Melbourne Google Street View imagery dataset

The data presented in this article is related to the research article entitled "Urban design using generative adversarial networks: optimising citizen health and wellbeing" (Wijnands et al 2018). The data consists of Google Street View (Google Maps, 2017) imagery (4,473,991 images, 8-bit JPEG at 256x256 resolution) from four headings (0, 90, 180, and 270 degrees) at 1,118,534 locations in the greater metropolitan area of Melbourne, Australia. Locations were determined using the nodes of the vector lines in the PSMA Street Network dataset (PSMA 2018) and data was post-processed by removing indoor images. Please cite this paper if you use the dataset. The data is broken up into four archives, 000.zip, 090.zip, 180.zip, and 270.zip, containing the imagery from each compass heading. A csv file (contained in MelbourneStreetViewImagesData.zip) provides a mapping between the filenames, location names, direction, latitude, and longitude.