The following projects and initiatives use satellite imagery, oftentimes in combination with various forms of computer vision, to track phenomena of interest in the built and natural world. Until recently, much of this work has been tuned for specific geospatial problems.
Abelson, B.; Varshney, K.; and Sun, J. "Targeting direct cash transfers to the extremely poor". In Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining, 2014. 1563–1572. ACM.
Begall, Sabine et al. "Magnetic alignment in grazing and resting cattle and deer". Proceedings of the National Academy of Sciences 105, no. 36 (9/9/2008), pp. 13,451-13,455. DOI: 10.1073/pnas.0803650105
Begley, Josh. "Prison Map", 2012. prisonmap.com. Accessed 5/16/2016.
Coolidge, Matthew and The Center for Land Use Interpretation. "Land Use Database". clui.org. Accessed 5/14/2016.
Finer, M.; Novoa S.; and Olexy, T. "MAAP #30: Gold Mining Invasion of Tambopata National Reserve Intensifies". maaproject.org, Monitoring of the Andean Amazon Project, 4/12/2016. Accessed 5/15/2016.
Fretwell, Peter T.; Staniland Iain J.; and Forcada, Jaume. "Whales from Space: Counting Southern Right Whales by Satellite". PLoS ONE 9(2): e88655. doi:10.1371/journal.pone.0088655.
Garling, Caleb."Startup Promises Business Insights from Satellite Images". TechnologyReview.com, MIT Technology Review, 3/16/2015. Accessed 5/14/2016.
Gross, Benedikt and Lee, Joey. "Aerial Bold". A typeface comprised of letters found in satellite images. aerial-bold.com, 2016. Accessed 5/14/2016.
Hogenboom, Melissa. "Watching penguins, and their poo, from space". BBC.com. 12/10/2014. Accessed 5/14/2016.
Howard, Brian C. "Tiny Team Uses Satellites to Bust Illegal Fishing Worldwide". NationalGeographic.com, 6/15/2015. Accessed 5/14/2016.
Howard, Brian C. "New Theory Behind Dozens of Craters Found in Siberia". news.nationalgeographic.com, National Geographic Society, 2/27/2015. Accessed 5/14/2016.
Novoa, S.; Fuentes, M.T.; Finer, M.; Pena N.; and Julca, J. "MAAP #18: Proliferation of Logging Roads in the Peruvian Amazon". maaproject.org, Monitoring of the Andean Amazon Project, 10/30/2015. Accessed 5/14/2016.
Odell, Jenny. Satellite Collections (2009-2011). Collages of similar-looking things that the artist has cut out from Google satellite imagery. Accessed 5/14/2016.
Onformative Design. Google Faces. 2013. Faces in satellite imagery, discovered with a face tracker.
Pell, Richard W. and Allen, Lauren B. "Preface to a Genealogy of the Postnatural". In Intercalations 2: Land & Animal & Non-Animal, Edited by Anna-Sophie Springer & Etienne Turpin. K. Verlag and the Haus der Kulturen der Welt, Berlin. 2015.
Raymond, Nathaniel A.; Card, Brittany L.; Baker, Isaac L.; and Al Achkar, Ziad. "Satellite Imagery Interpretation Guide: Intentional Burning of Tukuls". http://hhi.harvard.edu. Signal Program on Human Security and Technology at the Harvard Humanitarian Initative, 2014. Accessed 5/15/2016.
Raymond, Nathaniel A.; Vidan, Gili; and the staff of the Signal Program on Human Security and Technology at the Harvard Humanitarian Initiative (Isaac Baker, Ziad Achkar, Brittany Card, Benjamin Davies, and Steve Juntunen). "Tents and Tukuls: Lessons from the Development of AMALGAM". thetechchallenge.org, Mass Atrocity Prevention Tech Challenge, 12/10/2014. Accessed 5/15/2016.
Rothfeld, Michael and Patterson, Scott. "Traders Seek an Edge With High-Tech Snooping". The Wall Street Journal, 12/18/2013. Accessed 5/15/2016.
Trevi, Alexander. "Atomic Gardens". Pruned.blogspot.com. 4/20/2011. Accessed 5/14/2016.
Venkataramanan, Madhumita. "Space archaeologist discovers lost cities with satellite imagery". Wired.co.uk, Wired Media, 11/14/2014. Accessed 5/15/2016.
The following research projects and open-source codebases are specifically concerned with the application of 'deep-learning' techniques to satellite imagery. Several of these projects also employ OpenStreetMap labels as training data, as we do.
Learning to Detect Roads in High-Resolution Aerial Images by Volodymyr Mnih and Geoffrey E. Hinton (2010), a seminal paper in the application of machine learning to the problem of making assertions about satellite imagery.
Machine Learning for Aerial Image Labeling, Volodymyr Mnih's 2013 doctoral thesis from the University of Toronto, under the advisement of Geoff Hinton.
Humanitarian Mapping with Deep Learning by Stanford graduate student, Lars Roemheld (2016). Like the Terrapattern project, this project uses OpenStreetMap (OSM) to help train a neural net, in order to help support map creation in the developing world.
OSM-Crosswalk-Detection by Marcel Huber (2015). Developed at the University of Applied Sciences Rapperswil, this project again trains deep learning models with OSM labels to locate Swiss crosswalks.
DeepOSM by Andrew L. Johnson, Kevin Lacker, Zain Memon, and Dan Silberman (2016). DeepOSM trains a neural network with OSM labels and aerial imagery. DeepOSM uses RGB+infrared images from the U.S. National Agriculture Imagery Program (NAIP), with raw OSM extracts from Geofabrik.
skynet-data (2016), by Anand Thakker at DevelopmentSEED, is "a pipeline to simplify building a set of training data for aerial-imagery-based and OpenStreetMap-based machine learning." Skynet uses OSM QA Tiles to generate ground truth images where each color represents some category derived from OSM features.
Costea, Dragos and Leordeanu, Marius. "Aerial image geolocalization from recognition and matching of roads and intersections". arXiv:1605.08323, 5/26/2016.