Method for subdividing tourist market by using computer vision model of tourist online photos

Abstract

The invention discloses a method for subdividing tourist markets by using a computer vision model of tourist online photos, which belongs to the technical field of Internet information processing, is used for subdividing tourist markets, and comprises the following steps: collecting online photo data uploaded by tourists from a social media platform and cleaning and processing the data; a visual content analysis data set is obtained after cutting, duplicate removal and elimination processing; performing visual content analysis on the tourist photos by using a computer vision model URNet model; subdividing tourist markets according to photo theme attributes; obtaining a tourist market distribution rule from time and space dimensions; and the correlation of different tourist markets in time and space is analyzed. Compared with the prior art, the method has the advantages that the preferences and demands of tourists are deeply obtained, the gap between marketing products and the demands of the tourists is reduced, and the tourism sustainable development of destinations is promoted; the system is simple in structure, low in use cost and wide in application range, can carry out more detailed group segmentation on tourists, and realizes personalized service and differentiated marketing.

Description

Technical Field

[0001] The invention discloses a method for segmenting a tourist market by using a computer vision model of online photos of tourists, and belongs to the technical field of Internet information processing. Background Art

[0002] With the surge in competition in the tourism market, it is found that using online photo data as material and analyzing it through computer vision models is an effective means of segmenting the tourist market. As the tourism industry rapidly emerges as the most promising industry in the world, fierce market competition has promoted the intensified monitoring of market trends. In order to gain an advantage in this highly competitive environment, segmenting the tourist market has become an indispensable strategic means. This strategy requires dividing tourists into groups based on records that reflect their preferences or needs, and developing personalized service and marketing plans for each group.

[0003] Although photos are crucial in assessing market competitiveness and sustainability, market segmentation based on tourist photos is still lacking in-depth research. Existing studies mainly rely on questionnaires and online texts, and pay insufficient attention to the analysis of the visual content of photos. In recent years, with the development of computer technology and deep learning models, large-scale photo content analysis has become possible, providing strong support for tourist segmentation based on photo materials. Computer vision models can help destination managers understand tourists' interests and preferences more accurately by identifying elements such as attractions, activities, and people, providing an effective tool for segmenting the tourist market, making personalized travel experiences and customized marketing strategies possible. Summary of the invention

[0004] The purpose of the present invention is to provide a method for segmenting the tourist market by using a computer vision model of tourists' online photos, so as to solve the problems in the prior art of lack of personalized services in the tourist market, large gap between marketing products and tourists' needs, and high cost of use.

[0005] A method for segmenting tourist markets using computer vision models of tourists' online photos, comprising:

[0006] S1. Collect online photo data uploaded by tourists from social media platforms and perform data cleaning and processing;

[0007] S2. Use the computer vision model URNet model to perform visual content analysis on tourist photos;

[0008] S3. Segment the tourist market based on photo theme attributes;

[0009] S4. Obtain the distribution law of tourist market from the dimensions of time and space;

[0010] S5. Analyze the correlation between different tourist markets in time and space.

[0011] S1 includes the acquisition of online photo data by using an application programming interface (API) to collect photo data and metadata uploaded by visitors of the photo sharing website Flickr. The metadata includes the photo ID, user ID Owner, shooting time Datetaken, longitude Lon, latitude Lat, download link Url_z and title Title.

[0012] S1 includes data processing, which is to match the photo data with the attribute data, and obtain the visual content analysis data set after regional cropping, deduplication, and elimination of local resident data.

[0013] S2 includes a computer vision model URNet model, which inputs photo data into the URNet model to extract photo scenes, divides the processed visual content analysis data set into a training data set and a test data set, trains the URNet model with the training data set, verifies the accuracy of the URNet model with the test data set, evaluates the performance of the model, and adjusts the model parameters to improve the recognition accuracy.

[0014] S2 includes using the trained URNet model to perform image recognition and classification on photo data from the photo sharing website Flickr, outputting the classification results as keywords, obtaining the confidence scores of the keywords, using the keywords with the highest confidence as the scene labels of the photos, and dividing the recognized scenes into multiple categories based on the classification method of tourism-related scenes.

[0015] S3 includes segmenting the tourist market by using the K-Means clustering algorithm to divide the photos into themes. The K-Means clustering algorithm is used, and the Euclidean distance is used as a measure of the similarity of photo attributes. The longitude and latitude coordinates of each tourist's photos and the proportion of 10 types of identified scene photos are input for clustering.

[0016] S3 includes K-Means clustering, which is a clustering algorithm from n data points {x1,x2,x3,...,x n}, randomly select k samples as the initial cluster centers {c1,c2,c3,…,c k}, divide the samples into the nearest cluster center, calculate the cluster center by iterative relocation, and complete clustering when the cluster center no longer changes or the change is less than the set threshold;

[0017] The distance d from the data point to the cluster center is:

[0018]

[0019] In the formula, x i is the i-th data point, c j is the jth cluster center;

[0020] The silhouette coefficient and elbow method were used to determine the number of clusters of tourist photo themes, and the clustering results were verified by analysis of variance.

[0021] S4 includes obtaining the statistical results in time and space. The statistical results in time and space include obtaining the distribution law of the tourist market from the time and space dimensions, which is to spread the coordinates of the tourist photo theme clusters according to the kilometer grid, use the indicators of frequency density and type ratio to calculate the proportion of the number of photos taken by different types of tourists in the grid, and determine the spatial distribution law of the tourist photo theme clusters, corresponding to the types of tourists popular in the area, including local food lovers, lifestyle lovers, animal and plant lovers, portrait infrastructure lovers, natural scenery lovers and cultural stage lovers;

[0022] Indicator frequency density F i1 Calculated as:

[0023]

[0024] In the formula, i1 represents the type of tourist, n i1 Indicates the number of photos in the grid, N i1 Indicates the total number of photos.

[0025] S4 includes type scale C i1 The calculation is:

[0026]

[0027] Statistics were collected on the time distribution of photography activities of various types of tourists, the proportion of photos taken by different types of tourists at different time periods, and the time behavior patterns of various types of tourists as well as their different preferences and characteristics in terms of travel time.

[0028] S5 includes the Spearman rank correlation coefficient Spearman method to analyze the correlation of different tourist markets in time and space. The statistical results in time and space in S4 are combined with the Spearman method correlation analysis to calculate the correlation of photographic behaviors of various types of tourists. The Spearman correlation coefficient ρ is calculated as:

[0029]

[0030] Where, d i Represents the difference in rank values ​​of the i-th data point pair.

[0031] Compared with the prior art, the present invention has the following beneficial effects: deeply obtaining the preferences and needs of tourists, reducing the gap between marketing products and tourist needs, and promoting the sustainable development of tourism in the destination; low cost of use, wide scope of application, and the ability to perform more detailed group segmentation of tourists to achieve personalized services and differentiated marketing. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is a flow chart of the present invention;

[0033] Figure 2 This is the URNet model structure diagram;

[0034] Figure 3 It is the time distribution map of tourists’ photography activities;

[0035] Figure 4 It is a correlation analysis diagram of tourists’ photography activities in the spatial dimension;

[0036] Figure 5 It is a correlation analysis chart of tourists' photography activities in the time dimension. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] A method for segmenting the tourist market using a computer vision model of tourists' online photos. The flow chart of the present invention is as follows: Figure 1 As shown, the steps include:

[0039] S1. Collect online photo data uploaded by tourists from social media platforms and perform data cleaning and processing;

[0040] S2. Use the computer vision model URNet model to perform visual content analysis on tourist photos;

[0041] S3. Segment the tourist market based on photo theme attributes;

[0042] S4. Obtain the distribution law of tourist market from the dimensions of time and space;

[0043] S5. Analyze the correlation between different tourist markets in time and space.

[0044] S1 includes the acquisition of online photo data by using an application programming interface (API) to collect photo data and metadata uploaded by visitors of the photo sharing website Flickr. The metadata includes the photo ID, user ID Owner, shooting time Datetaken, longitude Lon, latitude Lat, download link Url_z and title Title.

[0045] The URNet model structure is as follows Figure 2As shown, S1 includes data processing, which is to match the photo data with the attribute data, and obtain the visual content analysis data set after regional cropping, deduplication, and elimination of local resident data.

[0046] S2 includes a computer vision model URNet model, which inputs photo data into the URNet model to extract photo scenes, divides the processed visual content analysis data set into a training data set and a test data set, trains the URNet model with the training data set, verifies the accuracy of the URNet model with the test data set, evaluates the performance of the model, and adjusts the model parameters to improve the recognition accuracy.

[0047] S2 includes using the trained URNet model to perform image recognition and classification on photo data from the photo sharing website Flickr, outputting the classification results as keywords, obtaining the confidence scores of the keywords, using the keywords with the highest confidence as the scene labels of the photos, and dividing the recognized scenes into multiple categories based on the classification method of tourism-related scenes.

[0048] S3 includes segmenting the tourist market by using the K-Means clustering algorithm to divide the photos into themes. The K-Means clustering algorithm is used, and the Euclidean distance is used as a measure of the similarity of photo attributes. The longitude and latitude coordinates of each tourist's photos and the proportion of 10 types of identified scene photos are input for clustering.

[0049] S3 includes K-Means clustering, which is a clustering algorithm from n data points {x1,x2,x3,…,x n}, randomly select k samples as the initial cluster centers {c1,c2,c3,...,c k}, divide the samples into the nearest cluster center, calculate the cluster center by iterative relocation, and complete clustering when the cluster center no longer changes or the change is less than the set threshold;

[0050] The distance d from the data point to the cluster center is:

[0051]

[0052] In the formula, x i is the i-th data point, c j is the jth cluster center;

[0053] The silhouette coefficient and elbow method were used to determine the number of clusters of tourist photo themes, and the clustering results were verified by analysis of variance.

[0054] S4 includes obtaining the statistical results in time and space. The statistical results in time and space include obtaining the distribution law of the tourist market from the time and space dimensions, which is to spread the coordinates of the tourist photo theme clusters according to the kilometer grid, use the indicators of frequency density and type ratio to calculate the proportion of the number of photos taken by different types of tourists in the grid, and determine the spatial distribution law of the tourist photo theme clusters, corresponding to the types of tourists popular in the area, including local food lovers, lifestyle lovers, animal and plant lovers, portrait infrastructure lovers, natural scenery lovers and cultural stage lovers;

[0055] Frequency density of indicator F i1 Calculated as:

[0056]

[0057] In the formula, i1 represents the type of tourist, n i1 Indicates the number of photos in the grid, N i1 Indicates the total number of photos.

[0058] S4 includes type scale C i1 The calculation is:

[0059]

[0060] Statistics on the time distribution of various types of tourists' photography activities. The time distribution of tourists' photography activities is shown in the figure below. Figure 3 As shown in the figure, Local cuisine enthusiasts are local cuisine enthusiasts, Lifestyle enthusiasts are lifestyle enthusiasts, Flora and fauna enthusiasts are flora and fauna enthusiasts, Portraits infrastructureenthusiasts are portrait infrastructure enthusiasts, Natural scenery enthusiasts are natural scenery enthusiasts, Cultural herttage enthusiasts are cultural stage enthusiasts, and the proportion of photos taken by different types of tourists at different times can be obtained to obtain the time behavior patterns of various types of tourists and their different preferences and characteristics in travel time.

[0061] The relevant analysis of tourists' photography activities in the spatial dimension is shown in the figure below: Figure 4 As shown in the figure, * indicates 0.01<ρ≤0.05, and ** indicates ρ≤0.01. The relevant analysis diagram of tourists' photography activities in the time dimension is shown in Figure 5 As shown in the figure, * indicates 0.01<ρ≤0.05, and ** indicates ρ≤0.01.

[0062] S5 includes the Spearman rank correlation coefficient Spearman method to analyze the correlation of different tourist markets in time and space. The statistical results in time and space in S4 are combined with the Spearman method correlation analysis to calculate the correlation of photographic behaviors of various types of tourists. The Spearman correlation coefficient ρ is calculated as:

[0063]

[0064] Where, d i Represents the difference in rank values ​​of the i-th data point pair.

[0065] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features may be replaced by equivalents, and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for segmenting the tourist market using a computer vision model of tourists' online photos, characterized in that: include: S1. Collect online photo data uploaded by tourists from social media platforms and perform data cleaning and processing; S2. Use the computer vision model URNet model to perform visual content analysis on tourist photos; S3. Segment the tourist market based on photo theme attributes; S4. Obtain the distribution law of tourist market from the dimensions of time and space; S5. Analyze the correlation between different tourist markets in time and space.

2. A method for segmenting the tourist market using a computer vision model of tourists' online photos according to claim 1, characterized in that: S1 includes the acquisition of online photo data by using an application programming interface (API) to collect photo data and metadata uploaded by visitors of the photo sharing website Flickr. The metadata includes the photo ID, user ID Owner, shooting time Datetaken, longitude Lon, latitude Lat, download link Url_z and title Title.

3. The method for segmenting the tourist market using a computer vision model of tourists' online photos according to claim 2, characterized in that: S1 includes data processing, which is to match the photo data with the attribute data, and obtain the visual content analysis data set after regional cropping, deduplication, and elimination of local resident data.

4. The method for segmenting the tourist market using a computer vision model of tourists' online photos according to claim 3, characterized in that: S2 includes a computer vision model URNet model, which inputs photo data into the URNet model to extract photo scenes, divides the processed visual content analysis data set into a training data set and a test data set, trains the URNet model with the training data set, verifies the accuracy of the URNet model with the test data set, evaluates the performance of the model, and adjusts the model parameters to improve the recognition accuracy.

5. A method for segmenting the tourist market using a computer vision model of tourists' online photos according to claim 4, characterized in that: S2 includes using the trained URNet model to perform image recognition and classification on photo data from the photo sharing website Flickr, outputting the classification results as keywords, obtaining the confidence scores of the keywords, using the keywords with the highest confidence as the scene labels of the photos, and dividing the recognized scenes into multiple categories based on the classification method of tourism-related scenes.

6. A method for segmenting the tourist market using a computer vision model of tourists' online photos according to claim 5, characterized in that: S3 includes segmenting the tourist market by using the K-Means clustering algorithm to divide the photos into themes. The K-Means clustering algorithm is used, and the Euclidean distance is used as a measure of the similarity of photo attributes. The longitude and latitude coordinates of each tourist's photos and the proportion of 10 types of identified scene photos are input for clustering.

7. A method for segmenting the tourist market using a computer vision model of tourists' online photos according to claim 6, characterized in that: S3 includes K-Means clustering, which is a clustering algorithm from n data points {x1,x2,x3,...,x n }, randomly select k samples as the initial cluster centers {c1,c2,c3,...,c k }, divide the samples into the nearest cluster center, calculate the cluster center by iterative relocation, and complete clustering when the cluster center no longer changes or the change is less than the set threshold; The distance d from the data point to the cluster center is: In the formula, x i is the i-th data point, c j is the jth cluster center; The silhouette coefficient and elbow method were used to determine the number of clusters of tourist photo themes, and the clustering results were verified by analysis of variance.

8. The method for segmenting the tourist market using a computer vision model of tourists' online photos according to claim 7, characterized in that: S4 includes obtaining the statistical results in time and space. The statistical results in time and space include obtaining the distribution law of the tourist market from the time and space dimensions, which is to spread the coordinates of the tourist photo theme clusters according to the kilometer grid, use the indicators of frequency density and type ratio to calculate the proportion of the number of photos taken by different types of tourists in the grid, and determine the spatial distribution law of the tourist photo theme clusters, corresponding to the types of tourists popular in the area, including local food lovers, lifestyle lovers, animal and plant lovers, portrait infrastructure lovers, natural scenery lovers and cultural stage lovers; Frequency density of indicator F i1 Calculated as: In the formula, i1 represents the type of tourist, n i1 Indicates the number of photos in the grid, N i1 Indicates the total number of photos.

9. The method for segmenting the tourist market using a computer vision model of tourists' online photos according to claim 8, characterized in that: S4 includes type scale C i1 The calculation is: Statistics were collected on the time distribution of photography activities of various types of tourists, the proportion of photos taken by different types of tourists at different time periods, and the time behavior patterns of various types of tourists as well as their different preferences and characteristics in terms of travel time.

10. The method for segmenting the tourist market using a computer vision model of tourists' online photos according to claim 9, characterized in that: S5 includes the Spearman rank correlation coefficient Spearman method to analyze the correlation of different tourist markets in time and space. The statistical results in time and space in S4 are combined with the Spearman method correlation analysis to calculate the correlation of photographic behaviors of various types of tourists. The Spearman correlation coefficient ρ is calculated as: Where, d i Represents the difference in rank values ​​of the i-th data point pair.

Images