A heat map-based scenic spot tourist navigation method

Abstract

The application discloses a scenic spot tourist navigation method based on a heat map, relates to a scenic spot tourist navigation method, and provides multiple navigation routes for selection for tourists, and generates a heat map by predicting real-time positions of the tourists by using historical data; the predicted heat map is then corrected by a pre-trained convolutional neural network to ensure conformity with actual conditions; when the heat map shows that the crowd density of a certain area is about to exceed a threshold value, the system recommends a navigation route with high uniformity to new tourists to ensure balanced distribution of people flow in the scenic spot. The application solves the problem of possible local congestion of tourists in the scenic spot.

Description

Technical Field

[0001] This invention relates to a method for navigating tourists in scenic areas, and more specifically, to a method for navigating tourists within scenic areas based on heat maps. Background Technology

[0002] With the rapid development of the tourism industry, more and more people are choosing to visit various scenic spots, parks, and tourist destinations during holidays or weekends. These places offer tourists a wealth of cultural, entertainment, and natural experiences, but they also face a significant management challenge: how to ensure that large numbers of tourists can visit smoothly and safely, especially during peak seasons or at certain popular attractions.

[0003] Since scenic areas are generally large, the lack of planned navigation routes can easily lead to overcrowding in certain areas. This not only diminishes the visitor experience but can also pose safety hazards, such as stampedes. Therefore, how to accurately predict and manage such localized overcrowding in real time has become a pressing issue for scenic area managers. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a tourist navigation method based on heatmaps within scenic areas, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A heatmap-based tourist navigation method within a scenic area includes the following steps:

[0007] S1: Offers multiple navigation routes for tourists to choose from and receives navigation route selections based on tourists' personal preferences;

[0008] S2: Based on each tourist's departure time and selected navigation route, and using historical data to calculate the average walking speed and average stay time at each attraction, calculate each tourist's real-time location within the scenic area, and generate a predicted heat map accordingly.

[0009] S3: Correct the predicted heatmap using a pre-trained convolutional neural network; wherein the training process of the convolutional neural network is as follows: collect actual heatmap data, and pair the predicted heatmap with the actual heatmap data to form a training dataset; use the training dataset to train the model through the convolutional neural network, with the goal of minimizing the difference between the predicted heatmap and the actual heatmap.

[0010] S4: When the population density in any preset area in the predicted and corrected heat map exceeds the preset threshold, the following navigation route recommendation method is adopted for newly entered tourists: For each navigation route, calculate the corrected heat map that will be generated in the future if it is assigned to newly entered tourists. For each corrected heat map, calculate its uniformity over a period of time in the future. The navigation route corresponding to the heat map with higher uniformity is preferentially recommended to tourists.

[0011] Preferably, the navigation route is a Hamiltonian path that starts at the scenic area entrance, ends at the exit, and passes through all attractions.

[0012] Preferably, in the heat map, the darker the color, the higher the population density.

[0013] Preferably, the uniformity of the heatmap at any given time is calculated as follows:

[0014] Extract the population density values ​​of all points on the heatmap to form a dataset, where each point represents a unit area; calculate the standard deviation of this dataset; the smaller the standard deviation, the higher the uniformity.

[0015] In some embodiments, the reciprocal of the standard deviation is taken as a measure of uniformity.

[0016] Preferably, the uniformity over a period of time is represented by the integral of the standard deviation over time; the smaller the integral value, the higher the uniformity.

[0017] In some embodiments, the uniformity over a period of time is equal to the reciprocal of the integral value.

[0018] In some embodiments, the time period is from the moment when the uniformity is calculated to the moment when the scenic area closes.

[0019] In some embodiments, the preset area mentioned in S4 is 50 square meters.

[0020] In some embodiments, the preset threshold is 0.2 people / square meter.

[0021] The advantages of this invention over existing technologies are that it combines predictive heatmaps, convolutional neural networks, and dynamic route recommendation technologies, enabling real-time and accurate prediction and management of pedestrian flow distribution within scenic areas, especially for potential localized congestion. Furthermore, it can provide navigation route recommendations that make pedestrian flow distribution more even based on existing pedestrian flow patterns. This not only significantly improves the visitor experience and safety but also provides scenic area managers with an efficient and scientific tool for managing pedestrian flow. Attached Figure Description

[0022] Figure 1It is a schematic diagram of the method of the present invention. Detailed implementation manners

[0023] The following describes the detailed implementation manners of the present invention with reference to the accompanying drawings.

[0024] As Figure 1 shown, the method of the present invention includes the following steps: The deeper the color, the denser the crowd.

[0025] S1: Provide multiple navigation routes for tourists to choose from, and receive the navigation route selected by the tourists based on their personal preferences;

[0026] S2: Based on the departure time of each tourist and the selected navigation route, use the average walking speed of tourists and the average stay time at each scenic spot statistically obtained from historical data to calculate the real-time position of each tourist in the scenic area, and generate a predicted heat map accordingly;

[0027] S3: Use a pre-trained convolutional neural network to correct the predicted heat map; wherein, the training process of the convolutional neural network is: collect actual heat map data, and pair the predicted heat map with the actual heat map data to form a training data set; use the training data set to perform model training through the convolutional neural network, and its goal is to minimize the difference between the predicted heat map and the actual heat map;

[0028] S4: When the crowd density within any preset area in the future in the predicted and corrected heat map exceeds the preset threshold, for newly entered tourists in the scenic area, adopt the following navigation route recommendation method: For each navigation route, calculate the corrected heat map that will be caused in the future if it is assigned to the newly entered tourists in the scenic area, and for each corrected heat map, calculate its uniformity within a future period of time. Among them, the navigation route corresponding to the heat map with a higher uniformity is preferably recommended to the tourists.

[0029] In the above method, for the navigation route, it starts from the entrance and ends at the exit, and is a Hamiltonian path that can pass through all scenic spots. Generally speaking, the navigation route is several navigation routes recommended by the scenic area staff based on their understanding of the terrain of the scenic area, etc., and can be formulated according to the specific situation of the scenic area.

[0030] In the above method, the so-called average walking speed and average stay time at each attraction in S2 refer to the statistical analysis of a large number of tourists' walking speeds and stay times at each attraction in historical tourist data. This data is then averaged to obtain the average walking speed and average stay time at each attraction. After averaging, the average data is used to determine the location of each tourist in subsequent large numbers of visitors. While this may not be accurate for individual tourists, the overall change in the heat map caused by a large number of tourists will not differ significantly. This heat map displays the crowd density in various areas of the scenic area at a specific point in time; the darkness of the color indicates the size of the crowd, with darker colors indicating a denser crowd.

[0031] Additionally, generally, we have a given order of routes to attractions and roads connecting them. When a tourist arrives at an attraction, since an attraction often encompasses a large area, if we can calculate the number of tourists at that attraction at a given moment, we can divide the number of tourists by the area of ​​the attraction to obtain the density at that attraction. Furthermore, since the location of tourists on roads is predictable, and roads are generally relatively narrow, we can calculate the density on the road based on the point a tourist reaches on that road.

[0032] Heatmaps can be created using the Python-based open-source library Matplotlib. After inputting the calculated crowd density, a heatmap can be generated. These heatmaps can then be displayed in real-time on screens at the scenic area and its management office for reference.

[0033] In addition, to further reduce the discrepancy between the calculated heatmap and the actual heatmap, this invention also designs a convolutional neural network to correct the heatmap:

[0034] First, actual heatmaps need to be collected from historical data, ensuring they have consistent resolution and size. These actual heatmaps are generated by counting the number of visitors in the scenic area using infrared cameras; that is, they are the actual observed heatmaps of the area. The predicted heatmaps, on the other hand, are calculated based on the average walking speed of visitors and the average time spent at each attraction. The predicted heatmaps are used as input, and the actual heatmaps as the target output, forming a training data pair. For the network architecture, a basic convolutional neural network architecture can be chosen. Typically, one can start with a few simple convolutional layers and then increase the number of layers for better performance. Each convolutional layer can be followed by a batch normalization layer and a ReLU activation function to increase the model's non-linearity. Finally, the network should have one or more fully connected layers, and the output size should be the same as the actual heatmap.

[0035] Since the task is regressive, mean squared error (MSE) can be used as the loss function to measure the difference between the predicted and actual heatmaps. Adaptive optimizers such as Adam or RMSprop can also be used for training.

[0036] During training, the dataset is divided into training and validation sets. The model is trained using the training set, and performance is periodically checked on the validation set to prevent overfitting. Early stopping techniques are considered, i.e., training is stopped when the validation loss no longer decreases significantly. For the generated revised heatmap, thresholding techniques can be used for further processing, such as filtering out areas with extremely low population density.

[0037] In step S4, a key question is how to assess the uniformity of the heat map over a future period. This can be addressed using the standard deviation calculation method.

[0038] In any given heatmap, each pixel or region represents the population density at that specific location. The uniformity of a heatmap actually refers to how evenly the population is distributed throughout the scenic area. If the population is relatively evenly distributed across different areas, the visitor experience and mobility will be improved, while congestion and potential safety hazards will be reduced.

[0039] Standard deviation is a commonly used statistical tool to measure the dispersion of a set of data. In this invention, standard deviation is used to measure the dispersion of population density at various points on a heatmap. The specific steps are as follows:

[0040] First, extract the crowd density values ​​for all points on the heatmap (each point represents a specific area of ​​the scenic spot, such as a 50-square-meter area). These values ​​will form a dataset representing the expected number of people in each area of ​​the scenic spot.

[0041] Next, calculate the average value of this dataset. This value tells us the average population density within the scenic area. Theoretically, if the population density in each area is close to this average, then the population distribution within the scenic area is relatively uniform.

[0042] To understand how much the population density in each region deviates from the mean, we calculate the difference between the mean and the population density of each data point (i.e., each region). Then, we square each difference and sum them. Finally, we divide this sum by the number of data points to obtain the variance. The square root of the variance is the standard deviation.

[0043] A lower standard deviation means that the population density in most areas is close to the average, indicating a relatively even distribution. Conversely, a higher standard deviation indicates that the population density in some areas is much higher or lower than the average, suggesting the existence of crowded or sparsely populated areas within the scenic area.

[0044] This method allows us to obtain the uniformity of each heatmap. By integrating the uniformity calculated at each moment over time, we can determine the overall uniformity over a given period. This period typically refers to the time from the moment of calculation to the venue's closing time.

[0045] This uniformity can be used to recommend navigation routes for new visitors to the scenic area, thereby striving to distribute visitor flow as evenly as possible within the area. In other words, navigation routes that contribute to a more uniform heatmap are prioritized. Navigation routes can be sorted according to their uniformity and then recommended to visitors.

[0046] Furthermore, generally speaking, the threshold mentioned in S4 regarding "crowd density exceeding the preset threshold" is not a threshold indicating actual overcrowding, but rather a warning value for potential overcrowding, such as 0.2 people per square meter. By setting such a warning threshold, tourists can be given ample choice when there are few visitors, and navigation routes can be planned in advance to prevent potential congestion.

[0047] This invention uses a fictional scenic area as an example to illustrate its technical solution:

[0048] Imagine a large theme park called "The Wizard of Oz". This park has four main areas: the Enchanted Forest, the Fantasy Castle, the Magic Academy, and the Fairy Tale Village.

[0049] Step S1: The park provides three navigation routes for visitors:

[0050] Route A: First visit the Magic Forest, then proceed to the Fantasy Castle, followed by the Magic Academy, and finally the Fairy Tale Village.

[0051] Route B: First visit the fairytale village, then the fantasy castle, then the magic forest, and finally end at the magic academy.

[0052] Route C: First, go to the Magic Academy, then the Magic Forest, then the Fantasy Castle, and finally visit the Fairy Tale Village.

[0053] After entering the park, Alice, a tourist, chose route A.

[0054] Step S2: Based on historical data, the park predicts that Alice will stay in the Enchanted Forest for 45 minutes, the Fantasy Castle for 1 hour, the Magic Academy for 1.5 hours, and the Fairy Tale Village for 30 minutes. Based on this data and Alice's departure time, the system predicts her route and travel time. Similar predictions are made for all other visitors in the park, generating a predicted heatmap.

[0055] Step S3: Previously, the park had collected actual crowd heat maps and used this data to train a convolutional neural network. The park used this model to refine the predicted heat maps.

[0056] Step S4: The system notices that the revised heat map shows a very high crowd density at the Enchanted Castle within the next hour. To avoid this overcrowding, when new visitor Bob enters the park, the system recommends Route C, as the heat map predicting the order of attractions along this route shows the most even distribution of people, especially at the popular Enchanted Castle.

[0057] In this way, the park not only ensures that every visitor can visit all the attractions, but also effectively manages the flow of people, making the distribution of visitors more even and improving the overall visitor experience.

[0058] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A tourist navigation method within a scenic area based on heatmaps, characterized in that, Includes the following steps: S1: Offers multiple navigation routes for tourists to choose from and receives navigation route selections based on tourists' personal preferences; S2: Based on each tourist's departure time and selected navigation route, and using historical data to calculate the average walking speed and average stay time at each attraction, calculate each tourist's real-time location within the scenic area, and generate a predicted heat map accordingly. S3: Correct the predicted heatmap using a pre-trained convolutional neural network; wherein the training process of the convolutional neural network is as follows: collect actual heatmap data, and pair the predicted heatmap with the actual heatmap data to form a training dataset; use the training dataset to train the model through the convolutional neural network, with the goal of minimizing the difference between the predicted heatmap and the actual heatmap. S4: When the population density in any preset area in the predicted and corrected heat map exceeds the preset threshold, the following navigation route recommendation method is adopted for newly entered tourists: For each navigation route, calculate the corrected heat map that will be generated in the future if it is assigned to newly entered tourists. For each corrected heat map, calculate its uniformity over a period of time in the future. The navigation route corresponding to the heat map with higher uniformity is preferentially recommended to tourists.

2. The tourist navigation method within a scenic area based on a heat map according to claim 1, characterized in that, The navigation route is a Hamiltonian path that starts at the entrance of the scenic area and ends at the exit, passing through all the attractions.

3. The tourist navigation method within a scenic area based on a heat map according to claim 1, characterized in that, In the heat map, the darker the color, the higher the population density.

4. The tourist navigation method within a scenic area based on a heat map according to claim 1, characterized in that, The uniformity of the heatmap at any given time is calculated as follows: Extract the population density values ​​of all points on the heatmap to form a dataset, where each point represents a unit area; calculate the standard deviation of this dataset; the smaller the standard deviation, the higher the uniformity.

5. The tourist navigation method within a scenic area based on a heat map according to claim 4, characterized in that, The reciprocal of the standard deviation is used as a measure of uniformity.

6. The tourist navigation method within a scenic area based on a heat map according to claim 4, characterized in that, The uniformity over a period of time is represented by the integral of the standard deviation over time; the smaller the integral value, the higher the uniformity.

7. The tourist navigation method within a scenic area based on a heat map according to claim 6, characterized in that, The uniformity over a given period of time is equal to the reciprocal of the integral value.

8. The tourist navigation method based on heatmaps within a scenic area according to claim 6 or 7, characterized in that, The time period is from the moment the uniformity is calculated to the moment the scenic area closes.

9. The tourist navigation method within a scenic area based on a heat map according to claim 1, characterized in that, The preset area mentioned in S4 is 50 square meters.

10. The tourist navigation method within a scenic area based on a heat map according to claim 1, characterized in that, The preset threshold is 0.2 people per square meter.

Images