Implementation method of geospatial agent

Abstract

The invention discloses an implementation method of a geographic space agent, and relates to the technical field of geographic space. Comprising the steps of acquiring data: acquiring a geographic data set, an attribute data set, a personnel information data set, an intelligent multimedia equipment data set and a monitoring equipment data set, and setting up a server: setting up a cloud server, and setting a database in the cloud server to store each data set. Through a quick query method, a customer can use a mobile terminal to transmit missing person information to intelligent multimedia equipment, a query model firstly performs feature extraction on a missing person picture, and then performs feature comparison query based on monitoring data in a shopping mall to obtain a missing person. The missing person behavior track and the final appearance position information of the target person can be obtained, the target person position can be obtained by combining the target person behavior track and the final appearance position of the target person, finally the target person position is transmitted to the mobile terminal of the customer through the wireless communication technology, and the problem encountered by the customer is rapidly solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of geospatial technology, in particular to a method for implementing a geospatial agent. BACKGROUND

[0002] With the improvement of living standards, existing shopping malls are also increasingly commercialized and scaled, so the construction of shopping malls is also getting larger and larger. People in the shopping mall are basically simple plane map display or sign to find the required business for shopping. Customers will spend a long time looking for the corresponding business, especially some customers who have clear consumption targets but are not familiar with the shopping mall. It is difficult for customers to find the desired goods in time, which will reduce their purchase experience. Some high-end shopping malls build intelligent navigation systems in the room through indoor positioning technology to accurately determine the location of customers in the shopping mall, thereby providing detailed navigation paths to help customers quickly and accurately find their destinations. However, customers need to perform some tedious operations, which is cumbersome. Therefore, the shopping mall needs a method for implementing a geospatial agent to establish its own geospatial agent and provide customers with a good shopping experience.

[0003] For example, patent publication No. CN104765924A, a city traffic emergency evacuation simulation system and control method, the system includes: a geographic space module for establishing a virtual geographic space structure; an agent module for building and managing dynamic entities in a virtual geographic space; a motion and behavior module for simulating vehicle behavior and describing vehicle behavior according to embedded motion behavior models to obtain vehicle motion processes and target planning for vehicle motion processes; a simulation kernel module for planning different types of agents, complete workflows between agents, and evacuation paths in normal and emergency states, and capturing macroscopic data generated by the agent module; a human-computer interaction module for presenting simulation results of the city traffic emergency evacuation simulation system to users, and having the advantages of accuracy and reliability for emergency traffic evacuation.

[0004] The above technology focuses on using multiple agents combined with a geographic space structure to simulate the scene during city traffic emergencies, giving corresponding evacuation paths based on the scene and feeding back to the user, which can quickly evacuate the city traffic. However, it can only be used for macroscopic scheduling and cannot be based on different feedback for different customer needs in the shopping mall. At the same time, children or the elderly often go missing in shopping malls, and it is also difficult to find missing persons based on the characteristics of missing persons. Therefore, the above technology cannot be well applied to this specific scenario of shopping malls. SUMMARY

[0005] The present application aims to provide a method for implementing a geospatial agent to solve the problems raised in the background art.

[0006] To achieve the above object, the present application provides the following technical solutions: an implementation method of a geographic space agent, the method comprising:

[0007] Data acquisition: acquire geographic data sets, attribute data sets, personnel information data sets, intelligent multimedia device data sets, and monitoring device data sets;

[0008] Server building: build a cloud server, and the cloud server is provided with a database for storing various data sets;

[0009] Creation of a space agent: establish a geographic space structure based on the geographic data sets and the attribute data sets, and establish a space agent based on the geographic space structure;

[0010] Intervention carrier: based on the space agent, integrate the space agent into an intelligent multimedia device;

[0011] Creation of a command set: the space agent outputs corresponding information according to the command set;

[0012] The method for creating the space agent comprises:

[0013] S1: Building a geographic space structure, using GIS technology based on geographic data sets and attribute data sets to obtain a geographic space structure;

[0014] S2: Data acquisition: build a private local area network based on a cloud server, establish contact between the cloud server and multiple intelligent multimedia devices and multiple monitoring devices through the private local area network, and obtain monitoring data sets;

[0015] S3: Creation of a query model: establish a query model based on the monitoring data sets and the personnel information data sets;

[0016] S4: Integrated output: obtain a space agent based on the geographic space structure and the query set, the space agent can obtain a target location path according to a target path generation method, and the space agent can obtain a target personnel location according to a fast query method.

[0017] Further, the method for creating the query model comprises:

[0018] A1: Establish a learning library to collect the monitoring data sets and the personnel information data sets, label the image information in the monitoring data sets and the image information in the personnel information data sets, and divide the learning library into a learning set, a training set, and a test set;

[0019] A2: Select a convolutional neural network as a feature extractor to extract target features in the image information;

[0020] A3: The feature extractor uses the training set and selects a corresponding query model for training and optimization, thereby obtaining an initial model for analyzing human appearance features.

[0021] A4: Evaluate the performance of the initial model using cross-validation techniques, and adjust various hyperparameters based on the evaluation results to optimize the query model.

[0022] Further, the target path generation method includes:

[0023] B1: Information receiving, a separate processor is set on the intelligent multimedia device to receive information and process execution information;

[0024] B2: Information confirmation, based on the execution information, the target location coordinate information and the current coordinate information of the intelligent multimedia device are confirmed based on the geographic space structure;

[0025] B3: Path generation, based on the target location coordinate information and the current coordinate information of the intelligent multimedia device, a path finding algorithm is used to generate a finding path, i.e. a target location path;

[0026] B4: Information output, the target location path is transmitted to the mobile terminal by using wireless communication technology.

[0027] Further, the execution information creation method includes:

[0028] V1: Function determination, determine the function of the required spatial agent through the interactive interface of the intelligent multimedia device;

[0029] V2: Information input, input target information based on the interactive interface of the intelligent multimedia device;

[0030] V3: Information integration, based on the target information and the function of the required spatial agent, the execution information is obtained.

[0031] Further, the rapid query method includes:

[0032] C1: Select function, determine the function of the required spatial agent through the interactive interface of the intelligent multimedia device;

[0033] C2: Personnel information input, input target personnel information through the interactive interface of the intelligent multimedia device in cooperation with the mobile terminal;

[0034] C3: Personnel position output, based on the target personnel information, the target personnel behavior trajectory and the last appearance location of the target personnel are obtained by using the query model in cooperation with the monitoring data set, the target personnel position is obtained by combining the target personnel behavior trajectory and the last appearance location of the target personnel, and the target personnel position is transmitted to the mobile terminal by using wireless communication technology.

[0035] Further, the personnel behavior trajectory acquisition method includes:

[0036] T1: Based on the query model, multiple target personnel collection images are obtained from the monitoring data set, and the target personnel appearance time is sorted;

[0037] T2: The feature points of each collection image are tracked using the optical flow method. By changing the position of the feature points in different images, the planar motion trajectory of the feature points in the two-dimensional plane can be obtained;

[0038] T3: Build a three-dimensional trajectory, use stereo vision technology and establish three-dimensional information based on multiple collection images to obtain the three-dimensional motion trajectory of the feature points in three-dimensional space;

[0039] T4: Trajectory optimization, using spatial averaging method to optimize the planar motion trajectory and three-dimensional running trajectory, so that the planar motion trajectory and three-dimensional running trajectory are close to the real situation.

[0040] Further, the method for obtaining the last appearance location of the target personnel includes:

[0041] Y1: According to the sorting result of the multiple target personnel collection images, the last collection image of the target personnel is selected as the last appearance collection image of the target personnel;

[0042] Y2: Based on the last appearance collection image of the target personnel, the image processing library is used for preprocessing and feature extraction to obtain a matching photo;

[0043] Y3: Based on the matching photo, the image processing library is used to match the monitoring data set to obtain the corresponding monitoring data of the matching picture;

[0044] Y4: Based on the metadata of the corresponding monitoring data of the matching picture, the corresponding monitoring device position information, i.e. the last appearance location information of the target personnel, is obtained by cooperating with the geographic space structure.

[0045] Further, the method for creating the command set includes:

[0046] D1: Wake-up command, set multiple trigger keywords to match and establish contact with each command;

[0047] D2: Command organization, based on multiple trigger keywords, add corresponding input parameters, and let the corresponding trigger keywords match and establish contact with the corresponding multiple input parameters. The corresponding trigger keywords and the corresponding different input parameters can obtain different output formats;

[0048] D3: Command output, based on the intelligent multimedia device, use voice or text to output different output formats.

[0049] Further, the mobile terminal includes but is not limited to smart phones and smart watches, and the wireless communication technology includes but is not limited to Bluetooth transmission and NFC transmission.

[0050] Compared with the prior art, the application has the beneficial effects that:

[0051] By the target path generation method, the route searching function of the intelligent multimedia device can be selected by voice or touch screen, and then the target merchant or target commodity is input by voice or touch screen input method, the corresponding merchant coordinate information is obtained by querying the target merchant or target commodity, and the corresponding route searching algorithm can be used to obtain the required searching path based on the current coordinate of the intelligent multimedia device and the coordinate information of the corresponding merchant, and then the searching path is transmitted to the mobile terminal of the customer by using wireless communication technology, so that the customer does not need to perform complex operations, which helps the customer to quickly find the target merchant to purchase the required commodity, and improves the customer shopping experience.

[0052] By the quick query method, the personnel query function of the intelligent multimedia device can be selected by voice or touch screen, the customer can transmit the photo of the missing person to the intelligent multimedia device by using the mobile terminal, and the appearance features of the missing person are input by voice or touch screen to assist, then the query model extracts the features of the missing person photo, and then the features are compared and queried based on the monitoring data in the shopping mall to obtain the missing person, and the behavior trajectory of the missing person is obtained by the method of obtaining the behavior trajectory of the missing person through the personnel behavior trajectory, and the last appearance position information of the target person is obtained by the method of obtaining the last appearance position of the target person, so that the target person position is obtained by combining the target person behavior trajectory and the last appearance position of the target person, and finally the target person position is transmitted to the mobile terminal of the customer by using wireless communication technology, so that the problem of missing of the customer's relatives and friends is quickly solved. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 is the structural principle diagram of the application;

[0054] Figure 2 is the space intelligent body creation process schematic diagram of the application;

[0055] Figure 3 is the 1F plane schematic diagram of the shopping mall of the application;

[0056] Figure 4 is the schematic diagram of the intelligent multimedia device of the application. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0058] The implementation method of the geospatial agent can establish a geospatial agent belonging to a shopping mall, load the spatial agent into a smart multimedia device, establish the geospatial structure of the shopping mall by using GIS technology, and then query the positions of corresponding merchants by voice input and touch screen input information, and broadcast the positions of the corresponding merchants by voice. Further, the target path generation method can be used to obtain the path of the target merchant, and the wireless communication technology can be used to transmit the path to the mobile terminal of the customer. In addition, the query model can be used in cooperation with the fast query method to query the last appearance position of the missing person and the behavior trajectory of the missing person, so as to help the customer quickly query the scattered or missing relatives and friends, improve the customer's consumption experience in the shopping mall, and the like. For example, when the customer needs to buy a pair of sneakers, the customer inputs "I want to buy a pair of sneakers" by voice or text input through the smart multimedia device. Based on the above information, the geospatial intelligence quickly queries the merchants selling sneakers in the shopping mall, and displays the merchants by voice or text through the smart multimedia device. When the customer inputs "navigate to the Nike store" by voice or text through the smart multimedia device, the finding path from the current position to the Nike store can be obtained according to the target path generation method, and the finding path is transmitted to the mobile terminal of the customer through the wireless communication technology. When the customer inputs "query missing person" by voice or text through the smart multimedia device, the behavior trajectory and the last appearance position of the target person can be obtained according to the fast query method, and the behavior trajectory and the last appearance position are transmitted to the mobile terminal through the wireless communication technology, thereby effectively improving the customer's consumption experience.

[0059] As shown in Embodiment 1, Figures 1-4 The present application provides a technical solution: an implementation method of a geospatial agent, which comprises:

[0060] Data acquisition: acquire a geospatial data set, an attribute data set, a personnel information data set, a smart multimedia device data set, and a monitoring device data set. The geospatial data set includes a shopping mall floor plan, merchant positions, and passage layout. The attribute data set includes store names, positions, business categories, brands, and promotion activity time information of the merchants. The personnel information data set includes shopping mall staff information and merchant staff information. The smart multimedia device data set includes smart multimedia device numbers. The monitoring device data set includes monitoring device numbers.

[0061] Server building: build a cloud server, and the cloud server is provided with a database for storing various data sets. The cloud server is used for overall planning of multiple smart multimedia devices.

[0062] Creating a spatial agent: based on the geographic dataset and the attribute dataset, a geographic spatial structure is established, and a spatial agent is established based on the geographic spatial structure, the geographic dataset and the attribute dataset are cleaned and preprocessed, then the geographic dataset and the attribute dataset are integrated, and the two are associated through a unique identifier (such as ID), then the spatial data is combined with the attribute data using geographic information system (GIS) software to form a complete geographic spatial dataset, and the spatial analysis function of the GIS software is used to analyze the spatial relationship in the geographic spatial dataset, according to the analysis result, the geographic spatial structure in the shopping mall is identified, the geographic spatial structure is expressed in the form of a model or a chart using GIS software or other visualization tools, and the accuracy and rationality of the established geographic spatial structure are verified through actual observation and data analysis, and the structure is modified and adjusted according to the verification result, the GIS can select ArcGIS platform to support data storage, query, analysis and visualization, then a query model is established and a path finding algorithm, a target path generation method and a fast query method are introduced to establish a spatial agent;

[0063] Intervention carrier: based on the spatial agent, the spatial agent is integrated into the intelligent multimedia device;

[0064] Create a command set: the spatial agent outputs corresponding information according to the command set;

[0065] The method for creating the spatial agent comprises:

[0066] S1: Building a geographic spatial structure, based on the geographic dataset and the attribute dataset, a geographic spatial structure is obtained using GIS technology, S2: Data acquisition: based on a cloud server, a special local area network is built, the cloud server is connected with a plurality of intelligent multimedia devices and a plurality of monitoring devices through the special local area network to obtain a monitoring dataset, S3: Creating a query model: based on the monitoring dataset and a personnel information dataset, a query model is established, S4: Integrated output: based on the geographic spatial structure and the query set, a spatial agent is obtained, the spatial agent can obtain a target location path according to the target path generation method, and the spatial agent can obtain a target personnel location according to the fast query method.

[0067] It should be noted that the method for building the special local area network comprises: N1: Information acquisition, obtaining basic networking information of the intelligent multimedia device and the monitoring device, the basic networking information includes device IP, N2: Device connection, the cloud server automatically allocates a communication channel to match the IP of the intelligent multimedia device with the IP of the monitoring device, N3: Configuration networking, the cloud server configures corresponding network services and security measures for the communication channel to create a special local area network.

[0068] The method for creating the query model comprises:

[0069] A1: Collecting the monitoring data set and the personnel information data set, labeling the image information in the monitoring data set and the image information in the personnel information data set, and dividing the learning library into a learning set, a training set, and a test set, A2: Selecting a convolutional neural network as a feature extractor to extract target features in the image information, A3: The feature extractor uses the training set and selects a corresponding query model for training and optimization, thereby obtaining an initial model for analyzing human appearance features, A4: Using cross-validation technology to evaluate the performance of the initial model, and adjusting various hyperparameters based on the evaluation results to optimize the query model.

[0070] It should be noted that the training optimization method includes: M1: One-time optimization, in the training process of the query model, the backpropagation algorithm calculates the loss according to the difference between the predicted results of the model and the true labels, then the algorithm propagates these losses back to each layer of the model, calculates the gradient of each layer parameter, and finally updates the parameters of the model according to the calculated gradient, in order to reduce the loss in the next iteration, M2: Second optimization, in the training of the query model, select the cross-entropy loss function as the objective function to measure the difference between the predicted results of the model and the true labels, then select a suitable optimizer such as Adam or RMSprop to update the parameters of the model according to the calculated gradient, the selection of the optimizer will affect the training speed and final performance of the model, therefore it needs to be debugged according to the specific task and data set, M3: Third optimization, in the training process of the query model, divide the data set into training set, validation set and test set, after each training period, use the validation set to evaluate the model and calculate its performance on the validation set (such as accuracy, recall rate, etc.), according to the performance change of the validation set, it can be judged whether the model has overfitting or underfitting problem, and the hyperparameters are adjusted accordingly (such as reducing the learning rate, increasing the regularization term, etc.), through continuous iteration and adjustment, a query model with good performance on the validation set can be finally obtained.

[0071] The target path generation method includes:

[0072] B1: Information receiving, a separate processor is provided on the intelligent multimedia device to receive information and process execution information, B2: Information confirmation, based on the execution information, the target position coordinate information and the current coordinate information of the intelligent multimedia device are confirmed based on the geographic space structure, B3: Path generation, based on the target position coordinate information and the current coordinate information of the intelligent multimedia device, a path finding algorithm is used to generate a finding path, i.e. a target position path, B4: Information output, the target position path is transmitted to the mobile terminal by using wireless communication technology.

[0073] It should be noted that the path finding algorithm is A*(A-Star), the customer issues an execution information through the intelligent multimedia, the execution information is based on the geographic space structure and the ArcGIS platform to query the target position coordinate information and the current coordinate information of the intelligent multimedia device, and then the current position and the target position are based on the geographic space structure and the A*(A-Star) path finding algorithm to obtain the best path finding path.

[0074] Further, the execution information creation method comprises:

[0075] V1: function determination, the function of the required space intelligent body is determined through the interactive interface of the intelligent multimedia device, V2: information input, the target information is input based on the interactive interface of the intelligent multimedia device, V3: information integration, the execution information is obtained based on the target information and the function of the required space intelligent body.

[0076] It should be noted that the function determination is confirmed by the customer through voice and touch screen interaction with the intelligent multimedia device, the function to be confirmed in the execution information is the path finding function, on the basis of which, the target information is added, and finally the execution information is obtained.

[0077] Further, the rapid query method comprises:

[0078] C1: selection function, the function of the required space intelligent body is determined through the interactive interface of the intelligent multimedia device, C2: personnel information input, the target personnel information is input through the interactive interface of the intelligent multimedia device in cooperation with the mobile terminal, C3: personnel position output, the target personnel behavior track and the last appearance place of the target personnel are obtained based on the target personnel information by using the query model in cooperation with the monitoring data set, the target personnel position is obtained in combination with the target personnel behavior track and the last appearance place of the target personnel, and the target personnel position is transmitted to the mobile terminal through wireless communication technology.

[0079] It should be noted that the target personnel information includes a photo or a video of the target personnel, and a text description of the appearance of the current target personnel information, based on the photo or the video of the target personnel, the query model can extract the facial features, body features and clothing features of the target personnel, and then the appearance of the current target personnel information is described in combination with the above text, and finally the target personnel is found through the query model in cooperation with the monitoring data, and the target personnel behavior track and the last appearance place of the target personnel are obtained and transmitted to the mobile terminal of the customer.

[0080] Further, the personnel behavior track acquisition method comprises:

[0081] T1: Obtain multiple target personnel collection images based on the query model combined with the monitoring dataset, and sort them according to the appearance time of the target personnel, T2: Track the feature points of each collection image using the optical flow method, and by the position change of the feature points in different images, the planar motion trajectory of the feature points in the two-dimensional plane can be obtained, T3: Build a three-dimensional trajectory, use stereo vision technology and establish three-dimensional information based on multiple collection images to obtain the three-dimensional motion trajectory of the feature points in the three-dimensional space, T4: Trajectory optimization, use the spatial average method to optimize the planar motion trajectory and the three-dimensional running trajectory, and make the planar motion trajectory and the three-dimensional running trajectory close to the real situation.

[0082] It should be noted that the retrieved target personnel images are sorted according to the timestamp in the image or the timeline of the camera, and then the FAST corner detection algorithm is used to extract the feature points in the collection image, and then the optical flow method is used to calculate the displacement of these feature points between consecutive frames, thereby obtaining the two-dimensional trajectory motion trajectory of the target personnel, or the similar regions of multiple collection images are located, and then the optical flow is calculated through the displacement of the similar regions. The feature points can be human body parts, clothing appearance, etc. By tracking the position change of these feature points in different pictures, the two-dimensional trajectory motion trajectory of the target personnel can also be obtained. Stereo vision technology captures images from multiple cameras, uses stereo matching or depth estimation technology to obtain three-dimensional information of objects in the scene, uses multiple cameras to capture images of the target personnel at the same time to obtain data from different angles, and based on the three-dimensional information obtained by stereo vision technology, the tracking results of the feature points are combined to calculate the motion trajectory of the target personnel in the three-dimensional space.

[0083] Further, the method for obtaining the last appearance location of the target personnel comprises:

[0084] Y1: Select the last collection image of the target personnel as the last appearance collection image of the target personnel according to the sorting result of the multiple collection images of the target personnel, Y2: Preprocess and feature extraction based on the last appearance collection image of the target personnel using the image processing library to obtain a matching photo, Y3: Match the matching photo with the monitoring dataset based on the image processing library to obtain the corresponding monitoring data of the matching photo, Y4: Obtain the corresponding monitoring device location information, i.e. the last appearance location information of the target personnel, based on the metadata of the corresponding monitoring data of the matching photo and the geographic space structure.

[0085] It should be noted that the image processing library is OpenCV, and the metadata includes the ID, location information (such as installation floor, corridor), viewing direction, etc. of the monitoring device. Through these information, the specific location of the last appearance of the target personnel can be further determined, and the monitoring layout around the location can be understood.

[0086] Further, the method for creating the command set comprises:

[0087] D1: wake-up command, set multiple trigger keywords to match each command to establish a connection, D2: command organization, based on multiple trigger keywords, add corresponding input parameters respectively, and let the corresponding trigger keywords match the corresponding multiple input parameters to establish a connection, and the corresponding trigger keywords and the corresponding different input parameter organization can obtain different output formats. D3: command output, based on intelligent multimedia devices, output different output formats using voice or text.

[0088] It should be noted that the space agent uses natural language processing technology to analyze the customer's question in order to extract key information from the user's customer, and uses natural language generation technology to generate an answer in the form of natural language output format.

[0089] Further, the mobile terminal includes but is not limited to smart mobile terminals such as smartphones and smartwatches, and wireless communication technology includes but is not limited to Bluetooth transmission and NFC transmission.

[0090] It should be noted that Bluetooth is a short-range wireless communication technology that can transmit data between devices, and NFC transmission is a wireless communication technology based on radio frequency identification (RFID) technology, allowing devices to communicate within a short distance (usually between a few centimeters to tens of centimeters).

[0091] Although embodiments of the present application have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A method of implementing a geospatial agent, the method comprising: The method comprises: Acquiring data: acquiring geographic data sets, attribute data sets, personnel information data sets, intelligent multimedia device data sets, and monitoring device data sets; Server building: building a cloud server, and setting a database in the cloud server to store various data sets; Creating a spatial agent: establishing a geographic space structure based on the geographic data sets and the attribute data sets, and establishing a spatial agent based on the geographic space structure; Intervening carriers: integrating the spatial agent into the intelligent multimedia device based on the spatial agent; Creating a command set: the spatial agent outputs corresponding information according to the command set; The method for creating the spatial agent comprises: S1: Building a geographic space structure, obtaining the geographic space structure by using GIS technology based on the geographic data sets and the attribute data sets; S2: Data acquisition: building a special local area network based on the cloud server, establishing contact between the cloud server and multiple intelligent multimedia devices and multiple monitoring devices through the special local area network, and obtaining monitoring data sets; S3: Creating a query model: establishing a query model based on the monitoring data sets and the personnel information data sets; S4: Integrated output: obtaining a spatial agent based on the geographic space structure and the query set, the spatial agent can obtain a target location path according to a target path generation method, and the spatial agent can obtain a target personnel location according to a rapid query method.

2. The method of claim 1, wherein: The method for creating the query model comprises: A1: Establishing a learning library to collect the monitoring data sets and the personnel information data sets, labeling image information in the monitoring data sets and image information in the personnel information data sets, and dividing the learning library into a learning set, a training set, and a test set; A2: Selecting a convolutional neural network as a feature extractor to extract target features in the image information; A3: The feature extractor uses the training set and selects a corresponding query model for training and optimization, thereby obtaining an initial model for analyzing human appearance features; A4: Using a cross-validation technique to evaluate the performance of the initial model, and adjusting various hyperparameters based on the evaluation results to optimize the query model.

3. The method of claim 1, wherein: The target path generation method comprises: B1: Information receiving, a separate processor is arranged on the intelligent multimedia device to receive information and process execution information; B2: Information confirmation, confirming target location coordinate information and intelligent multimedia device current coordinate information based on the execution information and the geographic space structure; B3: Path generation, generating a finding path, i.e., a target location path, based on the target location coordinate information and the intelligent multimedia device current coordinate information by using a pathfinding algorithm; B4: Information output, transmitting the target location path to a mobile terminal by using wireless communication technology.

4. The method of claim 3, wherein: The method for creating the execution information comprises: V1: Function determination, determining the function of the required spatial agent through the interactive interface of the intelligent multimedia device; V2: Information input, inputting target information based on the interactive interface of the intelligent multimedia device; V3: Information integration, obtaining execution information based on the target information and the function of the required spatial agent.

5. The method of claim 1, wherein: The rapid query method comprises: C1: Function selection, determining the function of the required spatial agent through the interactive interface of the intelligent multimedia device; C2: personnel information input, through the interactive interface of the intelligent multimedia device to input the target personnel information with the mobile terminal; C3: personnel position output, based on the target personnel information, the target personnel behavior trajectory and the last appearance location of the target personnel are obtained by using the query model and the monitoring data set, the target personnel position is obtained by combining the target personnel behavior trajectory and the last appearance location of the target personnel, and the target personnel position is transmitted to the mobile terminal through wireless communication technology.

6. The method of claim 5, wherein: The method for obtaining the personnel behavior trajectory comprises: T1: based on the query model and the monitoring data set, a plurality of target personnel collection images are obtained, and the target personnel appearance time is sorted; T2: the feature points of each collection image are tracked by using the optical flow method, and the plane motion trajectory of the feature points in the two-dimensional plane can be obtained by the position change of the feature points in different images; T3: a three-dimensional trajectory is built, three-dimensional information is established based on the plurality of collection images by using the stereo vision technology, and the three-dimensional motion trajectory of the feature points in the three-dimensional space is obtained; T4: trajectory optimization, the plane motion trajectory and the three-dimensional running trajectory are optimized by using the spatial average method, so that the plane motion trajectory and the three-dimensional running trajectory are close to the real situation.

7. The method according to claims 5 and 6, wherein: The method for obtaining the last appearance location of the target personnel comprises: Y1: the last collection image of the target personnel is selected as the last appearance collection image of the target personnel according to the sorting result of the plurality of target personnel collection images; Y2: based on the last appearance collection image of the target personnel, the image processing library is used for pre-processing and feature extraction to obtain a matching photo; Y3: based on the matching photo, the image processing library is used to match with the monitoring data set to obtain the corresponding monitoring data of the matching picture; Y4: based on the metadata of the corresponding monitoring data of the matching picture, the corresponding monitoring device position information, i.e. the last appearance location information of the target personnel, is obtained by cooperating with the geographic space structure.

8. The method of claim 1, wherein: The method for creating the command set comprises: D1: wake-up command, a plurality of trigger keywords are matched with each command to establish a connection; D2: command organization, based on the plurality of trigger keywords, corresponding input parameters are added, and the corresponding trigger keywords are matched with the corresponding plurality of input parameters to establish a connection, so that different output formats can be obtained by organizing the corresponding trigger keywords and the corresponding different input parameters; D3: command output, the intelligent multimedia device outputs different output formats by using voice or text.

9. The method of claim 3, wherein: The mobile terminal comprises a smart phone and a smart watch, and the wireless communication technology comprises a Bluetooth transmission mode and an NFC transmission mode.

Images