data_code_MeasuringGeodiversity1

Datasets, code and results for paper 'Measuring Geodiversity in Large Overhead Imagery Datasets'1) Synthetic shapes dataset .zip (Count:3)Used to test GeoFID and GeoIS computations in the paper.Contains 3 separate zip files for 'polygon', 'star', and 'ellipse' shape classes. Each .zip file contains 2,000 images in .png format, 1,000 of which are randomized images along with 1,000 control images as described in the paper. Image filenames are referenced in external polygon GIS files (.geojson format) to facilitate research experiments in the paper.2) Synthetic shapes dataset creation script .py (Count:1)Used to create the dataset in (1) with the pycairo package3) Sample locations .csv (Count:3)Used to specify XY coordinate locations (WGS1984) for each shape class of the synthetic shapes dataset in (1)4) Polygon sub-regions creation script .py (Count:1)Used to create subregion grids in GeoJSON format with the shapely package5) Polygon sub-region grids .geojson (Count:12)Used to link images in synthetic dataset to geospatial regions based on point locations in (3). Used to store GeoFID and GeoIS values calculated for each subregion grid6) GeoFID/GeoIS implementation Jupyter Notebook .pynb (Count:1)Used in conjunction with Google Cloud Services to train deep learning models & calculate GeoFID/GeoIS values stored in (5)