David B Dean

Machine Learning Researcher and Software Engineer

OpenStreetMap and Machine Learning

OpenStreetMap, Community and Evangelism

I’ve been an OpenStreetMap member since 2007, and was responsible for a lot of the original mapping and social events in Brisbane, Australia.

I have organised around 25 Brisbane Mapping Events over the past decade. I have also evangelized OpenStreetMap to universities and local community groups on many occasions, such as this recent Missing Maps Presentation in October 2017:

If you need to be convinced why OpenStreetMap is a force for good around the world, please watch this great video:

Machine Learning and GIS - Object Detection from Aerial Imagery

These days, I’m mapping very seriously, but I’m also starting a focus on side projects for AI and machine-learning for GIS applications to help the OpenStreetMap community.

I have already done some work in this space, back in 2011, where I worked with Felix Guo to develop a dataset based on high-resolution aerial imagery from NearMap (which was CC-licensed at the time) and OpenStreetMap ground truth. We published our findings at the 2011 Digital Image Computing Techniques and Applications (DICTA) conference:

X. Guo, D. Dean, S. Denman, C. Fookes and S. Sridharan, “Evaluating Automatic Road Detection across a Large Aerial Imagery Collection,” 2011 International Conference on Digital Image Computing: Techniques and Applications, Noosa, QLD, 2011, pp. 140-145.

The paper can be downloaded here, and the slides from our presentation can be viewed below:

Unfortunately, due to lack of interest, the university has not made the database available, and due to newer data in this space, it is unlikely to happen moving forward.

Machine Learning for the Humanitarian OpenStreetMap Team

I’m currently doing open-source development work on the SkyNet project with Development Seed, focusing on approaches that can help the Humanitarian OpenStreetMap Team to more efficiently map buildings, roads, and other features for disadvantaged populations across the world.