METHOD OF IMPROVING CLASSIFICATION ACCURACY OF SNS IMAGE DATA FOR TOURISM USING SPACE INFORMATION DEEP LEARNING, RECORDING MEDIUM AND DEVICE FOR PERFORMING THE METHOD

摘要

The method for improving the accuracy of SNS image data classification for tourism purposes using spatial information deep learning technology, acquires image data posted on SNS by geotagging, and constructs a composite product neural network (CNN) based on tourist activity information (What). Classifying image data according to an image-only classification system for tourism purposes; Extracting a cluster for each category detail item through density-based spatial clustering of applications with noise (DBSCAN) using the category information labeled on the classified image data; Comparing the location information extracted from the cluster and the labeled value of the activity information (What), and if they do not match, updating the activity information (What) based on the location information; Re-classifying image data according to an image-only classification system for tourism purposes through a convolutional neural network (CNN) based on the updated activity information (What); And measuring the accuracy of image data classification through superimposition analysis of the updated activity information (What) and the actual location. Accordingly, it is possible to improve the classification accuracy of the image for tourism purposes by checking the location information through DBSCAN of the photo data labeled with the category and updating the location information if there is an error.