Data and Outputs for Explainable Cross-View Geo-Localisation

The file named data.tar.gz includes both real street-view images (SVIs) from the validation set of the CVUSA dataset (the original dataset can be requested here) and synthetic SVIs generated by each framework. Additionally, it contains the corresponding Grad-CAM heatmaps generated by the trained models in .tif format, as well as segmentation masks produced by the pre-trained Mask2Former model and the Output Space Adaptation pipeline in .png format. The code associated with this data can be accessed at https://github.com/ayushdabra/explainable_cross_view_geolocalisation