Scenic area visitor flow prediction method and system, medium, equipment and terminal

By deploying facial and behavioral recognition cameras in scenic areas and combining them with deep learning models, the system can monitor and predict visitor flow in real time, solving the problems of refined and predictive visitor flow monitoring and enabling intelligent management and improved visitor experience.

CN118942036AActive Publication Date: 2024-11-12HUBEI UNIV FOR NATITIES

Patent Information

Application Number
CN202410996559.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-24
Publication Date
2024-11-12
Estimated Expiration
2044-07-24

AI Technical Summary

Technical Problem

The existing methods for monitoring visitor flow in scenic areas are not comprehensive or mature enough, making it impossible to monitor visitor behavior in a refined manner or predict visitor gathering behavior at special attractions and popular spots. This results in insufficient intelligence in scenic area management and affects visitor satisfaction.

Method used

A deep learning-based method for predicting visitor flow in scenic areas is adopted. This method uses facial and behavioral recognition cameras to identify and analyze the behavior of tourists. It combines deep learning models to predict the location of tourists and the distribution of visitor flow, adjusts the visitor management plan in real time, and uses fiber optic networks to transmit data for real-time calculation and early warning in the central control center.

Benefits of technology

It enables high-precision real-time monitoring and prediction of visitor flow in scenic areas, improving the efficiency of scenic area management and visitor experience. It can predict peak periods in advance and conduct scientific crowd control, thereby enhancing the safety and operational management level of scenic areas.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118942036A_ABST
    Figure CN118942036A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of spatial and temporal distribution characteristic analysis of crowd gathering behaviors in public places and public safety management, and discloses a scenic spot visitor flow prediction method and system, a medium, equipment and a terminal. Analyzing the sightseeing habit of the tourist based on deep learning, and estimating the travel position of the tourist; and according to the estimated travel positions of the tourists at the following time points, predicting the dynamic distribution of the visitor flow of the scenic area, pre-judging the crowded time periods of the special scenic spots and the red spots in the scenic area, and starting a tourist guidance plan. According to the invention, the face recognition accuracy is high; and the behavior characteristics of the tourists are identified and transmitted to the server as shared information, so that the accuracy of predicting the number of the tourists in each scenic spot in the scenic area and the advancement of tourist diversion and guidance by the whole system are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of temporal and spatial distribution characteristics analysis of crowd gathering behavior in public places and public safety management technology, and in particular to a method, system, medium, equipment and terminal for predicting the flow of people in a scenic spot. Background Art

[0002] At present, with the improvement of people's living standards and the pursuit of high-quality life, leisure tourism has become one of the main ways for people to relax and enjoy life. Preventing excessive crowding in scenic spots is a necessary means to protect the safety of tourists' lives and property and reduce the risk of infectious disease transmission. It is also an important measure to strengthen the intelligent management of scenic spots and enhance tourists' experience. Based on infrared sensor technology, wireless radio frequency identification technology, wireless communication positioning technology, video image recognition technology, etc., the statistics and analysis of scenic spot passenger flow are the main methods for monitoring the distribution of scenic spot passenger flow and tourist behavior habits, but the integrity and maturity of the technology need to be improved. It cannot meet the requirements of refined monitoring of scenic spot passenger flow and multi-dimensional description of tourist behavior. It cannot predict the gathering behavior of tourists at special scenic spots and popular spots in scenic spots. It has not fundamentally enhanced the intelligence of scenic spot management and effectively improved the satisfaction of tourist experience.

[0003] Through the above analysis, the problems and defects of the existing technology are as follows: the integrity and maturity of the existing scenic spot passenger flow monitoring method need to be improved, it cannot meet the requirements of refined monitoring of scenic spot passenger flow and multi-dimensional description of tourist behavior, it cannot predict the gathering behavior of tourists at special attractions and popular spots in the scenic area, and it cannot fundamentally enhance the intelligence level of scenic spot management and effectively improve the satisfaction of tourists' experience. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a method, system, medium, device and terminal for predicting the flow of people in a scenic spot, and more particularly, relates to a method, system, medium, device and terminal for predicting the flow of people in a scenic spot based on deep learning.

[0005] The present invention is implemented as follows: a method for predicting the flow of people in a scenic area, which includes: using face and behavior recognition cameras to perform group calibration and identity recognition on the facial information of tourists entering the scenic area; analyzing the tourists' sightseeing habits based on deep learning, and estimating the tourists' next travel locations at different times; based on the time and location estimation of the tourists' entire tour, predicting the dynamic distribution of the flow of people at each point in the scenic area at each time, predicting the crowded periods of special attractions and popular spots in the scenic area, and judging whether to activate the tourist diversion plan.

[0006] Further, the scenic spot flow prediction method includes the following steps:

[0007] Step 1: Deploy face and behavior recognition cameras according to the scenic area’s tour routes, identify faces and perform identity calibration through video and image acquisition, and identify tourist behavior characteristics through dynamic analysis;

[0008] Step 2: The image data of the camera is transmitted to the server in real time via optical fiber. The server analyzes the tourist information collected by multiple cameras and calibrates the tourists into groups.

[0009] Step 3: The server analyzes the data uploaded by all cameras in the scenic area to determine the specific area of ​​the scenic area where all groups entering the scenic area are currently located;

[0010] Step 4: The server determines the interval location of the tourist group at the current moment and the behavioral characteristics of the tourists in each group, determines the number of tourists at each scenic spot with high traffic flow at the current and next moments, and determines whether to start the tourist diversion plan;

[0011] Step 5: The server updates the estimated number of tourists at each scenic spot on the tour route in real time based on the actual travel data of the tourists, adjusts the tourist diversion plan in real time and determines whether to provide voice prompts.

[0012] Furthermore, in step 4, the server determines the interval position of the tourist group at the current moment and the behavioral characteristics of the tourists in each group, and determines the estimated number of tourists for each scenic spot and Internet celebrity spot in the next time period according to the time series; if the estimated number of tourists for a certain scenic spot or Internet celebrity spot exceeds the threshold for accommodating tourists, the server sends an early warning message to the scenic spot operation management department to apply for the initiation of the tourist diversion plan;

[0013] The length of the next time period can be set arbitrarily, and the typical value is 0.5 hours.

[0014] Furthermore, in step 4, the tourist diversion plan is a diversion program pre-set in the scenic area control center, which reminds tourists to adjust their tour progress through voice IP broadcasts on the tour route, including:

[0015] ① Provide voice IP broadcast reminders to groups gathering at popular spots or those with slow tour speeds;

[0016] ② Provide voice IP broadcast reminders for attractions that are of low tourist interest, missed attractions, or have a large passenger capacity, or for groups with a fast tour speed.

[0017] Furthermore, in step five, the server updates the estimated number of tourists at each scenic spot in real time based on the actual travel data of the tourists, and adjusts the tourist diversion plan in real time; the diversion prompt information of each voice IP broadcast is adjusted in real time according to the behavioral characteristics of the current group of tourists. If the estimated number of tourists at all attractions is lower than the tourist capacity threshold of the attraction, no voice prompt will be given.

[0018] Another object of the present invention is to provide a scenic spot passenger flow prediction system using the scenic spot passenger flow prediction method, the scenic spot passenger flow prediction system comprising a face and behavior recognition camera, a fiber optic network, a fiber optic switch, a storage server and an application server.

[0019] Face and behavior recognition cameras are installed at the entrance of the scenic area, walking trails, special places, scenic spots and exits, including fixed-angle gun cameras and ball cameras with pan / tilts. The ball cameras are used for high-altitude shooting and tracking shooting.

[0020] The fiber optic switch is used to connect all facial and behavior recognition cameras in the scenic area to the server;

[0021] The servers are divided into application servers and storage servers, which are directly connected to the fiber switches;

[0022] The storage server includes sufficient disk arrays to meet the scenic spot's storage requirements for video data;

[0023] The application server has sufficient computing speed and stores programs for numbering the face information collected by each face and behavior recognition camera, and for calculating the estimated number of tourists at each scenic spot at each moment in the next period of time in real time, and comparing it with the tourist accommodation threshold determined in advance by the scenic spot;

[0024] The application server regularly broadcasts all calibrated identity numbers and facial information of the day to each facial recognition camera through the fiber optic switch, and the update cycle does not exceed 10 seconds.

[0025] Furthermore, the face and behavior recognition cameras calibrate the face information, and upload it to the server immediately after successful calibration; the server assigns identity numbers to all calibrated face information of the day in real time, and broadcasts the identity numbers and face image information to each face recognition camera, with an update interval of no more than 10 seconds / time;

[0026] The face and behavior recognition camera stores the face information and ID number broadcast by the server in the local storage. When a new face enters the camera, it captures the behavior of the tourist and compares it with the face information in the storage. If it already exists, the behavior characteristics of the tourist are stored according to the ID number and face information saved in the local storage, and the video and recognition data are transmitted to the server in real time through optical fiber; if there is no face information and ID number of the tourist in the local storage, the video, face recognition and tourist behavior characteristic information are directly transmitted to the server. The local storage only stores the tourist identity code, the corresponding face image information and the tourist behavior characteristic information, and all are cleared at 24:00 every day.

[0027] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the scenic spot passenger flow prediction method.

[0028] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the scenic spot passenger flow prediction method.

[0029] Another object of the present invention is to provide an information data processing terminal, which is used to implement the scenic spot passenger flow prediction system.

[0030] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0031] First, the present invention provides a scenic spot passenger flow monitoring system and method based on deep learning. It adopts face recognition cameras to identify the identities of tourists entering the scenic spot, and installs a certain density of cameras on the walking trails. Based on deep learning, the tourists' sightseeing habits are analyzed, and the travel positions of tourists at all time series points are estimated. On this basis, the dynamic distribution of the scenic spot passenger flow is predicted, and the crowded periods of special attractions and Internet celebrity spots in the scenic spot are predicted in advance. Safety precautions are taken in advance, and tourists are guided in a scientific, reasonable and beneficial manner, so as to enhance the tourists' experience satisfaction and the reputation of scenic spot management.

[0032] The scenic spot flow monitoring system based on deep learning provided by the present invention adopts a face and behavior recognition camera. The camera processes the collected original video and image information, and the face recognition accuracy is high; the behavior characteristics of tourists are identified and transmitted to the server as shared information; the camera and the server communicate through optical fiber, with a high communication rate, a large communication bandwidth and a strong anti-interference ability; in the centralized control center, the storage server and the application server are independent, which improves the computing speed of the application server.

[0033] The scenic spot passenger flow prediction system based on deep learning provided by the present invention adopts a fiber optic network architecture, which has the characteristics of high reliability and strong stability, thereby improving the accuracy of the entire system in predicting the number of tourists at each scenic spot.

[0034] Second, as auxiliary evidence of the inventiveness of the claims of the present invention, it is also reflected in the following important aspects:

[0035] (1) The expected benefits and commercial value of the technical solution of the present invention after transformation are as follows: The product corresponding to the technical solution of the present invention is a must-have choice for all provincial-level scenic area management, and can be promoted and deployed in 3A and above scenic spots across the country, with broad market prospects.

[0036] (2) The technical solution of the present invention fills the technical gap in the industry at home and abroad: The technical solution of the present invention adopts a deep learning method to predict the behavior of tourists in scenic spots and estimate the number of tourists at each scenic spot. The subdivision field is a domestic first, filling the domestic technical gap.

[0037] (3) The technical solution of the present invention solves a technical problem that people have been eager to solve but have never been able to solve successfully: the technical solution of the present invention solves the problems of the existing scenic spot tourist flow monitoring and management system being imprecise, having poor real-time performance, and having no predictive data.

[0038] (4) The technical bias overcome by the technical solution of the present invention is that the existing technical solution mainly solves the real-time data of the distribution of people in various scenic spots in the scenic area, while the technical solution of the present invention is mainly aimed at the number of tourists gathered in various scenic spots in the scenic area in the next period of time, and mainly solves the problem of early guidance based on the prediction results to alleviate the problem of tourist congestion.

[0039] Third, the traditional scenic area management method has deficiencies in monitoring and predicting tourist flow, mainly relying on manual statistics and simple monitoring equipment, making it difficult to achieve real-time and accurate collection and analysis of tourist flow data. Especially during peak tourist seasons, the lack of an efficient tourist flow prediction system can easily lead to congestion in scenic areas, safety hazards and other problems, affecting tourist experience and scenic area operations.

[0040] This paper proposes a tourist flow prediction system based on face recognition technology and deep learning. By deploying face recognition cameras at key locations in scenic spots, tourist image data is collected in real time. The system uses the YOLOv8-MP model and MobileNet network for efficient face feature recognition and learning, and uses the K-means clustering algorithm to optimize small-size face feature recognition. Combined with the deep learning passenger flow prediction model (DL-TFP), the system can accurately predict the tourist flow of each scenic spot in the scenic area, providing a scientific basis for scenic area management.

[0041] The significant technical progress of the present invention is that it realizes high-precision real-time monitoring and prediction of tourist flow in scenic spots. Through the integration of deep learning and computer vision technology, the system can not only effectively identify the facial features of tourists, but also predict future tourist flow trends based on historical data, providing dynamic management strategy support for scenic spots. In addition, the introduced regional enhanced attention mechanism further improves the model's ability to process spatiotemporal series data, making the prediction results more accurate and reliable.

[0042] The application of this tourist flow prediction system has significantly improved the management efficiency and tourist experience of scenic spots. Through real-time monitoring and early warning mechanisms, scenic spots can respond to peak tourist flow in a timely manner and effectively avoid congestion and safety hazards. At the same time, the multi-dimensional data analysis function provided by the system provides strong data support for scenic spot planning, resource allocation and marketing strategy formulation, which helps scenic spots achieve sustainable development and intelligent upgrades. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0044] Figure 1 is a flow chart of a method for predicting the flow of people in a scenic spot provided by an embodiment of the present invention;

[0045] Figure 2 It is a structural block diagram of a scenic spot passenger flow prediction system provided by an embodiment of the present invention;

[0046] Figure 3 It is a network structure diagram of a tourist flow prediction model provided by an embodiment of the present invention;

[0047] In the figure: 001, face and behavior recognition camera; 002, optical fiber; 003, optical fiber switch / router; 004, storage server; 005, application server. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] In view of the problems existing in the prior art, the present invention provides a method, system, medium, device and terminal for predicting the flow of people in a scenic spot. The present invention is described in detail below with reference to the accompanying drawings.

[0050] Example 1: Face recognition and tourist flow prediction system for large tourist attractions

[0051] Equipment configuration:

[0052] 1) Face recognition camera:

[0053] Entrance: deploy no less than 3 facial recognition cameras to cover the waiting area and ticket gate area. If there are multiple ticket gates in the scenic area, increase the number of cameras as needed to ensure that each entrance is covered by more than 3 cameras.

[0054] At the exit: Install a corresponding number of facial recognition cameras according to the number of exit gates, ensuring that each exit gate has at least one camera.

[0055] Inside the scenic area: distribution is based on the length of the tour route, the density of attractions and the tourist carrying capacity. Important attractions (popular spots) must be equipped with facial recognition cameras. The density of cameras is higher in the initial section of the tourist trail, and other sections are randomly distributed.

[0056] 2) Network transmission:

[0057] The face recognition camera is connected to the fiber optic switch, router, and server via optical fiber to ensure efficient and stable data transmission.

[0058] 3) Face recognition algorithm:

[0059] The face recognition camera uses the YOLOv8-MP model and the MobileNet lightweight deep neural network, and uses K-means clustering to optimize anchors to improve the speed and accuracy of facial feature recognition.

[0060] 4) Data storage and analysis:

[0061] Storage server: configured according to the number of cameras, video parameters, and storage time requirements. The storage time shall not be less than 180 days.

[0062] Application server: Stores and runs the passenger flow prediction program, based on the deep learning tourist flow prediction model (DL-TFP), to predict and analyze the tourist flow in the scenic area.

[0063] Workflow:

[0064] 1) Data collection: Facial recognition cameras collect images of tourists and transmit the data to the control center through the fiber optic network.

[0065] 2) Data processing: In the centralized control center, the YOLOv8-MP model and MobileNet are used for face recognition, facial features extraction, matching, and statistics.

[0066] 3) Tourist flow prediction: Based on the DL-TFP model, ConvLSTM and regional enhanced attention mechanism are used to perform real-time prediction of tourist flow.

[0067] 4) Result display: The current and predicted number of tourists are displayed on the display platform of the control center through heat maps, curve graphs, bar graphs, etc., and early warning information is provided.

[0068] Example 2: Face recognition and tourist flow prediction system for medium-sized urban parks

[0069] Equipment configuration:

[0070] 1) Face recognition camera:

[0071] Entrance: No less than 3 and 2 facial recognition cameras will be deployed at the main entrance and secondary entrance of the park respectively, covering the waiting area and ticket gate area.

[0072] Exit: Depending on the number of exits, 1-2 facial recognition cameras are installed at each exit to ensure coverage of all exits.

[0073] Inside the park: cameras are densely arranged at key scenic spots (such as gardens, lakeside) and starting points of major walking trails, and randomly arranged in other areas.

[0074] 2) Network transmission:

[0075] All cameras are connected to fiber switches, routers and servers via optical fibers to ensure efficient data transmission.

[0076] 3) Face recognition algorithm:

[0077] Use the YOLOv8-MP model and MobileNet deep neural network, combined with K-means clustering to optimize anchors and improve the speed and accuracy of face recognition.

[0078] 4) Data storage and analysis:

[0079] Storage server: Configured according to the number of cameras and storage requirements to ensure that video data storage time is not less than 180 days.

[0080] Application server: runs the passenger flow prediction program and uses the DL-TFP model to predict and analyze tourist flow.

[0081] Workflow:

[0082] 1) Data collection: Facial recognition cameras collect images of visitors at the entrances, exits and inside the park, and transmit them to the control center via a fiber optic network.

[0083] 2) Data processing: In the centralized control center, the YOLOv8-MP model and MobileNet are used to perform face recognition, extract features and perform statistics.

[0084] 3) Tourist flow prediction: Based on the DL-TFP model, combined with ConvLSTM and regional enhanced attention mechanism, the tourist flow in the park is predicted in real time.

[0085] 4) Result display: The display platform of the control center displays the current and predicted number of tourists through heat maps, curve graphs and bar graphs, and provides early warning information.

[0086] These two specific examples demonstrate how to deploy facial recognition cameras in scenic spots of different sizes, and use advanced deep learning algorithms to achieve real-time monitoring and prediction of tourist flow, ensuring efficient scenic spot management and tourist safety.

[0087] like Figure 1 As shown, the method for predicting the flow of people in a scenic spot provided by the embodiment of the present invention comprises the following steps:

[0088] S101, using a face and behavior recognition camera to calibrate the face information of the visitor and identify the identity;

[0089] S102, analyzing the tourists' sightseeing habits based on deep learning, and estimating the tourists' travel locations at each time in the next period of time;

[0090] S103, based on the tourists' travel positions at various times in the next period of time, the actual location of the tourists is estimated, the dynamic distribution of the flow of people in the scenic area is predicted, the crowded periods of special scenic spots and popular spots in the scenic area are predicted, and it is determined whether to activate the tourist diversion plan.

[0091] As a preferred embodiment, the method for predicting the flow of people in a scenic spot provided by the embodiment of the present invention comprises the following steps:

[0092] 1. First, according to the scenic area's tour routes, cameras with face and behavior recognition functions are appropriately deployed. The cameras can recognize faces and perform identity calibration through video and image acquisition, and can also distinguish tourist behavior characteristics through dynamic analysis;

[0093] 2. The image data of the camera is transmitted to the server in real time through optical fiber. The server analyzes the tourist information collected by multiple cameras, calibrates the tourists into groups, and distinguishes which group a specific tourist belongs to. The number of people in a group can be 1, 2, 3, ... N, where N is a positive integer;

[0094] 3. The server analyzes the data uploaded by all cameras in the scenic area to determine the specific area of ​​the scenic area where all groups entering the scenic area are currently located;

[0095] 4. The server determines the interval location of the tourist group at the current moment and the behavioral characteristics of the tourists in each group, and determines the estimated number of tourists for each scenic spot and popular spot in the next time period (typical value is 0.5 hours) according to the time series; if the estimated number of tourists for a scenic spot or popular spot exceeds the threshold for accommodating tourists, it will send an early warning message to the scenic spot operation management department to apply for the activation of the tourist diversion plan;

[0096] 5. The tourist guidance plan is a guidance program pre-set in the scenic area control center. Specifically, it reminds tourists to adjust their tour progress through voice IP broadcasts on the tour route. There are two ways: Method 1: For people gathering at popular spots or groups with a slow tour speed, voice IP broadcasts remind "The scenery ahead is more beautiful, come on!" Method 2: For scenic spots that are not highly concerned by tourists, or are missed, or have a large passenger capacity, or groups with a fast tour speed, voice IP broadcasts remind "You must be tired, stop and breathe the pure air with super high negative oxygen ions!"

[0097] 6. The server updates the estimated number of tourists at each scenic spot in real time based on the current location data of all tourists in the park, and adjusts the tourist diversion plan in real time. The diversion prompt information of each voice IP broadcast is adjusted in real time according to the behavioral characteristics of the current group of tourists. If the estimated number of tourists at all attractions is lower than the tourist capacity threshold of the attraction, no voice prompt will be given.

[0098] like Figure 2 As shown, the scenic spot flow prediction system provided by the embodiment of the present invention includes: face and behavior recognition cameras installed at the entrance, walking trails, special places, scenic spots (including popular spots), and exits of the scenic spot, optical fiber networks, optical fiber switches, storage servers, and application servers.

[0099] Among them, face and behavior recognition cameras include fixed-angle gun cameras and ball cameras with pan / tilts. The ball cameras are mainly used for high-altitude shooting and tracking shooting.

[0100] The face and behavior recognition cameras can calibrate the face information, and upload it to the server immediately after successful calibration. The server will assign ID numbers to all calibrated face information of the day in real time, and broadcast the ID numbers and face image information to each face recognition camera, with the update interval not exceeding 10 seconds / time;

[0101] The face and behavior recognition camera stores the face information and identity number broadcast by the server in the local memory. When a new face enters the lens and captures the behavior of the tourist, it is compared with the face information in the memory. If it already exists, the behavior characteristics of the tourist are stored according to the identity number and face information saved in the local memory, and the video and recognition data are transmitted to the server in real time through optical fiber. If the local memory does not have the face information and identity number of the tourist, the video, face recognition and tourist behavior characteristic information are directly transmitted to the server; the local memory only stores the tourist identity code, the corresponding face image information and the tourist behavior characteristic information, and all are cleared at 24:00 every day;

[0102] The fiber optic switch is used to connect all facial and behavior recognition cameras in the scenic area to the server;

[0103] The servers are divided into application servers and storage servers, which are directly connected to the fiber switches;

[0104] The storage server has enough disk arrays to meet the scenic spot's storage requirements for video data;

[0105] The application server has sufficient computing speed and a program to identify the face information collected by each face and behavior recognition camera, and calculate the estimated number of tourists at each scenic spot at each moment in the next period of time in real time, and compare it with the tourist capacity threshold determined in advance for the scenic spot;

[0106] The application server regularly broadcasts all calibrated identity numbers and facial information of the day to each facial recognition camera through the fiber optic switch, and the update cycle does not exceed 10 seconds.

[0107] The technical solution of the present invention is mainly used for monitoring the flow of tourists in scenic spots, specifically for predicting the number of tourists at various scenic spots in the scenic spot in the next period of time, and providing data support for the scenic spot to carry out scientific and effective tourist guidance.

[0108] In order to facilitate tourists' visits, scenic spots generally design one-way or circular tour routes. Therefore, it is more convenient to deploy cameras on the tour routes for image acquisition. First, deploy no less than 3 face recognition cameras from the waiting area at the entrance of the scenic spot to the ticket gate. If the scenic spot is large, there are g I If there are no more than 30 ticket gates, the facial recognition cameras installed in this area should be no less than 3g I At the exit of the scenic area, according to the number of exit gates g O Install g O Face recognition cameras. In the scenic area, n face recognition cameras are installed and deployed according to the length of the tour route, the density of scenic spots, and the carrying capacity of tourists. The total number of cameras is (3g I +g O +n) items.

[0109] The n facial recognition cameras installed in the scenic area provided by the embodiment of the present invention are non-uniformly distributed, requiring that facial recognition cameras must be installed in important scenic spots (popular spots), and the density of cameras installed in the initial section of the walking trail is higher than that in other sections, and the other sections are randomly distributed.

[0110] The face recognition camera is connected to the fiber optic switch, router, and server via optical fiber.

[0111] The facial feature learning and recognition program included in the face recognition camera uses the YOLOv8-MP model, a lightweight deep neural network MobileNet built based on deep separable convolution, and uses K-means clustering to reset anchors to speed up model convergence, thereby improving the recognition speed of facial features.

[0112] The output size O of the YOLOv8-MP model used by the face recognition camera can be expressed as:

[0113]

[0114] Among them, I is the size of the input image, K is the size of the convolution kernel, P is the number of padding required for the convolution, and S is the step size.

[0115] The lightweight deep neural network MobileNet used by the face recognition camera provided in the embodiment of the present invention decomposes the standard convolution operation into a depth convolution and a 1×1 point convolution.

[0116] In order to avoid the problem of difficulty in recognizing small-size facial features caused by camera angle, depth of field, etc., the embodiment of the present invention obtains the optimal number of anchors and facial feature dimensions through the K-means clustering algorithm, where the clustering distance reflects the error between the feature value and the true value (the initial area uses the feature value) through the overlap intersection over union (IoU), and the error formula is D(fp, center) = 1-IoU(fp, centroid).

[0117] Among them, center is the center of all clusters, fp is the clustering result of sample features, and IoU() represents the intersection-over-union ratio of all sample features and center.

[0118] The application server provided by the embodiment of the present invention is loaded with a human flow prediction program, and its main data source is the face recognition cameras installed at the entrance of the scenic area, the walking trail and the exit. The human flow prediction program is a personnel density prediction model based on deep learning. The hardware system framework required by the model is a three-layer structure. The first layer is the data acquisition layer, which is mainly composed of face recognition cameras installed at the entrance of the scenic area, the walking trail and the exit; the second layer is the network transmission layer, which is mainly composed of optical fibers, optical fiber switches, and routers; the third layer is the application layer, which is mainly composed of storage servers and application servers.

[0119] The storage capacity of the storage server is calculated based on the number of cameras, video parameters, and storage time requirements. In order to achieve better data analysis results, the storage time must be no less than 180 days.

[0120] The application server stores a program for predicting passenger flow, wherein the passenger flow prediction program adopts a tourist flow prediction model based on deep learning (DL-TFP, Deep Learning-basedTouristFlowPrediction).

[0121] The network structure diagram of the tourist flow prediction model based on deep learning is as follows Figure 3 shown.

[0122] The tourist flow prediction system provided by the embodiment of the present invention numbers all face recognition cameras in the scenic area according to the installation position relationship and IP, with a total of k=(3g I +g O +n), the image sequence of each camera is The image sequence is the input of the system.

[0123] The tourist flow prediction model (DL-TFP) provided in the embodiment of the present invention uses convolution operation to capture the spatial correlation between different image acquisition points, that is, the lens area of ​​the face recognition camera, and adopts the ConvLSTM model to calculate its temporal correlation; wherein, the tourist flow prediction system regards the entire scenic area as an independent space, taking into account the interrelationship of the spaces of all camera acquisition areas.

[0124]

[0125] in, is the output of the t-th convolution of the area contained in the i-th face recognition camera image, is the input of the t-th convolution layer, express The weight matrix of Represents the bias term, and ReLU() represents the rectified linear unit as the activation function.

[0126] The convolutional short-term memory network ConvLSTM model used in the tourist flow prediction system provided in the embodiment of the present invention is a recurrent neural network that can process high-dimensional spatial sequence data and model the time dependency in the sequence.

[0127] The tourist flow prediction system provided by the embodiment of the present invention adopts a regional enhanced attention mechanism to describe the spatial correlation between frames of different time steps.

[0128]

[0129] Among them, score() represents the attention score function, W s and W r represents the learning parameters, represents the rate of change of the number of people based on the output of the single-head attention mechanism, q represents the query vector, p represents the key vector, and v represents the value vector. represents the number of people output based on the multi-head attention mechanism, W m represents the learning parameters.

[0130] The tourist flow prediction system provided by the embodiment of the present invention performs weighted calculation on the attention scores of the learned features to predict the number of tourists at the i-th (i∈(1,m)) scenic spot (the total number of scenic spots to be predicted is m) after Δt time from the current moment The value range of the warning time Δt is (0, 24)h.

[0131]

[0132] The application server provided by the embodiment of the present invention and equipped with a tourist flow prediction system is connected to the scenic area control center. The display platform of the control center displays the current number of people at each scenic spot in the scenic area and the number of people at each scenic spot at the warning time Δt (the default value is 0.5h) through heat maps, curve graphs, bar graphs, etc. For a certain scenic spot, you can view the curve of the change of the number of tourists at the scenic spot over time from the opening time in the morning (which can be set by the user) to the current time, as well as the estimated number of tourists at each time point from the current time to the closing time of the same evening (which can be set by the user) and three curves of the optimal number of tourists and the maximum number of tourists at the scenic spot.

[0133] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0134] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for predicting the flow of people in a scenic spot, characterized in that: The method of predicting the flow of people in scenic spots includes: using facial and behavior recognition cameras to calibrate the facial information of tourists entering the scenic spot and identify their identities; analyzing tourists' sightseeing habits based on deep learning, and estimating the tourists' travel locations at the next moment; predicting the dynamic distribution of the flow of people in the scenic spot, predicting the crowded periods of special attractions and popular spots in the scenic spot, and determining whether to activate the tourist diversion plan.

2. The method for predicting the flow of people in a scenic spot according to claim 1, characterized in that: The method for predicting the flow of people in a scenic spot includes the following steps: Step 1: Deploy face and behavior recognition cameras according to the scenic area’s tour routes, identify faces and perform identity calibration through video and image acquisition, and identify tourist behavior characteristics through dynamic analysis; Step 2: The image data of the camera is transmitted to the server in real time via optical fiber. The server analyzes the tourist information collected by multiple cameras and calibrates the tourists into groups. Step 3: The server analyzes the data uploaded by all cameras in the scenic area to determine the specific area of ​​the scenic area where all groups entering the scenic area are currently located; Step 4: The server determines the interval location of the tourist group at the current moment and the behavioral characteristics of the tourists in each group, determines the number of tourists at each high-density scenic spot at the current and next moments, and determines whether to start the tourist diversion plan; Step 5: The server updates the estimated number of tourists at each scenic spot in real time based on the actual location data of the tourists, adjusts the tourist diversion plan in real time, and determines whether to provide voice prompts.

3. The method for predicting the flow of people in a scenic spot as claimed in claim 2, characterized in that: In step 4, the server determines the interval position of the tourist group at the current moment and the behavioral characteristics of the tourists in each group, and determines the estimated number of tourists for each scenic spot and Internet celebrity spot in the next time period according to the time series; if the estimated number of tourists for a certain scenic spot or Internet celebrity spot exceeds the threshold for the scenic spot to accommodate tourists, the server sends an early warning message to the scenic spot operation management department to apply for the activation of the tourist diversion plan; The length of the next time period can be set arbitrarily, with a typical value of 0.5h.

4. The method for predicting the flow of people in a scenic spot as claimed in claim 2, characterized in that: In step 4, the tourist guidance plan is a guidance program pre-set in the scenic area control center, which reminds tourists to adjust their tour progress through voice IP broadcasts on the tour route, including: (1) Providing voice IP broadcast reminders to groups of people gathering at popular spots or groups with slow tour speeds; (2) Voice IP broadcast reminders are provided for scenic spots that are of low interest to tourists, are missed, have a large passenger capacity, or are visited by groups with a fast speed.

5. The method for predicting the flow of people in a scenic spot as claimed in claim 2, characterized in that: In step five, the server updates the estimated number of tourists at each scenic spot in real time based on the actual travel data of the tourists, and adjusts the tourist diversion plan in real time; the diversion prompt information of each voice IP broadcast is adjusted in real time according to the behavioral characteristics of the current group of tourists. If the estimated number of tourists at all attractions is lower than the tourist capacity threshold of the attraction, no voice prompt will be given.

6. A scenic spot passenger flow prediction system using the scenic spot passenger flow prediction method according to any one of claims 1 to 5, characterized in that: The scenic spot flow prediction system includes face and behavior recognition cameras, fiber optic networks, fiber optic switches, storage servers, and application servers; Face and behavior recognition cameras are installed at the entrance of the scenic area, walking trails, special places, scenic spots and exits, including fixed-angle gun cameras and ball cameras with pan / tilts. The ball cameras are used for high-altitude shooting and tracking shooting. The fiber optic switch is used to connect all facial and behavior recognition cameras in the scenic area to the server; The servers are divided into application servers and storage servers, which are directly connected to the fiber switches; The storage server includes sufficient disk arrays to meet the scenic spot's storage requirements for video data; The application server has sufficient computing speed and stores programs for numbering the face information collected by each face and behavior recognition camera, and for calculating the estimated number of tourists at each scenic spot at each moment in the next period of time in real time, and comparing it with the tourist accommodation threshold determined in advance by the scenic spot; The application server regularly broadcasts all calibrated identity numbers and facial information of the day to each facial recognition camera through the fiber optic switch, and the update cycle does not exceed 10 seconds.

7. The scenic spot flow prediction system as claimed in claim 6, characterized in that: The face and behavior recognition cameras calibrate the face information and upload it to the server immediately after successful calibration; the server assigns identity numbers to all calibrated face information of the day in real time, and broadcasts the identity numbers and face image information to each face recognition camera, with an update interval of no more than 10s / time; The face and behavior recognition camera stores the face information and identity number broadcast by the server in the local storage. When a new face enters the lens, the tourist's behavior is captured and compared with the face information in the storage. If it already exists, the tourist's behavior characteristics are stored according to the identity number and face information saved in the local storage, and the video and recognition data are transmitted to the server in real time via optical fiber; if there is no tourist face information and identity number in the local storage, the video, face recognition and tourist behavior characteristic information are directly transmitted to the server; the local storage only stores the tourist identity code, the corresponding face image information and tourist behavior characteristic information, and all are cleared at 24:00 every day.

8. A computer device, characterized in that: The computer device includes a memory and a processor. The memory stores a computer program. When the computer program is executed by the processor, the processor executes the steps of the scenic spot passenger flow prediction method as described in any one of claims 1 to 5.

9. A computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to execute the steps of the scenic spot passenger flow prediction method as claimed in any one of claims 1 to 5.

10. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the scenic spot passenger flow prediction system as described in any one of claims 6 to 7.

Citation Information

Patent Citations

  • A pedestrian flow early warning system for tourist scenic spots

    CN109272153A

  • Deep learning-based rail transit passenger flow demand prediction method and device

    CN110991775A

  • Scenic spot playing track prediction system based on multi-feature fusion

    CN111079968A

  • Automatic driving bus dispatching system and bus dispatching method based on passenger flow prediction

    CN117172461A

  • Group person trajectory prediction method based on spatio-temporal characteristics

    CN117953009A

Cited By

  • Smart tourism management system and method

    CN119579359A

  • Online passenger flow monitoring analysis data error correction method

    CN119807601A

  • Campus safety informatization management method based on artificial intelligence

    CN120088728A

  • Scenic spot passenger flow volume prediction method based on multi-source information fusion

    CN120822673A

  • Public area intelligent management method and device, equipment and storage medium

    CN120997757A