Darjeeling Flickr Dataset
Description
• H1: Hierarchical Density-Based Clustering of geotagged images based on their geospatial information can aid in the discovery of regions of interest. • H2: A combination of unsupervised and supervised classification over traditional image classification algorithms can effectively identify specific tourist interests and imageability within Regions of Interest. • H3: Topic modelling of image captions can support and validate the results of image classification while providing further insight into tourist attraction types.
Files
Steps to reproduce
A QGIS plugin of flickrAPI was used to download the images and associated metadata (https://github.com/arka816/flickrforqgis). HDBSCAN algorithm in ArcGIS was used for clustering. Unsupervised and supervised classification done using Orange3.