A method for managing user entry and exit of a store

Abstract

The present invention relates to the technical field of user data management, and in particular to a method for managing users entering and exiting a store. The method comprises the following steps: obtaining user bracelet data, and performing user device authentication based on the user bracelet data to obtain user device authentication data; collecting user image data based on the user device authentication data to obtain user image data, and extracting user behavior features based on the user image data to obtain user behavior feature data; obtaining user indoor location data, and performing location mapping based on the user behavior feature data and the user indoor location data to obtain user behavior location mapping data; tracking entry and exit behaviors based on the user behavior location mapping data to obtain user entry and exit behavior data, so as to perform user entry and exit store management operations. The present invention improves the response speed and analysis accuracy of the system through senseless identity authentication and multi-dimensional behavior feature analysis, so that users can enjoy a convenient experience when entering the store.

Description

Technical Field

[0001] The present invention relates to the technical field of user data management, and in particular to a method for managing users entering and exiting a store. Background Art

[0002] Traditional foot spa access management mainly relies on manual records or access cards, front desk registration and other means. These methods are inefficient and prone to misjudgment or omissions, making it difficult to meet modern management needs. In addition, traditional management methods cannot analyze customers' activities and behaviors in the foot spa in real time, thus limiting the provision of individual services and the optimization of spatial layout. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a method for managing users entering and exiting a store to solve at least one of the above technical problems.

[0004] The present application provides a method for managing user entry and exit of a store, the method comprising:

[0005] S1. Control the wristband communication device to exchange user wristband data within the sensing range to obtain user wristband data, and perform user device authentication according to the user wristband data to obtain user device authentication data;

[0006] S2. Collecting user image data according to the user device authentication data to obtain user image data, and extracting user behavior features according to the user image data to obtain user behavior feature data;

[0007] S3, obtaining the user's indoor location data, and performing location mapping according to the user's behavior feature data and the user's indoor location data to obtain user behavior location mapping data;

[0008] S4. Track the entry and exit behavior according to the user behavior location mapping data to obtain the user's entry and exit behavior data to perform user entry and exit management operations.

[0009] In the present invention, user device authentication is performed through a wristband communication device, and each user can be quickly and accurately identified, which reduces the time of identity authentication and avoids the error of manual verification. The wristband data exchange does not require manual operation by the user, making the identification process more convenient and suitable for places with high traffic. After the user device is authenticated, the user image data is collected by the camera, and the user's body shape, posture, gait and other behavioral characteristics are combined to accurately describe the user characteristics, and the fusion judgment is performed based on the user characteristics and the wristband data, so as to reduce the misjudgment caused by different users using the same wristband. The location mapping data can reflect the user's activity trajectory, so that merchants can identify high-frequency paths and hot zone distribution, and provide a basis for the layout optimization of shops. Location mapping can track the location of specific personnel in large-scale crowd monitoring, identify suspicious or improper behavior in time, and improve the safety control capabilities in the store. Through the behavior location mapping data, the specific time, path and behavior pattern of the user entering and leaving the store can be recorded in detail, which can not only improve management efficiency, but also accumulate long-term entry and exit records, and provide data support for analyzing passenger flow, activity areas, etc.

[0010] Optionally, S1 includes:

[0011] S11, detecting the sensing area and establishing a temporary connection through the wristband communication device to obtain sensing temporary connection data;

[0012] S12, exchanging wristband data according to the induction temporary connection data to obtain the user's wristband data;

[0013] S13: Transmit the user's wristband data to the edge computing node for verification and comparison to obtain user device authentication data.

[0014] In the present invention, this step realizes contactless automatic identity recognition. The user does not need to actively swipe the card or operate. The system can automatically detect the bracelet device through the sensing area and quickly establish a temporary connection, which improves the user's entry and exit experience, especially in the case of high traffic, and can avoid queuing and congestion caused by manual card swiping or code scanning. Through the induction of temporary connection, the system can exchange data with the bracelet device. This exchange mechanism realizes a real-time, low-power communication method, ensuring that the user's identity information can be collected instantly when the user passes through the sensing area, and enhancing the real-time and reliability of data acquisition. The user's bracelet data is transmitted to the edge computing node for real-time verification and comparison. The introduction of edge computing greatly reduces the dependence on the central server. The edge computing node can locally process user identity authentication requests, thereby reducing delays, so that user device authentication can be completed in almost real time. Through the distributed processing architecture of the edge computing node, even if there are multiple entrances or multiple sensing areas in the store, the data processing speed will not be affected by multi-point collection, thereby improving the scalability and concurrent processing capabilities of the system.

[0015] Optionally, S2 includes:

[0016] S21, activating the camera and collecting user image data according to the user device authentication data to obtain the user image data;

[0017] S22, extracting user body shape data according to the user image data to obtain user body shape data;

[0018] S23, extracting user posture features according to the user body shape data to obtain user posture feature data;

[0019] S24: Integrate the user's body shape data and the user's posture feature data to obtain user behavior feature data.

[0020] The activation of the camera in the present invention is based on device authentication, ensuring that image acquisition is performed only with user authorization, meeting the requirements of privacy protection, and ensuring the security of data during the acquisition and storage process. By analyzing the user's body shape characteristics, the system's recognition accuracy of the user can be enhanced. Especially in the scenario of frequent entry and exit of the store, body shape data can assist in identity recognition and effectively improve security. Posture feature data can be updated in real time, so that the system can identify the user's current status, which is very beneficial for improving the response speed of in-store services. For example, the system can identify users waiting to check out so that the clerk can provide corresponding services. By integrating the user's body shape data and posture feature data, the user's comprehensive behavioral feature data is formed, and a complete user behavior portrait is generated, covering the user's appearance and action information, which can more accurately describe the user's behavior pattern and provide a comprehensive basis for shops to analyze users.

[0021] Optionally, the user posture feature extraction includes:

[0022] Perform key point detection according to the user's body shape data to obtain the user's body shape key point data;

[0023] Perform skeleton marking according to the key point data of the user's body shape to obtain the user's skeleton marking data;

[0024] Coordinates are normalized and motion detection is triggered according to the user skeleton marking data to obtain user motion detection data;

[0025] Performing posture category recognition based on user motion detection data to obtain user posture category data;

[0026] Calculate the limb angle according to the user's posture category data to obtain the user's limb angle data;

[0027] Calculate the posture change rate according to the user's limb angle data to obtain the user's posture change rate data;

[0028] Performing posture symmetry analysis based on the user posture change rate data to obtain the user posture symmetry data;

[0029] Performing posture stability adjustment according to the user posture change rate data to obtain user posture stability adjustment data;

[0030] Perform posture dynamic amplitude measurement according to user posture symmetry data and user posture stability adjustment data to obtain user posture dynamic amplitude measurement data;

[0031] A posture feature vector is generated according to the user posture category data, the user posture change rate data and the user posture dynamic amplitude measurement data to obtain the user posture feature data.

[0032] In the present invention, through key point detection based on user body data, the system can accurately identify the user's body parts (such as head, shoulders, knees, ankles, etc.), and construct the user's posture skeleton model through skeleton marking, which can not only accurately depict the user's static posture, but also can be used for real-time dynamic posture recognition. Through coordinate standardization, the data deviation caused by differences in user height and other factors can be eliminated, ensuring that the posture analysis of different users is consistent and comparable. At the same time, the triggering of motion detection can capture the user's dynamic changes in real time, ensuring that the system can adjust and respond instantly when the user moves. Angle calculation can provide specific limb feature data, which is convenient for the system to more accurately identify complex postures (such as squatting, bending, stretching, etc.), and further increase the analysis dimension. The posture change rate reflects the speed of change of the user's posture, and can further distinguish different posture modes (such as fast walking, slow walking, etc.). The calculation of the change rate helps to identify the urgency or comfort of the user's behavior and provide support for real-time decision-making. Symmetry analysis can identify the balance of the user's left and right limbs, such as whether the user has a one-side center of gravity offset, thereby mining the uniqueness of user data in different dimensions to achieve more in-depth data analysis. Through stability adjustment, the system can remove short-term posture jitter or abnormal changes, making the posture data more stable. This ensures the accuracy of the analysis and avoids identifying temporary jitter as unstable posture. Through posture dynamic amplitude measurement, it is possible to quantify the amplitude of the user's specific movements (such as step width, arm swing amplitude, etc.), further understand the user's movement state, and help determine the user's movement intensity and mode (such as easy walking or fast walking).

[0033] Optionally, the gesture category recognition includes:

[0034] Perform dynamic and static posture classification according to the user motion detection data to obtain the user's dynamic and static posture data;

[0035] Perform posture turning point recognition based on the user's dynamic and static posture data to obtain posture turning point data;

[0036] Perform preliminary screening of hierarchical classification based on posture turning point data to obtain posture hierarchical classification data;

[0037] Detailed features are extracted according to the posture hierarchical classification data to obtain posture detail feature data;

[0038] Perform multi-level behavior classification based on posture detail feature data and posture hierarchical classification data to obtain posture multi-level behavior data;

[0039] Divide the action sequence according to the multi-level behavior data of the posture to obtain the posture behavior action sequence data;

[0040] Extracting posture transition data according to posture behavior action sequence data to obtain posture transition data;

[0041] Matching is performed based on the posture transition data and the preset posture category template data to obtain user posture category data.

[0042] In the present invention, the user's motion detection data is first classified into dynamic and static categories, which can quickly distinguish whether the user is currently stationary or in motion. In this way, unnecessary data processing can be effectively screened out, so that the system reduces the amount of calculation when processing the static state, and focuses on the detailed analysis in the dynamic state. By identifying the turning points of the user's actions (such as the transition point from standing to walking, the point from walking to stopping, etc.), the important nodes in the user's behavior can be accurately identified. Through the preliminary screening of the posture hierarchical classification, the user's actions can be roughly divided into categories of different levels, such as the basic "walking", "standing", "sitting", etc., which can provide data structure support for the extraction of detailed features. Further extracting the detailed features in the posture (such as the angle, amplitude, speed, etc. of the posture) can more carefully describe the user's specific action mode, thereby identifying the slight differences in the action, such as distinguishing "normal walking" from "fast walking", etc., laying the foundation for more accurate behavior analysis. By classifying the posture feature data at multiple levels, the user's behavior category can be further refined. For example, the basic level "walking" posture can be divided into "slow walking", "fast walking", etc. after segmentation, which enhances the depth of the system's behavior analysis and can more accurately adapt to the needs of different scenarios. Dividing the user's continuous actions into action sequences can capture the overall process of the user's behavior. Through posture transition analysis, the transformation features between action sequences can be captured (such as the transition from walking to sitting, the transition from standing to walking, etc.). The transition features help to further understand the user's behavioral intentions, such as whether the user is ready to leave, whether he is looking for something, etc. By matching the posture transition data with the preset posture category template data, the user's specific posture category can be quickly and accurately identified. The matching method not only improves the efficiency of recognition, but also ensures the consistency of classification.

[0043] Optionally, the dynamic and static posture classification includes:

[0044] Extracting acceleration features and speed features based on the user motion detection data to obtain user motion acceleration feature data and user motion feature data;

[0045] Preliminarily classify the user motion detection data according to the user motion acceleration feature data and the user motion feature data to obtain primary user dynamic and static posture data;

[0046] Perform multi-joint consistency analysis and center of gravity stability analysis based on the user motion detection data to obtain the user motion multi-joint consistency data and the user motion center of gravity stability data respectively;

[0047] Perform secondary classification on the primary user's dynamic and static posture data according to the user's motion multi-joint consistency data and the user's motion center of gravity stability data to obtain secondary user's dynamic and static posture data;

[0048] Performing time series division on the user motion detection data to obtain user motion time series division data;

[0049] Calculate the activity change rate based on the user's motion time series data to obtain the user's motion activity change rate data;

[0050] Perform energy calculation based on the user's sports activity change rate data to obtain the user's sports activity change rate energy data;

[0051] Energy-weighted dynamic and static posture classification is performed on the secondary user dynamic and static posture data according to the preset time series division parameter data corresponding to the user movement time series division data and the energy data of the user movement activity change rate to obtain the user dynamic and static posture data.

[0052] In the present invention, by extracting acceleration and speed features from motion detection data, the system can accurately capture the user's motion intensity and speed changes. Acceleration features help distinguish static and dynamic behaviors, while speed features can reflect the user's motion state, such as walking, running, etc. By preliminarily classifying the motion data through acceleration and speed features, most static and dynamic behaviors can be quickly distinguished. By analyzing the motion consistency of each joint of the user, the system can identify the user's motion coordination and stability. For example, when walking, the legs and arms should move in coordination, while the consistency between joints is high in static posture, further confirming the error in the primary classification. Analyzing the change in the user's center of gravity can help identify static and dynamic postures. For example, the user's center of gravity is basically stable in static posture, while the user's center of gravity moves within a certain range in dynamic posture. The center of gravity data enables the system to more accurately identify whether the user is completely still or has a small movement. Dividing the user's motion data by time allows the system to process the data in segments and calculate the activity intensity in different time periods. By calculating the rate of change of activity, the system can capture the rate of change of the user's motion state (such as the rate change from fast walking to slow walking). Based on the activity change rate data, the user's exercise energy data is calculated to evaluate the user's exercise intensity from a physical perspective. Through energy data, the system can better distinguish between high-energy activities (such as brisk walking and running) and low-energy activities (such as standing and sitting). Combining multi-joint consistency, center of gravity stability, activity change rate and energy data can significantly improve the accuracy of dynamic and static classification. Through energy-weighted dynamic and static classification, the system can better cope with small posture changes, thereby reducing misclassification.

[0053] Optionally, the posture change rate calculation includes:

[0054] Perform time series segmentation according to the user's limb angle data to obtain user limb angle segmentation data;

[0055] Calculating the long-term and short-term change rates of the user's limb angle segmentation data to obtain limb angle long-term and short-term change rate data, wherein the limb angle long-term and short-term change rate data includes limb angle long-term change rate data and limb angle short-term change rate data;

[0056] The fluctuation of the change rate is calculated based on the long-term and short-term change rate data of the limb angle, and the fluctuation data of the change rate of the limb angle is obtained;

[0057] According to the short-term change rate data of the limb angle, the change rate directionality analysis is performed to obtain the limb angle change rate directionality data;

[0058] According to the short-term change rate data of the limb angle, short-term change rate classification is performed to obtain the short-term change rate classification data of the limb angle;

[0059] The user posture change rate data is obtained by performing feature vector conversion based on the long-term and short-term change rate data of the limb angle, the volatility data of the limb angle change rate, the directional data of the limb angle change rate and the classification data of the short-term change rate of the limb angle.

[0060] In the present invention, the user's limb angle data is segmented in time series, and the changes in the user's movements can be analyzed according to different time periods. Through segmented processing, the system can capture the changes in the user's posture in a relatively short time, thereby improving the accuracy of data processing. Through the calculation of long-term and short-term change rates, the system can identify the changing trends of the user's posture at different time scales. The long-term change rate can reveal the overall trend of the user's movements, such as the stability of standing and walking; the short-term change rate reflects the instantaneous dynamic changes, such as the speed of the steps and the frequency of the arm swing. By analyzing the volatility of the limb angle change rate, the stability or jitter of the movement can be quantified. High volatility usually reflects unstable movements (such as irregular walking or sudden pauses), while low volatility represents more stable movements (such as continuous standing or regular steps). By performing directional analysis on the short-term change rate of the limb angle, the system can identify the directional characteristics of the movement (such as rising, falling, sideways, etc.), which is suitable for identifying the regularity and directional changes of movements, and is suitable for identifying complex movements, such as the up and down swinging of the arms. By classifying the short-term rate of change data, the system can classify different types of actions into categories such as "fast", "slow" or "medium", allowing the system to quickly distinguish different short-term behavioral characteristics, thereby improving recognition speed. By integrating the long-term and short-term rate of change, volatility, directionality and short-term rate of change classification data of limb angles into a posture change rate feature vector, the system can generate comprehensive data containing various characteristics of the user's actions.

[0061] Optionally, S3 includes:

[0062] S31, obtaining user real-time location data through an indoor positioning system;

[0063] S32: Perform multi-source positioning data fusion according to the user's real-time location data to obtain the user's indoor location data.

[0064] S33: performing association matching based on the user behavior feature data and the user indoor location data to obtain user behavior location mapping data.

[0065] In the present invention, an indoor positioning system (such as Bluetooth beacon, Wi-Fi positioning, ultra-wideband UWB, etc.) is used to obtain the user's location data in real time in indoor environments such as shops and venues. The real-time positioning can accurately determine the user's specific location, so that the system can continuously track the user's dynamics and provide support for real-time response and services. For example, when the user moves in different areas such as the lobby, massage area, and tea area, the system can accurately identify the user's location and update it in real time when the user changes location. Multi-source positioning data fusion refers to the integration of data from different positioning sources (such as Bluetooth, Wi-Fi, UWB, etc.) to obtain a more accurate and stable user location. Fusion positioning can make up for the limitations of a single positioning system in certain scenarios. For example, when the Bluetooth signal is blocked, Wi-Fi data can be used as a supplement to improve the robustness of the system. The user's behavioral feature data is associated and matched with the indoor location data to generate user behavior location mapping data, which can not only show the user's behavior pattern at a specific location, but also help understand the user's intention. For example, if the system detects that the user stays in the product display area for a long time and has observation actions, it can be inferred that the user is interested in the products in the area. Behavioral location mapping data can support the identification of users' preferred areas within the foot spa. For example, if a user stays in the physiotherapy area for a long time, the system can infer that they are interested in physiotherapy projects. Based on this data, the system can push relevant service recommendations (such as new therapies, discounts, etc.) at the right time to achieve precision marketing.

[0066] Optionally, S4 includes:

[0067] S41, tracking the user behavior trajectory according to the user behavior location mapping data to obtain user behavior trajectory data;

[0068] S42, performing user path analysis according to the user behavior trajectory data to obtain user path analysis data;

[0069] S43. Extracting the user's entry and exit behavior based on the user path analysis data to obtain the user's entry and exit behavior data for user entry and exit management.

[0070] The behavior trajectory tracking in the present invention can capture the specific path of the user in the store, including the resident position, moving direction, speed change, etc., so that the system can conduct in-depth analysis of the user's behavior pattern. For example, it can identify whether the user has frequent stops, detours, etc. The system can dynamically capture the user's trajectory changes and track the user's path and behavior status. For example, the system can identify the user's moving speed, residence time, and switching frequency between different areas. Through the trajectory data, the system can identify the user's behavioral characteristics and activity patterns to provide support for operation management. Path analysis can identify the user's activity preferences in specific areas. Through path analysis data, the system can identify high-frequency visited areas (such as physiotherapy areas, massage areas) and less visited areas, thereby inferring the user's interest preferences. Through path analysis data, the system can accurately identify the user's entry and exit behavior, including the time, path, and visited areas of entering and leaving the store. The system monitors the customer's entry and exit records throughout the process to achieve seamless entry and exit management. The system can arrange services and allocate resources according to the customer's entry and exit records to ensure efficient customer management. Through path analysis data, the system can accurately identify the user's entry and exit behavior, including the time of entry and exit, the path, and the area visited, and support the system to monitor the customer's entry and exit records throughout the process to achieve seamless entry and exit management. The system can arrange services and allocate resources based on the customer's entry and exit records to ensure efficient customer management.

[0071] Optionally, the user path analysis includes:

[0072] Perform path segmentation according to user behavior trajectory data to obtain user behavior path segmentation data;

[0073] Calculate curvature and speed characteristics based on user behavior path segmentation data to obtain path curvature data and path speed characteristic data respectively;

[0074] Perform stay point detection based on user behavior path segmentation data to obtain user stay point data;

[0075] Perform path morphology analysis based on path curvature data, path speed characteristic data, and user stop point data to obtain user behavior path morphology data;

[0076] Perform hot spot area detection and area switching frequency calculation based on user behavior path morphology data to obtain user behavior hot spot area data and user behavior area switching frequency data;

[0077] Generate a typical user path based on the path curvature data, path speed feature data, user behavior hotspot area data, and user behavior area switching frequency data to obtain typical user path data;

[0078] User path analysis data is integrated based on path curvature data, path speed feature data, user behavior hotspot area data, user behavior area switching frequency data, and typical user path data to obtain user path analysis data.

[0079] The path segmentation in the present invention can divide the user behavior trajectory into smaller units, helping the system to analyze the specific behavior characteristics of the user in each section. Through path segmentation, the system can perform a more detailed analysis of the user's behavior at different stages (such as the path from the massage area to the tea area, the stay in the rest area, etc.). Curvature calculation can reveal the turning and curve characteristics of the user's path. A larger curvature usually indicates that the user is turning or detouring, while a smaller curvature reflects straight walking. Through curvature data, the system can identify the user's walking pattern in a specific area, for example, frequent turns indicate that the user is browsing products. Speed ​​feature data can reveal the user's moving speed in the path and identify fast-walking and slow-moving areas. A lower speed indicates that the user is browsing or resting, while a higher speed indicates that the user has a clear goal, such as going to a designated area. By identifying the stop points in the user's path, the system can determine the user's points of interest in a specific area. For example, a user's longer stay in the rest area means that the services in that area are more popular with him;

[0080] Stop point data provides strong support for space optimization and personalized recommendations. Path morphology analysis combines curvature, speed characteristics and stop point data to reveal user behavior patterns. For example, the system can identify typical path patterns of users, such as slow walking, pauses, frequent turns, etc. Different path patterns can reflect user behavior patterns in different areas (such as browsing, goal orientation, leisure mode, etc.). Hot spot area data can guide shops to optimize resource allocation, such as adding waiters or related facilities in high-frequency areas, thereby improving users' service experience;

[0081] The area switching frequency data can analyze the continuity of user behavior. For example, the continuous movement from the massage area to the health area indicates that the user is experiencing a full set of experiences. The system can guide the path based on this data to improve the consistency of the user experience. By generating typical user paths, the system can summarize different types of user behavior patterns. For example, the system can identify the typical paths of "leisure" users and "purpose-oriented" users to help shops better understand the needs of different customer groups. Typical path data can also provide reference for waiters to guide them to provide more considerate services to different types of users.

[0082] The purpose of the present invention is to automatically identify the user's identity through the wristband data exchange system when the user enters the store, and realize non-sensing authentication. There is no need for manual registration or code scanning, and the user's store entry process is smoother. For places with high traffic, this process greatly reduces the waiting time in line and improves the efficiency of entering the store. The system generates user behavior feature data based on the user's body shape, posture, movement and other characteristics through image data acquisition. The data not only includes the user's body parameters, but also captures the user's posture changes, such as standing, walking, sitting, etc., on the one hand, realizing security identification based on user characteristics, and on the other hand, providing data support for user services. High-precision indoor positioning is achieved through the fusion of multi-source data such as Bluetooth beacons, Wi-Fi and UWB. User behavior location mapping data can support spatial analysis of behavior data. For example, the system can identify the typical behavior and preferred areas of users in different areas. The system can identify the user's behavior pattern through path data. For example, the user's stay time, walking path, and area switching frequency in different areas can show their specific behavior preferences. By integrating user trajectory and path analysis data, the system can accurately judge the user's entry and exit behavior. The system can not only identify the user's entry and exit time and frequency, but also provide dynamic entry and exit data based on the user's stay time in the store, visited areas, peak entry and exit times, etc. Real-time monitoring allows managers to understand the flow of people in the store in a timely manner and provide customers with smoother services. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings:

[0084] Figure 1 A flowchart showing the steps of a method for managing user entry and exit of a store according to an embodiment is shown;

[0085] Figure 2 A flowchart showing a method for authenticating a user device according to an embodiment of the present invention is provided;

[0086] Figure 3 A flowchart showing a method for extracting user behavior features according to an embodiment of the present invention is shown;

[0087] Figure 4 A flowchart showing the steps of a user behavior location mapping method according to an embodiment is shown;

[0088] Figure 5 A flowchart showing a method for tracking user entry and exit behavior according to an embodiment of the present invention is provided;

[0089] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0090] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0091] In addition, the drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0092] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0093] The foot massage city has installed a wristband recognition system, camera, indoor positioning system and behavior analysis system to provide customers with timely, effective and comfortable services. In a day, the foot massage city has 200 customers, and the system optimizes customer access, service experience and product service recommendations.

[0094] When a customer enters the store, the bracelet he wears will establish a temporary connection with the bracelet communication device in the entrance sensing area. The system identifies the customer through the bracelet and performs device authentication. When the customer enters the foot bath city, the bracelet device ID6789 is recognized at the entrance. The system establishes a connection with the bracelet within 2 seconds and starts device authentication. The system verifies the bracelet ID through the edge computing node and compares the customer information. The verification success rate is 99% and the verification time is 0.5 seconds. The system identifies the customer as a new member B and stores the customer's membership level, last consumption record and other information. After the authentication is completed, the system records the customer's entry time (such as 14:02 PM) and transmits the information to the central database. The system completes the identity authentication when the customer enters the store and stores the customer's entry time and basic information. The non-sensing authentication process does not require customers to register manually, which is convenient and efficient.

[0095] After the customer enters the store, the system automatically activates the camera and collects the customer's image data. The system extracts the customer's body shape and posture information through the image data to support further authentication services and specific product service provision. The system automatically activates the camera and captures the image data of customer B. For example, if customer B is 170cm tall and weighs 65kg, the system recognizes the body shape data and compares it with the preset data template to identify that customer B is suitable for moderate massage. The system also monitors the customer's posture characteristics. According to the image data analysis, customer B's posture is standing, accompanied by a slight turn of the head to the left and right, indicating that he is observing the environment. After customer B sits down in the waiting area of ​​the lobby, the system captures his sitting posture and monitors that he remains still for more than 5 minutes. The system records the behavior and generates behavioral characteristic data {body shape: {170cm, 65kg}, posture: sitting, stay time: 5 minutes, action: none}. The system generates body shape and behavioral characteristic data for the customer. Through body shape and posture analysis, the system can recommend a more suitable massage plan and provide data support for the next step.

[0096] The customer's wristband is connected to the Bluetooth beacon and Wi-Fi positioning system in the foot bath city. The system obtains the customer's location in real time through multi-source data fusion, and generates behavioral location mapping data based on the customer's location and behavioral characteristics. The system locates that customer B walks from the waiting area in the lobby to the massage area in 1 minute. The positioning accuracy is ±1 meter, and the system records the location every 3 seconds. Through positioning, the system updates the real-time location of customer B as he moves. When passing through the physical therapy area on the way, customer B briefly stops (about 10 seconds), and the system records the location {location: physical therapy area, stay: 10 seconds}, and generates behavioral location mapping data {location: massage area, action: walk for 1 minute} in combination with the customer's behavioral characteristic data. In the process of approaching the massage area, the system detects that the area where customer B stays contains the health-preserving medicinal bath experience area. According to the system's recommendation model, the system sends personalized recommendation information for health-preserving experience to customers when they approach the medicinal bath area. Through behavioral location mapping data, the system records the customer's real-time location, stay area and walking path. When customer B arrives at the massage area, the system has pushed a preferential information for the experience of medicinal bath.

[0097] The system records the behavior trajectory of customers in the foot bath city, identifies their behavior patterns through path analysis, and determines their entry and exit status. The system records the customer's stay and movement in different areas through the behavior trajectory to generate complete path data. For example, the path of customer B is {hall waiting area → physical therapy area → massage area → tea area}. Through path segmentation, the system identifies that the customer stays in the massage area for 40 minutes and stays in the tea area for 20 minutes. The system analyzes the path morphological characteristics of customer B (the curvature is close to 0, indicating that it is mainly a straight line movement). Through the speed characteristics, it is identified that the customer's speed from the waiting area to the massage area is 0.5 m / s, and the speed from the massage area to the tea area slows down to 0.3 m / s. The system generates path analysis data {area: {hall waiting area, physical therapy area, massage area, tea area}, stay time: {massage area: 40 minutes, tea area: 20 minutes}, action mode: relax, slow} based on the path curvature, speed, and stop point data. The system completes the customer's path analysis and identifies his behavior pattern as "relaxed and slow", which meets the needs of health and leisure experience, and provides further support for specific services and recommendations. Through behavioral characteristics, location mapping and path analysis data, the system generates a complete behavioral data map for customer B {behavioral characteristics: relaxed, slow, stay preference: massage area, tea area, interest area: medicinal bath area}. Based on customer B's stay data in the health area, the system recommends the special medicinal bath project of the Foot Bath City to him, and provides experience discount coupons in the push. The system also records customer B's preferred area as "health / physiotherapy" and gives priority to recommending physiotherapy packages or customized service options when he visits next time. If customer B leaves the massage area and enters the tea area, the system will automatically arrange for the waiter to deliver special refreshments to enhance the customer experience.

[0098] See also Figures 1 to 5 , the present application provides a method for managing user entry and exit of a store, the method comprising:

[0099] S1. Control the wristband communication device to exchange user wristband data within the sensing range to obtain user wristband data, and perform user device authentication according to the user wristband data to obtain user device authentication data;

[0100] Specifically, the wristband communication device at the entrance of the foot massage city detects the wristband signal when the customer enters. The system communicates with the wristband over a short distance to exchange data, including customer ID, membership status, health data, etc. The wristband sends the customer information to the local system via Bluetooth or RFID communication for identity verification. The system compares the received wristband data with the information in the customer database to verify the identity. If the customer is a registered member, the system generates "user device authentication data".

[0101] S2. Collecting user image data according to the user device authentication data to obtain user image data, and extracting user behavior features according to the user image data to obtain user behavior feature data;

[0102] Specifically, when the user's identity is confirmed, the camera or other sensor device begins to capture the user's image data, including facial expressions, height, body shape and other information. The image data is transmitted to the processing module, and the user's facial expressions, body language and other information are analyzed to extract the user's behavioral characteristics. For example, whether the user is browsing products, whether he has shopping intentions, whether he is queuing to check out, etc. Based on the image processing results, the user's behavioral characteristic data is stored for analysis in subsequent steps.

[0103] Specifically, after successful authentication, the system activates the in-store camera to take a picture of the customer. The camera can be installed at key locations such as the entrance, rest area, and massage area to capture the customer's facial expressions and body postures. The system analyzes the image data and extracts the customer's behavioral characteristics. For example, expression recognition and posture analysis are used to determine whether the customer is resting, waiting, or receiving service. The system generates "user behavior characteristic data", including the customer's emotional state (such as relaxation, pleasure) and behavior type (such as waiting, receiving service).

[0104] S3, obtaining the user's indoor location data, and performing location mapping according to the user's behavior feature data and the user's indoor location data to obtain user behavior location mapping data;

[0105] Specifically, positioning devices (such as Bluetooth beacons, Wi-Fi access points or UWB) are installed in the foot massage city to track the customer's location. The positioning system regularly records the customer's real-time coordinates in the foot massage city, for example (X=4.5, Y=2.0). The system determines the customer's specific activity area based on the customer's behavioral characteristics and location data. For example, if the customer is in the massage area and shows a relaxed posture, the system marks this behavior as "massage area-rest". Generate "user behavior location mapping data", associate the customer's behavior with the location, and provide data support for path tracking.

[0106] S4. Track the entry and exit behavior according to the user behavior location mapping data to obtain the user's entry and exit behavior data to perform user entry and exit management operations.

[0107] Specifically, the system uses location mapping data to track the customer's path, record the customer's specific location at different time points, and connect them into a path in chronological order. For example, the system records that the customer enters the massage area from the rest area and then goes to the tea break area. The system analyzes the customer's path data and identifies his entry and exit behavior. For example, when the customer's path data shows that he enters from the entrance and finally passes through the exit, the system marks it as a complete entry and exit behavior. The system generates "user entry and exit behavior data", including the customer's entry time, departure time, and length of stay in each area.

[0108] Optionally, S1 includes:

[0109] S11, detecting the sensing area and establishing a temporary connection through the wristband communication device to obtain sensing temporary connection data;

[0110] Specifically, the wristband communication device continuously scans the signal source within its sensing range to detect whether a wristband enters. The scan can be performed through Bluetooth Low Energy (BLE) technology to monitor the signal strength (RSSI) of a specific frequency band. When the wristband enters the sensing area, the wristband communication device detects the wristband signal and generates a temporary connection request. The communication device identifies the request as "sensing temporary connection data", which contains basic information such as wristband ID, wristband type, and signal strength. The communication device and the wristband establish a temporary communication connection based on the sensing temporary connection data. Communication protocols such as Bluetooth can be used for this connection, and a low-latency mode is set in the temporary connection to minimize the data exchange time. After the temporary connection is established, preparations for data exchange begin between the communication device and the wristband. At this time, the generated temporary connection data contains the device MAC address, connection strength information, etc., which are used for the next step of data exchange.

[0111] S12, exchanging wristband data according to the induction temporary connection data to obtain the user's wristband data;

[0112] Specifically, after the connection is established, the communication device sends a data exchange request. The wristband responds to the request and transmits the user's wristband data. This data may include the user ID, the device number of the wristband, the user's health status information, the membership level, etc. The communication device will perform an integrity check on the data exchange process to ensure that the transmitted data is correct. If the data transmission is successful, a "transmission success" signal will be fed back to the wristband. After receiving the data from the wristband, the communication device parses it. For example, key fields such as user ID and device status are extracted, and these data are combined into a "user wristband data packet".

[0113] S13: Transmit the user's wristband data to the edge computing node for verification and comparison to obtain user device authentication data.

[0114] Specifically, the communication device transmits the generated user bracelet data packet to the edge computing node for authentication and comparison operations. The edge computing node is usually located in the network node of the store and has higher computing power and data processing speed. The transmission process uses encryption protocols to ensure data security. For example, symmetric encryption technology is used to ensure that the data will not be tampered with during transmission. After receiving the user bracelet data packet, the edge computing node compares the user ID and device number therein with the user information in the database. For example, the user ID is compared with the record in the store member database to verify whether the user identity is valid. The authentication and comparison process includes two main operations: one is to determine whether the user is a member, and the other is to check whether the bracelet device number matches the device record in the database. If the comparison is passed, the edge computing node generates "user device authentication data", marks the user identity as authenticated, and marks its permission level (such as VIP, ordinary member, etc.). After the authentication is completed, the edge computing node feeds back the generated user device authentication data to the communication device and stores the authentication result. The communication device updates the connection status to keep the connection valid, or disconnects the connection as appropriate. Finally, the authentication data is transmitted to the store management system as input data for subsequent processes, providing support for subsequent image acquisition and behavior tracking.

[0115] Optionally, S2 includes:

[0116] S21, activating the camera and collecting user image data according to the user device authentication data to obtain the user image data;

[0117] Specifically, after the user device authentication data is confirmed, the system triggers the camera activation mechanism. The camera starts to collect images according to the preset activation conditions (such as the user approaches or passes through a certain area). The activated camera takes real-time photos of the user and transmits the image data to the edge computing device for further analysis. The camera collects user image data within a certain range, including the user's face, body posture, and overall body shape. The system performs preliminary filtering and clarity detection based on the quality of the captured image to ensure that the captured image meets the analysis requirements. The collected image data is converted into a digital image format and a "user image data packet" is generated on the edge device.

[0118] S22, extracting user body shape data according to the user image data to obtain user body shape data;

[0119] Specifically, the system uses the user's image data to identify the user's body area through image segmentation technology. For example, the system identifies the outlines of the user's head, torso, and limbs, and calculates their relative proportions. The image analysis process marks the user's body features such as height, shoulder width, and leg length as key data. For example, the system can estimate that the user's height is 1.75 meters and the shoulder width is 0.45 meters. The "user body data" generated by the system includes the user's basic body parameters, such as height, body outline, body proportions, etc.

[0120] S23, extracting user posture features according to the user body shape data to obtain user posture feature data;

[0121] Specifically, the system further analyzes the user's posture characteristics based on the user's body shape data. The image skeleton recognition method is used to determine the position information of the user's skeletal points, such as the shoulder, elbow, knee and other key points. Posture feature recognition includes the judgment of the user's limb movements, for example, whether the user is standing, walking, browsing, etc. The "user posture feature data" generated by the system contains information such as the user's limb angle and movement state. For example, the user's elbow is bent at an angle of 30 degrees and the shoulder is facing a certain direction.

[0122] S24: Integrate the user's body shape data and the user's posture feature data to obtain user behavior feature data.

[0123] Specifically, the system integrates the user's body shape data with the posture feature data to analyze the user's overall behavior. For example, the user's body shape and posture are combined to determine the user's specific behavior pattern (such as standing and observing, walking and browsing, etc.). The integrated data is used to generate "user behavior feature data", which contains body shape and posture information. The system can analyze user behavior based on this, for example, whether the user is focusing on a certain product area.

[0124] Optionally, the user posture feature extraction includes:

[0125] Perform key point detection according to the user's body shape data to obtain the user's body shape key point data;

[0126] Specifically, based on the user's body shape data, the system learns and extracts features in the image through stacked convolutional layers based on convolutional neural networks (deep learning models such as OpenPose, PoseNet, HRNet, etc.), identifies the human body structure, and thus identifies the user's key points, such as the head, shoulders, elbows, wrists, hips, knees, ankles, etc. The coordinates of each key point are marked and recorded to generate the user's body shape key point data. The first part of the model is used to extract image features. For example, the convolutional neural network extracts low-level and high-level features. The subsequent part of the model processes the feature map through a convolutional layer or a deconvolution layer to generate a heatmap of the human body key points, each heatmap representing the location of a key point. The output of the model is usually multiple heatmaps, each heatmap corresponding to a human body key point (such as head, shoulder, elbow, etc.). The value of each pixel in the heatmap indicates the nature of the position as the corresponding key point. By checking the positions with the highest activation values ​​in each heatmap one by one, the system can identify the candidate positions of each key point. The system finds the pixel position with the highest value in each heatmap, maps the position back to the coordinates in the original image, and obtains the corresponding key point coordinates. For example, a high-value point in the heat map is located at (30, 40), and the coordinates of the point in the original image are inferred to be (X=120, Y=160) through mapping. The main task of OpenPose is to identify the key points of humans from images and form a skeleton structure through a deep learning model. It is based on convolutional neural networks (CNNs) to extract features from images. The main steps include: the model first uses convolutional neural networks to extract low-level and high-level features of the image to capture the position and structure of the human body in the image. OpenPose generates a probability heat map for each key point, and each heat map represents the location of a specific human body part. For example, key points such as the head, shoulders, and elbows have their own heat maps.

[0127] Perform skeleton marking according to the key point data of the user's body shape to obtain the user's skeleton marking data;

[0128] Specifically, when the coordinates of all key points are determined, the system connects the key points into a skeleton structure according to the predefined skeleton connection rules. The skeleton connection rules include shoulder to elbow, elbow to wrist, etc. The system uses the key point data to draw the user's skeleton structure and connect the user's limbs into a skeleton model (i.e., joint connection lines) to intuitively present the user's posture. The skeleton marker data consists of line segments for each pair of connection points, such as shoulder to elbow, elbow to wrist, etc., to generate user skeleton marker data.

[0129] Coordinates are normalized and motion detection is triggered according to the user skeleton marking data to obtain user motion detection data;

[0130] Specifically, the system standardizes the coordinates in the skeleton marking data to eliminate the coordinate scale differences caused by different body shapes. Coordinate standardization maps all key points to a unified coordinate system through normalization or scaling. The system detects the user's motion changes based on the standardized coordinates. The system extracts the motion characteristics of each key point, such as position change, speed, acceleration, etc., based on the standardized skeleton data. The system first records the standardized coordinates of the previous frame and the current frame of each key point, and then calculates the displacement between the two frames. The system calculates the magnitude of the displacement as the motion amplitude. The system compares the motion amplitude with the preset motion threshold to determine whether the key point is moving. For example, the threshold is set to 0.05. If the motion amplitude exceeds this value, the key point is determined to be in motion. If multiple key points exceed the threshold at the same time (such as shoulders, arms, and legs), the system determines that the user is in motion as a whole. The system will mark whether each key point is in motion. For example, if the motion amplitude of the left and right shoulders exceeds the threshold, it is marked as "upper body movement". For example, the position difference of each key point in consecutive frames is analyzed to determine whether the user's limbs are moving. The motion detection trigger data contains whether the user is stationary or performing some kind of motion.

[0131] Performing posture category recognition based on user motion detection data to obtain user posture category data;

[0132] Specifically, based on the motion detection data, the system analyzes the user's posture category, for example, determining whether the user is standing, walking, sitting, bending, etc. The system generates user posture category data and marks the user's posture category using the relative position and angle changes of key points.

[0133] Calculate the limb angle according to the user's posture category data to obtain the user's limb angle data;

[0134] Specifically, the system calculates the angles between the joints based on the posture category data. For example, the system calculates the bending angle of the elbow joint through the relative positions of the shoulder, elbow, and wrist in the user's skeleton. The user's limb angle data is generated, which records the angle value of each joint, such as the right arm elbow bent 45 degrees.

[0135] Calculate the posture change rate according to the user's limb angle data to obtain the user's posture change rate data;

[0136] Specifically, the system analyzes the changes in the user's limb angle data between consecutive frames to determine the rate of change of the posture. The posture change rate data records the rate of change of each joint angle. For example, the user's right arm bending angle changes by 10 degrees per second, generating the user's posture change rate data.

[0137] Performing posture symmetry analysis based on the user posture change rate data to obtain the user posture symmetry data;

[0138] Specifically, the system analyzes the symmetry of the user's limbs based on the posture change rate data. For example, the angle change rates of the left and right arms are compared to determine whether the action is symmetrical. If there is a large difference between the left and right movements, the system marks the asymmetrical part. Generate user posture symmetry data to describe the symmetry of the action. The system pairs the left and right symmetrical key points according to the human body structure. For example: left shoulder and right shoulder, left elbow and right elbow, left knee and right knee, etc. The system extracts the change rate data of these key points from the posture change rate data to compare the left and right symmetry. The system records the posture change rate of each pair of symmetrical key points and generates a data set for each pair of key points. The system calculates the difference in posture change rate for each pair of symmetrical key points to evaluate the symmetry of the two sides. For example, for the left shoulder and the right shoulder, if the change rate of the left shoulder is 0.3 and the change rate of the right shoulder is 0.2, the difference is 0.1. The smaller the difference, the higher the left-right symmetry; the larger the difference, the asymmetry of the left and right movements.

[0139] Performing posture stability adjustment according to the user posture change rate data to obtain user posture stability adjustment data;

[0140] Specifically, the system analyzes the stability of the user's posture based on the posture change rate data. For example, it analyzes whether the user's posture is stable when standing. By adjusting the detected key point positions in real time, the system marks and processes unstable movements and generates user posture stability adjustment data. For each key point, the system calculates the mean and standard deviation of the rate of change over a period of time to evaluate its stability. Key points with low and stable rates of change indicate that the part is relatively still, while large rates of change indicate that the part is moving or swinging frequently. The system sets a rate of change threshold for each key point to determine whether it is stable. For example, the head and shoulders should remain stable when standing, while the ankles and knees can have small changes. If the rate of change of a key point exceeds the stability threshold, it is marked as "unstable". For example, the stability threshold of the head is set to 0.05. If the mean rate of change of the head is greater than 0.05, the head is considered unstable.

[0141] Perform posture dynamic amplitude measurement according to user posture symmetry data and user posture stability adjustment data to obtain user posture dynamic amplitude measurement data;

[0142] Specifically, the system measures the user's dynamic range of motion based on the posture symmetry data and posture stability adjustment data. For example, it measures the height and range of the user's hand and leg raises. It generates user posture dynamic range measurement data, records the amplitude values ​​of the user's limb movements, and describes the overall dynamic range of the movements.

[0143] A posture feature vector is generated according to the user posture category data, the user posture change rate data and the user posture dynamic amplitude measurement data to obtain the user posture feature data.

[0144] Specifically, the system generates a feature vector from the user's posture features based on the posture category data, posture change rate data, and posture dynamic amplitude measurement data. The feature vector data contains the core information of the user's overall posture, such as posture category (standing, walking), change rate, movement amplitude, etc., and is output as user posture feature data for subsequent analysis.

[0145] Optionally, the gesture category recognition includes:

[0146] Perform dynamic and static posture classification according to the user motion detection data to obtain the user's dynamic and static posture data;

[0147] Specifically, the system distinguishes whether the user is currently in a dynamic or static state based on the user's motion detection data. For example, the movement rate and frequency of key points can be used to determine whether the user is stationary, standing, walking, etc. If the user's key points do not move significantly for a long time, it is marked as "static"; otherwise, it is marked as "dynamic".

[0148] Perform posture turning point recognition based on the user's dynamic and static posture data to obtain posture turning point data;

[0149] Specifically, the system analyzes the dynamic and static posture data to identify the turning points of the user in the action, that is, the moment when the action starts, ends, or changes direction. For example, when the user suddenly changes from a static state to a dynamic state, this moment is marked as a turning point. The system records these turning point timestamps and generates posture turning point data, which provides key information for the hierarchical classification of actions.

[0150] Perform preliminary screening of hierarchical classification based on posture turning point data to obtain posture hierarchical classification data;

[0151] Specifically, based on the posture turning point data, the system performs hierarchical classification and divides the user's posture into different levels. If the user's speed is close to zero and the posture is stable, it is classified as "stationary"; if the stationary state lasts for a long time and the hip joint height decreases, it is classified as "sitting down"; if the height is high, it is classified as "standing". If the user's speed continues to increase, the pace changes significantly and is accompanied by a steady arm swinging movement, it is classified as "walking"; if the speed is fast and the stride increases, it is classified as "running". For example, the user's posture is first classified into basic posture categories (such as stationary, moving). The system generates posture hierarchical classification data and records the user's posture hierarchical classification for further detailed feature extraction.

[0152] Detailed features are extracted according to the posture hierarchical classification data to obtain posture detail feature data;

[0153] Specifically, the system identifies the movement trajectory of key points (such as ankles and knees) when the user walks, and measures the length of the step by calculating the horizontal distance between each foot lift and foot landing. The system analyzes the key point data in consecutive frames, determines the moment of foot lift and foot landing, and calculates the step length;

[0154] The system calculates the user's pace by calculating the length and time of the steps. The pace refers to the product of the number of steps and the length of the steps per unit time. During the continuous movement, the system records the start and end time of each step to calculate the average pace;

[0155] The system analyzes the swing amplitude of the user's joints (such as knees, ankles and hips) during walking, and measures the amplitude of movement by tracking the changing angles and positions of these key points. The amplitude of movement includes the amplitude of two dimensions, vertical and horizontal. For example, the system detects the height of the user's leg lift or the width of the stride to understand the overall amplitude of movement;

[0156] The system calculates the range of motion based on the highest and lowest positions of key points. For example, the height change when the knee is raised is part of the range of motion. The range of motion data can describe the user's "slight swing" or "large amplitude" step state while walking;

[0157] The system identifies the bending angles of the user's joints when walking, such as the degree of bending of the knee and hip joints. The bending angle can be determined by detecting the angle changes at the joints. For example, during walking, the bending angle of the knee is calculated by the angle between the thigh, knee and calf;

[0158] The system analyzes the swing amplitude of the user's arms and legs when walking. For example, by calculating the position changes of key points from shoulder to elbow and elbow to wrist, the system analyzes the swing of the arm. The swing amplitude refers to the amplitude or range of the arm swing during walking, which is expressed by the displacement of key points;

[0159] The stride length, average stride speed, range of motion, bending angle of each joint, and swing amplitude of arms and legs are integrated into posture detail feature data.

[0160] Perform multi-level behavior classification based on posture detail feature data and posture hierarchical classification data to obtain posture multi-level behavior data;

[0161] Specifically, the system combines posture hierarchical classification data with detailed feature data to perform multi-level classification of user postures. For example, the "movement" category can be further subdivided into "fast walking", "slow walking" or "running". The system classifies the user's behavior into a more specific level through the detailed features of each type of posture. Multi-level posture behavior data is generated to mark the user's behavior category at multiple levels.

[0162] Divide the action sequence according to the multi-level behavior data of the posture to obtain the posture behavior action sequence data;

[0163] Specifically, the system divides the multi-level behavior data into continuous action sequences in chronological order. For example, the user's "stand-walk-pause" is marked as a complete action sequence. After the action sequence is divided, the system analyzes the connection mode and transition of the actions based on the chronological order. The posture behavior action sequence data is generated, which contains the user's action sequence information in the time dimension.

[0164] Extracting posture transition data according to posture behavior action sequence data to obtain posture transition data;

[0165] Specifically, the system extracts transition data between actions based on the action sequence. For example, the transition period from "standing" to "walking" is transition data, which describes the connection and conversion pattern of the user's actions. The system analyzes the duration, speed change and other information of these transition segments. It generates posture transition data to describe the connection characteristics between each action.

[0166] Matching is performed based on the posture transition data and the preset posture category template data to obtain user posture category data.

[0167] Specifically, the system compares the posture transition data with the preset posture category template. The system presets different posture templates (such as "walk-pause-walk" and other patterns), and confirms the user's specific posture category through matching. For example, if the user's behavior meets the "walk-pause-walk" template, the system will classify it as "store patrol". Generate user posture category data and accurately mark the user's posture type.

[0168] Optionally, the dynamic and static posture classification includes:

[0169] Extracting acceleration features and speed features based on the user motion detection data to obtain user motion acceleration feature data and user motion feature data;

[0170] Specifically, the system extracts the acceleration information of each key point based on the user's motion detection data. For example, the acceleration of each point is obtained by calculating the displacement difference of the key point between consecutive frames. The generated user motion acceleration feature data contains the acceleration value of each key point. The system calculates the speed feature based on the key point displacement data. The speed is calculated by the displacement difference of consecutive frames and used together with the acceleration data for preliminary classification. The generated user motion feature data records the speed value of the key point.

[0171] Preliminarily classify the user motion detection data according to the user motion acceleration feature data and the user motion feature data to obtain primary user dynamic and static posture data;

[0172] Specifically, the system combines the acceleration feature data and the speed feature data to preliminarily classify the user's motion detection data into dynamic and static. For example, when the acceleration and speed are low, it is static, and when they are high, it is dynamic. Generate primary user dynamic and static posture data and mark the user's preliminary dynamic and static state.

[0173] Perform multi-joint consistency analysis and center of gravity stability analysis based on the user motion detection data to obtain the user motion multi-joint consistency data and the user motion center of gravity stability data respectively;

[0174] Specifically, the system analyzes the consistency of movements of multiple joints and checks whether the movement direction and amplitude of key points are synchronized. For example, during walking, the left and right legs and arms should show high consistency. The generated user movement multi-joint consistency data describes the consistency of the joints and indicates the degree of coordination of joint movements. The system analyzes the changes in the user's center of gravity trajectory and calculates the stability of the center of gravity. For example, the center of gravity should remain stable when standing, but will fluctuate greatly when jumping or running. The generated user movement center of gravity stability data describes the degree of stability of the center of gravity.

[0175] Perform secondary classification on the primary user's dynamic and static posture data according to the user's motion multi-joint consistency data and the user's motion center of gravity stability data to obtain secondary user's dynamic and static posture data;

[0176] Specifically, the system refines and classifies the primary dynamic and static posture data based on the multi-joint consistency data and the center of gravity stability data. For example, if the user's center of gravity is relatively stable and the joint consistency is high, it is in a static state; otherwise, it is in a dynamic state. The generated secondary user dynamic and static posture data contains the user's secondary classification results, marking more detailed dynamic and static information.

[0177] Performing time series division on the user motion detection data to obtain user motion time series division data;

[0178] Specifically, the system divides the user motion detection data into several time periods, for example, one time period per second, so as to analyze the motion changes. The generated user motion time series division data marks the time series information of the motion.

[0179] Calculate the activity change rate based on the user's motion time series data to obtain the user's motion activity change rate data;

[0180] Specifically, the system calculates the activity change rate of each time segment, that is, the change amplitude of the speed and acceleration of each key point. This data is used to determine the intensity and rhythm of exercise. The generated user exercise activity change rate data represents the rate of change of exercise in different time periods.

[0181] Perform energy calculation based on the user's sports activity change rate data to obtain the user's sports activity change rate energy data;

[0182] Specifically, the system calculates the user's exercise energy consumption based on the activity change rate data. For example, the activity change rate of each time segment is converted into an energy value. The user's exercise activity change rate energy data is generated to represent the user's exercise energy consumption for weighted classification. , It is the energy data of the user's sports activity change rate. It is the user's sports activity change rate data. is the user's weight (kg), as a weighting factor, is the gravitational acceleration ( ), used to calculate the physical forces in motion, The distance or range of motion generated by the activity change rate. If the activity change rate is small, it means that the user's exercise intensity is low, and the energy consumption is also low; vice versa.

[0183] Energy-weighted dynamic and static posture classification is performed on the secondary user dynamic and static posture data according to the preset time series division parameter data corresponding to the user movement time series division data and the energy data of the user movement activity change rate to obtain the user dynamic and static posture data.

[0184] Specifically, the system uses the time series division data, activity change rate energy data and preset time series division parameters to weight the secondary dynamic and static posture data. If the energy in a certain period is low and stable (based on the corresponding threshold judgment of body shape, or based on the fixed threshold judgment), it is marked as static; otherwise, it is dynamic. The generated user dynamic and static posture data is the weighted dynamic and static state classification result, which provides support for further posture category recognition.

[0185] Specifically, the system calculates the average energy consumption of the movement based on the energy data of the activity change rate of each time segment. For example, if the user's energy consumption in a period is 60 joules, the system uses a preset low energy threshold (such as 50 joules) to determine the activity intensity of the period. The system determines the dynamic and static characteristics of the user in the period by weighting the energy data. For example, if the energy is low and there is no significant fluctuation, the system marks the period as static or slightly moving. The system applies a set of dynamic and static classification rules to make weighted judgments on the energy of the period: Static: If the energy consumption of the period is lower than the set static energy threshold (such as 50 joules) and the rate of change is low, it is judged as static. Low dynamic: If the energy consumption of the period is lower than the medium energy threshold (such as 100 joules), but higher than the static threshold, and the rate of change is medium, it is judged as low dynamic. High dynamic: If the energy consumption of the period is higher than the medium energy threshold and the activity change rate is high, it is judged as high dynamic. The system uses these rules to perform energy weighted classification on each period and mark it as "static", "low dynamic" or "high dynamic". The system further weights the energy of each time period based on the user's body parameters and time series division parameters. For example, the heavier the body weight, the higher the weight coefficient to reflect the difference in energy consumption. Based on the above classification results, the system generates the user's dynamic and static posture data. Each time series segment is marked as static, low dynamic or high dynamic state.

[0186] Optionally, the posture change rate calculation includes:

[0187] Perform time series segmentation according to the user's limb angle data to obtain user limb angle segmentation data;

[0188] Specifically, the system divides the user's limb angle data into time series. For example, the angle data is divided into multiple time series segments at intervals of 0.5 seconds or 1 second. The generated user limb angle segmentation data records the changes in limb angles in each time period, forming a time series to prepare for the calculation of long-term and short-term change rates. For example, the system divides the user's elbow angle data into 1 second. From the 1st second to the 5th second, the angle data per second is 10 degrees, 12 degrees, 15 degrees, 17 degrees, and 20 degrees. The generated limb angle segmentation data is: [10, 12, 15, 17, 20].

[0189] Calculating the long-term and short-term change rates of the user's limb angle segmentation data to obtain limb angle long-term and short-term change rate data, wherein the limb angle long-term and short-term change rate data includes limb angle long-term change rate data and limb angle short-term change rate data;

[0190] Specifically, the system performs long-term and short-term change rate analysis on the limb angle segmentation data. The long-term change rate refers to the angle change trend within a longer time period (such as 5 seconds or 10 seconds), while the short-term change rate calculates the angle change rate within a shorter time period (such as 1 second or 2 seconds). The generated limb angle long-term and short-term change rate data includes limb angle long-term change rate data and limb angle short-term change rate data, reflecting the change trend of user actions in long and short time periods. For example, if the system calculates the long-term change rate every 5 seconds, the average change is (20-10) / 5=2 degrees / second. The system calculates the short-term change rate per second, and the results are 2 degrees / second, 3 degrees / second, 2 degrees / second, and 3 degrees / second respectively. The long-term change rate is 2 degrees / second, and the short-term change rate is [2, 3, 2, 3] degrees / second.

[0191] The fluctuation of the change rate is calculated based on the long-term and short-term change rate data of the limb angle, and the fluctuation data of the change rate of the limb angle is obtained;

[0192] Specifically, the system calculates the volatility of the rate of change based on the long-term and short-term rate of change data of the limb angle. For example, the fluctuation range of the short-term rate of change in different time periods can reflect the stability of the user's action. The generated limb angle rate of change volatility data contains a volatility index to describe the smoothness or jitter characteristics of the action. If the system calculates the variance of the volatility of the short-term rate of change and obtains a volatility of 0.5 (a small variance indicates a relatively smooth change), the generated rate of change volatility data is 0.5.

[0193] According to the short-term change rate data of the limb angle, the change rate directionality analysis is performed to obtain the limb angle change rate directionality data;

[0194] Specifically, the system analyzes the directionality of the short-term rate of change of limb angles and determines the direction of the angle change (e.g., rising, falling). This process is used to identify the continuity direction and trend of the action. The generated limb angle change rate direction data describes the positive and negative trends of the action direction in the short term, for example, marked as "rising" or "falling".

[0195] According to the short-term change rate data of the limb angle, short-term change rate classification is performed to obtain the short-term change rate classification data of the limb angle;

[0196] Specifically, the system classifies the short-term change rate data of limb angles (based on a threshold or a comparison of multiples of variance, such as weighting the variance to obtain a weight, weighting the initial value to obtain a classification threshold, and then classifying), and marks it into different categories according to the rate and trend of change, such as "rapid change", "steady change", etc. The generated short-term change rate classification data of limb angles classifies the short-term change rate into different categories, providing input for feature vector conversion. For example, the system classifies a short-term change rate of 2 degrees / second as "medium-speed change" and 3 degrees / second as "rapid change", and the generated short-term change rate classification data is "medium-speed change" and "rapid change".

[0197] The user posture change rate data is obtained by performing feature vector conversion based on the long-term and short-term change rate data of the limb angle, the volatility data of the limb angle change rate, the directional data of the limb angle change rate and the classification data of the short-term change rate of the limb angle.

[0198] Specifically, the system integrates the long-term and short-term change rate data of limb angles, change rate volatility data, directionality data, and short-term change rate classification data to generate a feature vector of the user's posture change rate. This feature vector is used to describe multiple dimensions such as the change rate, volatility, and directionality of the user's movements. The generated user posture change rate data contains all-round information about the user's posture changes, providing support for subsequent posture analysis. For example, the system integrates information such as the long-term change rate of 2 degrees / second, the short-term change rate of [2,3,2,3] degrees / second, volatility of 0.5, directionality "rising", and short-term change classification data to form a feature vector. The generated user posture change rate data is a feature vector, such as: (2 degrees / second, [2,3,2,3], 0.5, rising, [medium-speed change, rapid change]).

[0199] Optionally, S3 includes:

[0200] S31, obtaining user real-time location data through an indoor positioning system;

[0201] Specifically, the system obtains the user's real-time location through indoor positioning technologies such as Bluetooth beacons, Wi-Fi signals, ultra-wideband (UWB), infrared or RFID. Positioning devices are arranged in multiple locations within the store to form a positioning network that can cover different areas of the store. Each positioning device calculates the user's location coordinates based on data such as signal strength, signal arrival time or angle. For example, a Bluetooth beacon can estimate the distance between the user and the beacon through RSSI (received signal strength indication). The "user real-time location data" generated by the system includes the user's coordinates in the store (such as X=5.2 meters, Y=3.4 meters) for subsequent data fusion.

[0202] S32: Perform multi-source positioning data fusion according to the user's real-time location data to obtain the user's indoor location data.

[0203] Specifically, the system fuses positioning data from multiple positioning sources (such as multiple Bluetooth beacons or multiple Wi-Fi access points) to improve positioning accuracy. For example, by calculating the weighted average of the positions of each signal source or triangulating multiple signal data, more accurate position coordinates are obtained. Data fusion also includes denoising of position data to eliminate the impact of erroneous signal data on the final position estimate. The system generates user indoor position data to provide accurate user position coordinates for behavioral location mapping.

[0204] S33: performing association matching based on the user behavior feature data and the user indoor location data to obtain user behavior location mapping data.

[0205] Specifically, the system matches the user's behavior with a specific area in the store based on the user's indoor location data and behavior feature data. For example, if the system recognizes that the user's behavior feature is "resting" and the user is in a private room, the system will mark it as "private room-resting". The system identifies which area the user is in by comparing the user's current coordinates with the regional layout map of the foot bath city. The system then associates the location data with the behavior features to form behavior-location mapping data. For example, the user's behavior feature data is "waiting" and the location coordinates are (X=1.5 meters, Y=3.2 meters). The system recognizes that the location is in the waiting area, and generates the mapping data "waiting area-waiting". The generated "user behavior location mapping data" records the association between user behavior and location, such as "waiting area-waiting" or "massage area-receiving service".

[0206] Optionally, S4 includes:

[0207] S41, tracking the user behavior trajectory according to the user behavior location mapping data to obtain user behavior trajectory data;

[0208] Specifically, the system tracks the user's activity path in the store based on the user's behavioral location mapping data. By regularly collecting the user's location data (such as every few seconds or every minute), the system records the location in chronological order to form a complete behavior trajectory. The location data can include coordinate information and timestamps. For example, if the user goes from the waiting area to the private room and then to the rest area, the system will connect these location points to generate a path. The generated user behavior trajectory data includes the user's movement information and time points between different locations to show the user's complete activity trajectory.

[0209] S42, performing user path analysis according to the user behavior trajectory data to obtain user path analysis data;

[0210] Specifically, the system analyzes the path based on the user's behavior trajectory data, and calculates the user's stay time in different areas, the paths passed, and the frequency of activities. For example, the system can identify the frequency of users' movement between areas and the length of each stay. This step can also identify the user's movement patterns, such as whether there are many round-trip routes, or the frequency of passing through certain specific areas, which helps to understand the user's in-store behavior patterns, including the areas with the longest stay time or the user's main activity routes. The generated user path analysis data includes information such as the length of time the user stays in each area, the order of passing through, and repeated paths.

[0211] S43. Extracting the user's entry and exit behavior based on the user path analysis data to obtain the user's entry and exit behavior data for user entry and exit management.

[0212] Specifically, the system extracts the user's entry and exit behavior based on the path analysis data. The system detects whether the user passes through the store's entrance and exit areas to determine the user's entry and exit time and number of times. The system records the user's entry and exit time and calculates their total stay time in the store. If the user's path data shows that they enter from the entrance and eventually pass through the exit, the system will record it as a complete entry and exit behavior. The generated user entry and exit behavior data includes the user's entry and exit time, the length of time they stay in the store, and the specific path information they pass through, providing a basis for store entry and exit management operations.

[0213] Optionally, the user path analysis includes:

[0214] Perform path segmentation according to user behavior trajectory data to obtain user behavior path segmentation data;

[0215] Specifically, the system divides the user's path into multiple segments based on the user behavior trajectory data. For example, segmentation is performed based on significant changes in location or at certain time intervals (such as every minute). Through segmentation, the system can independently analyze the user's behavioral characteristics in each segment. The generated user behavior path segmentation data includes the segmentation start and end points of the path and the coordinate sequence of each segment. The system divides the user path into three segments based on the time interval: Segment 1: waiting area (X=1.0, Y=2.0) to private room (X=4.0, Y=3.0), Segment 2: private room (X=4.0, Y=3.0) to rest area (X=2.5, Y=4.5), Segment 3: rest area (X=2.5, Y=4.5) to front desk (X=1.2, Y=3.0), and the generated user behavior path segmentation data is 3 path segments.

[0216] Calculate curvature and speed characteristics based on user behavior path segmentation data to obtain path curvature data and path speed characteristic data respectively;

[0217] Specifically, the system calculates the path curvature of each segment based on the user behavior path segmentation data. Curvature calculation is used to measure the curvature of the path, for example, by calculating the rate of change of angle to estimate the curvature of the path. A higher curvature indicates that the path is more curved. The generated path curvature data describes the curvature of each segment of the path, helping the system understand the path morphology. The system calculates the speed characteristics of the user in each path segment. For example, the average speed and speed change within the segment are calculated by dividing the distance between the start and end points of the segment by the time. The generated path speed characteristic data provides information about the user's movement rate in different path segments. The curvature of segment 1 is 0.1 (lower, indicating a straight path) and the speed is 1.5 m / s. The curvature of segment 2 is 0.4 (higher, indicating a curved path) and the speed is 0.8 m / s. The curvature of segment 3 is 0.2 and the speed is 1.0 m / s. The path curvature data and speed characteristic data generated by the system are: [0.1, 0.4, 0.2] and [1.5 m / s, 0.8 m / s, 1.0 m / s] respectively.

[0218] Perform stay point detection based on user behavior path segmentation data to obtain user stay point data;

[0219] Specifically, the system detects that the user stayed in the middle of segment 2 for 20 seconds and stayed in the private room for 50 minutes. The generated user stay point data is (X=4.0, Y=3.0, stayed for 50 minutes) and the middle of segment 2 (X=3.5, Y=4.0, stayed for 20 seconds).

[0220] Specifically, the system identifies the stop points from the user's path segmentation data, that is, the situation where the user stops at a specific location. Stop points often appear in areas of interest to the user, indicating that the user is observing, browsing, or waiting. The system detects the duration of each stop point. If a certain path contains more stop points, it is marked as a "long-stop segment"; otherwise, it is marked as a "short-stop segment".

[0221] Perform path morphology analysis based on path curvature data, path speed characteristic data, and user stop point data to obtain user behavior path morphology data;

[0222] Specifically, the system analyzes the segmented data of the user behavior path and identifies the user's stay points. A stay point refers to a situation where a user stays at a specific location for a period of time without moving. For example, if the user's speed is close to zero in a certain section of the path, the system identifies it as a stay point. The generated user stay point data marks the location and duration of the user's stay. The system combines the path curvature data, speed feature data and stay point data to analyze the overall shape of the path. For example, analyze the speed change of the user on a relatively straight path segment, or the stay on a high curvature path segment. The generated user behavior path morphological data contains the morphological features of the path segment (such as curved, straight or segments containing multiple stay points) to help identify the user's activity pattern. The system analyzes that the path of segment 1 is a straight line, segment 2 is a curved route and contains stay points, and segment 3 is a straight line. The user behavior path morphological data generated by the system are marked as: "Segment 1-straight line", "Segment 2-curved", "Segment 3-straight line".

[0223] Specifically, the system compares the stop point data and the path speed feature data to analyze whether the user slows down or stops at a specific location. For example, if the user's speed decreases in a certain area and there are multiple stop points, the system will mark the area as a "high stop area". Combined with the curvature data, the system further analyzes the path shape. For example, in a section with high curvature and multiple stop points, the system can mark it as an "area of ​​interest", indicating that the user detoured here and observed the surroundings, and obtained the user behavior path shape data.

[0224] Perform hot spot area detection and area switching frequency calculation based on user behavior path morphology data to obtain user behavior hot spot area data and user behavior area switching frequency data;

[0225] Specifically, the system detects hot spots frequently visited by users based on the path morphology data. For example, if a user stops in an area many times, the area is marked as a hot spot. The generated user behavior hot spot area data describes the high-frequency areas of user activities. The system analyzes the frequency of users switching between different areas based on the user behavior path segmentation data. For example, by recording the number of times a user switches between areas, the system can identify the user's activity frequency. The generated user behavior area switching frequency data records the frequency of users switching between different areas, helping to understand the user's behavioral preferences. For example, the system detects that users frequently stay in private rooms based on the stop points and marks them as hot spots. The system calculates that the user's switching frequency from the private room to the rest area and then to the front desk is once per hour. The user behavior hot spot area data generated by the system is "private room", and the area switching frequency data is 1 time per hour.

[0226] Generate a typical user path based on the path curvature data, path speed feature data, user behavior hotspot area data, and user behavior area switching frequency data to obtain typical user path data;

[0227] Specifically, the system generates a typical user path based on path curvature data, speed feature data, hotspot area data, and area switching frequency data. For example, by identifying the common path patterns of users, the system generates a representative path. The generated typical user path data marks the user's regular activity path. The system generates a typical path for users based on path curvature, speed features, hotspot areas, and area switching frequency: "waiting area → private room → rest area → front desk".

[0228] User path analysis data is integrated based on path curvature data, path speed feature data, user behavior hotspot area data, user behavior area switching frequency data, and typical user path data to obtain user path analysis data.

[0229] Specifically, the system integrates path curvature, speed characteristics, hot spots, area switching frequency and typical path data to generate complete path analysis data for users. This step summarizes all path analysis features and provides stores with detailed user behavior patterns. The generated user path analysis data is a comprehensive analysis of user activities, which is used for further behavior management and store optimization. For example, the user path analysis data is: "Waiting area → private room (hot spot) → rest area → front desk (switching frequency once per hour)".

[0230] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0231] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be 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 the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for managing users entering and leaving a store, characterized in that: The method comprises: S1. Control the wristband communication device to exchange user wristband data within the sensing range to obtain user wristband data, and perform user identity authentication according to the user wristband data to obtain user identity authentication data; S2. Collecting user image data according to the user identity authentication data to obtain user image data, and extracting user behavior features according to the user image data to obtain user behavior feature data; S3, obtaining the user's indoor location data, and performing location mapping according to the user's behavior feature data and the user's indoor location data to obtain user behavior location mapping data; S4. Tracking the entry and exit behavior according to the user behavior location mapping data to obtain the user's entry and exit behavior data for user entry and exit management; S4 includes: Track the user's behavior trajectory according to the user's behavior location mapping data to obtain the user's behavior trajectory data; Perform user path analysis based on user behavior trajectory data to obtain user path analysis data; The user entry and exit behavior is extracted based on the user path analysis data to obtain the user entry and exit behavior data for user entry and exit management operations.

2. The method according to claim 1, characterized in that S1 includes: Perform sensing area detection and establish a temporary connection through the wristband communication device to obtain sensing temporary connection data; Exchange wristband data according to the induction temporary connection data to obtain the user's wristband data; The user's bracelet data is transmitted to the edge computing node for verification and comparison to obtain user identity authentication data.

3. The method according to claim 1, characterized in that S2 include: Activate the camera and collect user image data according to the user identity authentication data to obtain the user image data; Extracting user body shape data according to user image data to obtain user body shape data; Extracting user posture features according to user body shape data to obtain user posture feature data; The user's body shape data and user's posture feature data are integrated to obtain user behavior feature data.

4. The method according to claim 3, characterized in that The user posture feature extraction includes: Perform key point detection according to the user's body shape data to obtain the user's body shape key point data; Perform skeleton marking according to the key point data of the user's body shape to obtain the user's skeleton marking data; Coordinates are normalized and motion detection is triggered according to the user skeleton marking data to obtain user motion detection data; Performing posture category recognition based on user motion detection data to obtain user posture category data; Calculate the limb angle according to the user's posture category data to obtain the user's limb angle data; Calculate the posture change rate according to the user's limb angle data to obtain the user's posture change rate data; Performing posture symmetry analysis based on the user posture change rate data to obtain the user posture symmetry data; Performing posture stability adjustment according to the user posture change rate data to obtain user posture stability adjustment data; Perform posture dynamic amplitude measurement according to user posture symmetry data and user posture stability adjustment data to obtain user posture dynamic amplitude measurement data; A posture feature vector is generated according to the user posture category data, the user posture change rate data and the user posture dynamic amplitude measurement data to obtain the user posture feature data.

5. The method according to claim 4, characterized in that The posture category recognition includes: Perform dynamic and static posture classification according to the user motion detection data to obtain the user's dynamic and static posture data; Perform posture turning point recognition based on the user's dynamic and static posture data to obtain posture turning point data; Perform preliminary screening of hierarchical classification based on posture turning point data to obtain posture hierarchical classification data; Detailed features are extracted according to the posture hierarchical classification data to obtain posture detail feature data; Perform multi-level behavior classification based on posture detail feature data and posture hierarchical classification data to obtain posture multi-level behavior data; Divide the action sequence according to the multi-level behavior data of the posture to obtain the posture behavior action sequence data; Extracting posture transition data according to posture behavior action sequence data to obtain posture transition data; Matching is performed based on the posture transition data and the preset posture category template data to obtain user posture category data.

6. The method according to claim 5, characterized in that The dynamic and static posture classification includes: Extracting acceleration features and speed features based on the user motion detection data to obtain user motion acceleration feature data and user motion feature data; Preliminarily classify the user motion detection data according to the user motion acceleration feature data and the user motion feature data to obtain primary user dynamic and static posture data; Perform multi-joint consistency analysis and center of gravity stability analysis based on the user motion detection data to obtain the user motion multi-joint consistency data and the user motion center of gravity stability data respectively; Perform secondary classification on the primary user's dynamic and static posture data according to the user's motion multi-joint consistency data and the user's motion center of gravity stability data to obtain secondary user's dynamic and static posture data; Performing time series division on the user motion detection data to obtain user motion time series division data; Calculate the activity change rate based on the user's motion time series data to obtain the user's motion activity change rate data; Perform energy calculation based on the user's motion activity change rate data to obtain the user's motion activity change rate energy data; Energy-weighted dynamic and static posture classification is performed on the secondary user dynamic and static posture data according to the preset time series division parameter data corresponding to the user movement time series division data and the energy data of the user movement activity change rate to obtain the user dynamic and static posture data.

7. The method according to claim 4, characterized in that The attitude change rate calculation includes: Perform time series segmentation according to the user's limb angle data to obtain user limb angle segmentation data; Calculating the long-term and short-term change rates of the user's limb angle segmentation data to obtain limb angle long-term and short-term change rate data, wherein the limb angle long-term and short-term change rate data includes limb angle long-term change rate data and limb angle short-term change rate data; The fluctuation of the change rate is calculated based on the long-term and short-term change rate data of the limb angle, and the fluctuation data of the change rate of the limb angle is obtained; According to the short-term change rate data of the limb angle, the change rate directionality analysis is performed to obtain the limb angle change rate directionality data; According to the short-term change rate data of the limb angle, short-term change rate classification is performed to obtain the short-term change rate classification data of the limb angle; The user posture change rate data is obtained by performing feature vector conversion based on the long-term and short-term change rate data of the limb angle, the volatility data of the limb angle change rate, the directional data of the limb angle change rate and the classification data of the short-term change rate of the limb angle.

8. The method according to claim 1, characterized in that S3 include: Obtain user real-time location data through indoor positioning system; Based on the user's real-time location data, multi-source positioning data fusion is performed to obtain the user's indoor location data; According to the user behavior feature data and the user indoor location data, association matching is performed to obtain user behavior location mapping data.

9. The method according to claim 1, characterized in that: User path analysis includes: Perform path segmentation according to user behavior trajectory data to obtain user behavior path segmentation data; Calculate curvature and speed characteristics based on user behavior path segmentation data to obtain path curvature data and path speed characteristic data respectively; Perform stay point detection based on user behavior path segmentation data to obtain user stay point data; Perform path morphology analysis based on path curvature data, path speed characteristic data, and user stop point data to obtain user behavior path morphology data; Perform hot spot area detection and area switching frequency calculation based on user behavior path morphology data to obtain user behavior hot spot area data and user behavior area switching frequency data; Generate a typical user path based on the path curvature data, path speed feature data, user behavior hotspot area data, and user behavior area switching frequency data to obtain typical user path data; User path analysis data is integrated based on path curvature data, path speed feature data, user behavior hotspot area data, user behavior area switching frequency data, and typical user path data to obtain user path analysis data.

Images