: Step 101: Construct a velocity field based on multi-source sports and health data.
[0034] In the method of this embodiment, the multi-source sports and health data is multi-source heterogeneous data, which may specifically include heart rate data from smart wearable devices (such as wristbands), electrocardiogram signals from heart rate belts, trajectory data from GPS devices, and acceleration and angular velocity data from IMU sensors. Since different devices have different sampling frequencies (e.g., heart rate data is 1Hz, IMU sensor data is 100Hz) and inconsistent spatial reference systems, the multi-source sports and health data can be further aligned in multiple modes to obtain unified time-series data.
[0035] As a feasible solution, multi-source sports and health data can be synchronized by timestamp interpolation and spatial coordinate transformation through a multimodal alignment module, unifying all sports and health data under the same time axis and spatial coordinate system to obtain unified time-series data.
[0036] Subsequently, a velocity field is constructed based on unified time-series data. As a feasible approach, the unified time-series data is mapped onto a two-dimensional spatial grid, so that each grid point corresponds to a velocity vector. The velocity field is then generated based on the two-dimensional spatial grid with velocity vectors. The velocity vectors contain information about the direction and speed of motion.
[0037] Step 102: Calculate the motion dynamics tensor based on the dynamic properties of the velocity field to generate a two-dimensional discrete flow field.
[0038] In the method of this embodiment, dynamic properties including divergence (describing the degree of motion diffusion) and curl (describing the degree of rotational diffusion) can be further introduced on the basis of the velocity field to calculate the motion dynamics tensor. As an feasible approach, the divergence and curl are first calculated based on the velocity field, and the specific formulas are as follows: (1) (2) In the formula, div(v) is the divergence, v is the velocity vector field, and v x and v y denoted as the velocity components in the x and y directions, respectively, and curl(v) is the curl.
[0039] Subsequently, the kinematics tensor is calculated based on divergence and curl, providing physical constraints for subsequent encoding steps. As a feasible approach, the specific formula for calculating the kinematics tensor can be shown below: (3) In the formula, T is the kinematic tensor.
[0040] In the method of this embodiment, divergence and curl are introduced to construct a motion dynamics tensor, which can convert the original time series data into a two-dimensional flow field representation with physical meaning. This not only realizes the unified encoding of multi-source heterogeneous data, but also preserves the dynamic characteristics of motion, providing a physical constraint basis for subsequent authenticity verification.
[0041] Step 103: Calculate streamline trajectories based on the two-dimensional discrete flow field.
[0042] In the method of this embodiment, the streamline equations of the two-dimensional discrete flow field can be solved by numerical integration (such as the fourth-order Runge-Kutta method) to obtain several streamline trajectories. As an feasible solution, the streamline equation formula can be specifically shown as follows: (4) In the formula, r(s) is the streamline trajectory and s is the arc length parameter.
[0043] Step 104: Perform a sequence scan along the streamline trajectory to generate a flow field scan feature sequence.
[0044] In this embodiment, the uniform time-series data is scanned along each streamline trajectory, and the sampled points obtained from the scan are arranged in streamline order to finally form a flow field scanning feature sequence. This application's streamline scanning ensures that the information propagation direction conforms to the actual motion laws. Compared to traditional methods that scan along fixed directions (horizontal, vertical, diagonal, anti-diagonal), this embodiment's method of sequential scanning along streamline trajectories ensures that the information propagation direction conforms to the actual motion laws, avoids damage to the spatial neighborhood, and thus preserves the inherent physical structure of the motion data.
[0045] In the method of this embodiment, as an feasible approach, after obtaining the flow field scanning feature sequence, the flow field scanning feature sequence can be further encoded. Specifically, the flow field scanning feature sequence can be input into a curvature-aware dual-state Mamba encoder, which may specifically include a long-term physical state unit and a short-term motion state unit. The long-term physical state unit is used to record physical trends and training habits, while the short-term motion state unit is used to record instantaneous movements and burst behaviors.
[0046] Specifically, the encoder first calculates the curvature of each point on the streamlined trajectory, and then adaptively adjusts the state update rate based on the magnitude of the trajectory curvature, i.e., the update rate of long-term physical state and short-term action state. Among them, the trajectory curvature is positively correlated with the state update rate. For example, when the trajectory curvature is high (such as when there are sudden behaviors such as sudden stops, changes of direction, and turns), the corresponding state is updated quickly, while when the trajectory curvature is low (such as when there are continuous behaviors such as walking and jogging), the corresponding state is retained for a long time.
[0047] In the method of this embodiment, the trajectory curvature can be calculated based on the velocity and acceleration components of the flow field scanning feature sequence, and the specific formula is as follows: (5) In the formula, k is the trajectory curvature, α x and α y These are the acceleration components in the x and y directions, respectively.
[0048] In the method of this embodiment, ordinary encoders employ a uniform state update strategy, which cannot adapt to drastic changes in motion state. However, this embodiment introduces trajectory curvature as an adjustment factor for the state update rate. For high curvature movements (such as sudden stops, changes of direction, and turns), the state can be updated rapidly to capture transient features. For low curvature movements (such as walking and jogging), the state can be retained in memory for a longer period to preserve physical trends, thereby achieving adaptive encoding for different motion modes.
[0049] As a feasible approach, long-term physical condition and short-term motor function can be updated using the formula shown below: (6) (7) In the formula, Let t represent the long-term physical condition at time t. Let η be the short-term action state at time t. long (k) and η short (k) is the curvature-dependent learning rate function, Δh long Δh is the change in physical condition over a long period. short This refers to the change in short-term action state.
[0050] Subsequently, the curvature-aware dual-state Mamba encoder fuses long-term physical states and short-term motion states during the encoding process to generate fused state features. As a feasible approach, the formula for the fused state features can be expressed as follows: (8) In the formula, H fusion This refers to the characteristics of the fusion state.
[0051] In the method of this embodiment, a dual-state memory system of long-term physical fitness and short-term movement is constructed to record physical fitness trends, training habits and instantaneous movements and explosive behaviors, and after fusion, a comprehensive movement representation is generated, which can retain macroscopic movement patterns and capture microscopic movement details.
[0052] As a feasible approach, after obtaining the fused state features, these features can be further subjected to random projection and hash mapping to generate a trusted motion fingerprint. Specifically, this can be achieved by applying a random projection matrix to the fused state features for dimensionality reduction and obfuscation, followed by generating a fixed-length binary fingerprint through hash mapping. This trusted motion fingerprint possesses the characteristics of uniqueness, stability, and irreversibility. That is, different athletes have different fingerprints, the fingerprints of the same athlete remain stable, and the original data cannot be recovered from the fingerprint. Therefore, it can provide a unique identifier for the data while protecting privacy. Specifically, the formula for generating a trusted motion fingerprint can be shown below: (9) In the formula, F is the trusted motion fingerprint, W is the random projection matrix, and b is the bias term.
[0053] As a feasible solution, after obtaining a trusted motion fingerprint, the authenticity of the data can be verified during the transaction phase using this trusted motion fingerprint. Specifically, this can be achieved by first acquiring video data from the buyer in the transaction, i.e., the motion video provided by the buyer during the transaction phase corresponding to the data to be traded, and extracting the video encoding features from it using a video encoder. Then, the cosine similarity between this video encoding feature and the trusted motion fingerprint is calculated. The authenticity of the buyer's video data is determined based on the cosine similarity, thereby completing the trust verification and transaction. Specifically, the formula for calculating the cosine similarity can be shown below: (10) In the formula, sim(F, F video F represents the cosine similarity. video These are video coding features.
[0054] If the cosine similarity is greater than a preset threshold, the sensor data is determined to be consistent with the video evidence, and the data is considered authentic and reliable. If the cosine similarity is less than or equal to the preset threshold, the data is determined to be forged or abnormal. This verification process does not require the seller to provide the original sensor data; it can be completed solely through fingerprint comparison, effectively protecting the seller's data privacy.
[0055] In the method of this embodiment, by introducing video data provided by a third party (such as motion recordings taken on-site by the buyer) during the transaction stage, and extracting video encoding features and trusted motion fingerprints for cross-modal consistency verification, the authenticity of the data can be effectively verified without the seller providing the original data, thus resolving the contradiction between data privacy protection and authenticity verification in the transaction.
[0056] The method described in this embodiment unifies multi-source heterogeneous motion health data into a two-dimensional flow field, achieving standardized encoding of data from different devices and in different formats, thus solving the problem of severe data heterogeneity. It utilizes the dynamic properties of motion data to construct a motion dynamics representation, converting the original time-series data into a physically meaningful two-dimensional flow field representation, providing a physical constraint basis for subsequent encoding. Simultaneously, sequence sampling along the directional path of the flow field ensures that the information propagation direction conforms to the real motion laws, avoiding the spatial neighborhood destruction caused by traditional fixed-direction scanning and preserving the inherent physical structure of the motion data.
[0057] Furthermore, by introducing a curvature-aware dual-state encoding mechanism, the encoder can adaptively adjust the state update rate according to the intensity of the movement. It can quickly respond to transient features when the curvature is high and maintain the physical fitness trend in the long term when the curvature is low. At the same time, the dual-state system captures the long-term physical fitness trend and short-term movement details respectively, which significantly improves the encoding efficiency and representation ability of long-term time-series data.
[0058] Furthermore, by generating a reliable motion fingerprint that is unique, stable, and irreversible, both data verifiability and the seller's original data privacy are guaranteed. By introducing cross-modal cosine similarity verification during the transaction verification phase, data authenticity can be determined without disclosing the original data, effectively resolving the conflict between privacy protection and authenticity verification in transactions.
[0059] Example 2: This application discloses another method for processing sports and health data. Please refer to [link / reference]. Figure 2 This embodiment describes the entire process of sports and health data processing.
[0060] Step 201: Collect multi-source heterogeneous exercise and health data, and perform multimodal alignment to obtain unified time-series data.
[0061] Step 202: Map the unified time series data to a two-dimensional spatial grid, so that each grid point corresponds to a velocity vector, and generate a velocity field.
[0062] Step 203: Calculate divergence and curl based on the velocity field, and calculate the kinematics tensor based on divergence and curl to generate a two-dimensional discrete flow field.
[0063] Step 204: Solve the streamline equations of the two-dimensional discrete flow field by numerical integration to obtain the streamline trajectories.
[0064] Step 205: Perform a sequence scan along the streamline trajectory to generate a flow field scan feature sequence.
[0065] Step 206: Encode based on flow field scanning feature sequence, adaptively adjust the state update rate, and fuse long-term physical state and short-term action state to generate fused state features.
[0066] Step 207: Perform random projection and hash mapping on the fused state features to generate a reliable motion fingerprint.
[0067] Step 208: During the transaction phase, extract the video coding features from the video data from the buyer and calculate the cosine similarity between the video coding features and the trusted motion fingerprint.
[0068] Step 209: Determine if the cosine similarity is greater than a preset threshold. If yes, proceed to step 210. If no, proceed to step 211.
[0069] Step 210: Determine the authenticity and reliability of the data.
[0070] Step 211: Determine if the data is forged or abnormal.
[0071] Example 3: This application discloses another application scenario for a method of processing sports and health data. Specifically, athlete A wants to sell his / her week's running training data (including heart rate, GPS trajectory, and IMU data) on a data trading platform. First, athlete A's data is synchronized in time and space by a multimodal alignment module to obtain unified time-series data. A two-dimensional flow field construction module constructs a velocity field based on the GPS trajectory and calculates divergence and curl to generate a two-dimensional flow field. A flow field scanning module performs sequential scanning along streamlines to generate a flow field scanning feature sequence. A curvature-aware dual-state Mamba encoder adaptively adjusts the state update according to the curvature at different times during the run, fusing long-term physical state and short-term movement state to generate fused state features. A trusted fingerprint generation module generates a trusted motion fingerprint for athlete A and stores it on the blockchain.
[0072] Buyer B wishes to verify the authenticity of the data before making the purchase. Buyer B provides a short video of athlete A running at the same location, which they observed. The system extracts the video encoding features from the video and calculates the cosine similarity between this video and a trusted motion fingerprint on the blockchain. If the cosine similarity is greater than a preset threshold, the verification passes, Buyer B is convinced of the data's authenticity, and the transaction is completed. Throughout the process, seller A never discloses the original sensor data; buyer B completes the verification solely through fingerprint comparison, ensuring the privacy and rights of both parties are protected.
[0073] Example 4: This application discloses another application scenario for a method of processing sports and health data. Specifically, a malicious user C attempts to forge a marathon training record for a transaction. During the two-dimensional flow field construction stage, the forged data cannot form a physically consistent flow field structure due to physical constraints of divergence and curl. During the streamline scanning stage, the feature sequence obtained by sampling along the streamlines of the forged data differs significantly from the statistical distribution of the real motion pattern. During the fingerprint generation stage, the cosine similarity between the credible motion fingerprint generated by the forged data and the real motion video is far less than the threshold, resulting in verification failure and transaction rejection.
[0074] Based on the above-mentioned method for processing sports and health data, this application also discloses a system for processing sports and health data, used to execute the above-mentioned method. The system includes: a multimodal alignment module 301, a two-dimensional flow field construction module 302, a flow field scanning module 303, a curvature-sensing dual-state encoder 304, a trusted fingerprint generation module 305, and a trusted verification module 306. The multimodal alignment module 301 is used to acquire multi-source sports and health data and to perform multimodal alignment on the multi-source sports and health data. The two-dimensional flow field construction module 302 is used to construct a velocity field and introduce dynamic properties to construct a two-dimensional discrete flow field. The flow field scanning module 303 is used to calculate streamline trajectories and perform sequential scanning along the streamline trajectories to generate a flow field scanning feature sequence. The curvature-aware dual-state encoder 304 is used to adaptively adjust the state update rate according to the trajectory curvature and fuse long-term physical state and short-term action state to generate fused state features. The trusted fingerprint generation module 305 is used to perform random projection and hash mapping on the fused state features to generate a trusted motion fingerprint. The trusted verification module 306 is used to extract video encoding features from the video data from the buyer, and to calculate the cosine similarity between the video encoding features and the trusted motion fingerprint, and to determine the authenticity of the video data based on the cosine similarity.
[0075] Based on the method for processing sports and health data disclosed in the above embodiments, this embodiment correspondingly discloses an apparatus for processing sports and health data, including: a construction unit, a tensor calculation unit, a trajectory calculation unit, and a scanning unit; The construction unit is used to construct a velocity field based on multi-source sports and health data; the multi-source sports and health data includes at least two of the following: heart rate data from smart wearable devices, GPS trajectory data, and IMU sensor data; The tensor calculation unit is used to calculate the motion dynamics tensor based on the dynamic properties of the velocity field, and generate a two-dimensional discrete flow field. The trajectory calculation unit is used to calculate streamline trajectories based on the two-dimensional discrete flow field; The scanning unit is used to perform sequential scanning along the streamline trajectory to generate a flow field scanning feature sequence.
[0076] Optionally, the building unit includes: The alignment subunit is used to perform multimodal alignment on the multi-source motion health data to obtain unified time-series data; Construct sub-units to build the velocity field based on the unified time-series data.
[0077] Optionally, the building subunit includes: The mapping subunit is used to map the unified time-series data to a two-dimensional spatial grid, so that each grid point corresponds to a velocity vector; the velocity vector contains motion direction and speed information. Generate sub-units for generating the velocity field based on the two-dimensional spatial grid with the velocity vector.
[0078] Optionally, the tensor computation unit includes: The attribute calculation subunit is used to calculate the divergence and curl based on the velocity field to obtain the dynamic attributes; Tensor computation subunit is used to calculate the kinematics tensor based on the divergence and the curl.
[0079] Optionally, the trajectory calculation unit includes: The solution sub-element is used to solve the streamline equations of the two-dimensional discrete flow field through numerical integration, thereby obtaining the streamline trajectory.
[0080] Optionally, the device further includes: An encoding unit is used to encode based on the flow field scanning feature sequence; the update rate of the state during the encoding process is related to the trajectory curvature of the flow field scanning feature sequence; the state includes long-term physical state and short-term action state; The fusion unit is used to fuse the long-term physical state and the short-term action state during the encoding process to generate fused state features.
[0081] Optionally, the encoding unit includes: A curvature calculation subunit is used to calculate the trajectory curvature based on the velocity and acceleration components of the flow field scanning feature sequence. An update subunit is used to adaptively adjust the update rate according to the trajectory curvature; the trajectory curvature is positively correlated with the update rate.
[0082] Optionally, the device further includes: The mapping unit is used to perform random projection and hash mapping on the fused state features to generate a reliable motion fingerprint.
[0083] Optionally, the device further includes: The feature extraction unit is used to obtain the video encoding features of the video data from the buyer during the transaction phase. A consistency calculation unit is used to calculate the cosine similarity between the video coding features and the trusted motion fingerprint; A determining unit is used to determine the authenticity of the video data based on the cosine similarity.
[0084] This application also provides a sports and health data processing system, including functional modules configured to perform the steps of the above methods.
[0085] This application also provides an electronic device, including a processor and a memory, wherein the processor implements the above-described method when executing a program in the memory.
[0086] This application also provides a computer-readable storage medium storing a program that causes a processor to perform the above-described methods.
[0087] The embodiments in this specification are described in a progressive manner. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant details can be found in the method section.
[0088] It should also be noted that, in this application, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0089] The steps of the methods or algorithms described in conjunction with the embodiments disclosed in this application can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.
[0090] The features described in the embodiments of this specification can be substituted for or combined with each other, so that those skilled in the art can implement or use this application.
[0091] The foregoing description and accompanying drawings fully illustrate embodiments of this disclosure to enable those skilled in the art to practice them. Other embodiments may include structural, logical, electrical, procedural, and other changes. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the order of operation may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. Furthermore, the terminology used in this application is for descriptive purposes only and is not intended to limit the technical solutions recorded in this application. As used in this application, the term "and / or" refers to any and all possible combinations of one or more associated listed terms. In this application, each embodiment may emphasize differences from other embodiments, and similar or identical parts between embodiments may be referred to mutually. For methods, products, etc., disclosed in the embodiments, if they correspond to the method section of the embodiments, the relevant parts may be referred to the description of the method section.
[0092] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented using electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments disclosed herein. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0093] The methods and products (including but not limited to devices and equipment) disclosed in the embodiments of this application can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units may only be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the shown or discussed units may be through some interfaces, and the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to implement this embodiment according to actual needs. In addition, the functional units in the embodiments of this disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0094] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.
[0095] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for processing sports and health data, characterized in that, include: A velocity field is constructed based on multi-source sports and health data; The multi-source sports and health data includes at least two of the following: heart rate data from smart wearable devices, GPS trajectory data, and IMU sensor data. The motion dynamics tensor is calculated based on the dynamic properties of the velocity field to generate a two-dimensional discrete flow field; Calculate streamline trajectories based on the aforementioned two-dimensional discrete flow field; A sequence scan is performed along the streamline trajectory to generate a flow field scanning feature sequence.
2. The method according to claim 1, characterized in that, The construction of the velocity field based on multi-source sports and health data includes: Multimodal alignment is performed on the multi-source sports and health data to obtain unified time-series data; The velocity field is constructed based on the unified time-series data.
3. The method according to claim 2, characterized in that, The construction of the velocity field based on the unified time-series data includes: The unified time-series data is mapped to a two-dimensional spatial grid, so that each grid point corresponds to a velocity vector; the velocity vector contains motion direction and speed information. The velocity field is generated based on the two-dimensional spatial grid containing the velocity vector.
4. The method according to claim 1, characterized in that, The calculation of the motion dynamics tensor based on the dynamic properties of the velocity field includes: The divergence and curl are calculated based on the velocity field to obtain the dynamic properties; The kinematic tensor is calculated based on the divergence and the curl.
5. The method according to claim 1, characterized in that, The calculation of streamline trajectories based on the two-dimensional discrete flow field includes: The streamline trajectory is obtained by solving the streamline equation of the two-dimensional discrete flow field through numerical integration.
6. The method according to claim 1, characterized in that, After generating the flow field scanning feature sequence, the method further includes: Encoding is performed based on the flow field scanning feature sequence; the update rate of the state during the encoding process is related to the trajectory curvature of the flow field scanning feature sequence; the state includes long-term physical state and short-term action state; During the encoding process, the long-term physical state and the short-term action state are fused to generate fused state features.
7. The method according to claim 6, characterized in that, The encoding based on the flow field scanning feature sequence includes: The trajectory curvature is calculated based on the velocity and acceleration components of the flow field scanning feature sequence. The update rate is adaptively adjusted based on the trajectory curvature; the trajectory curvature is positively correlated with the update rate.
8. The method according to claim 6, characterized in that, After generating the fusion state features, the method further includes: The fused state features are randomly projected and hashed to generate a trusted motion fingerprint.
9. The method according to claim 8, characterized in that, After generating a trusted motion fingerprint, the method further includes: During the transaction phase, the video encoding characteristics of the video data from the buyer are obtained; Calculate the cosine similarity between the video coding features and the trusted motion fingerprint; The authenticity of the video data is determined based on the cosine similarity.
10. A system for processing sports and health data, characterized in that, The system for performing the method according to any one of claims 1-9 comprises: a multimodal alignment module, a two-dimensional flow field construction module, a flow field scanning module, a curvature-sensing dual-state encoder, a trusted fingerprint generation module, and a trusted verification module; The multimodal alignment module is used to acquire multi-source sports and health data and to perform multimodal alignment on the multi-source sports and health data. The two-dimensional flow field construction module is used to construct a velocity field and introduce dynamic properties to construct a two-dimensional discrete flow field. The flow field scanning module is used to calculate streamline trajectories and perform sequential scanning along the streamline trajectories to generate a flow field scanning feature sequence. The curvature-aware dual-state encoder is used to adaptively adjust the state update rate according to the trajectory curvature and fuse long-term physical state and short-term action state to generate fused state features. The trusted fingerprint generation module is used to perform random projection and hash mapping on the fused state features to generate a trusted motion fingerprint; The trusted verification module is used to extract video encoding features from the video data from the buyer, and to calculate the cosine similarity between the video encoding features and the trusted motion fingerprint, and to determine the authenticity of the video data based on the cosine similarity.
