Mapping Land Cover with Machine Learning

Lynker Analytics is a privately-owned data science and analytics company based in Wellington. In this presentation, I will discuss and demonstrate how machine learning—a subfield of artificial intelligence (AI)—has progressed to the stage where computer vision and deep learning in image analysis and classification is now viable for geospatial data production.

I will explain what’s involved in successfully apply machine learning to accurately map land use, land cover and land activity. I will explain how iterative semi-supervised machine learning can be used to optimize and minimize the amount of data needed to train industrial artificial neural network models and then show how this method can be used to generate a range of high-resolution GIS features including vegetation, deforested land, buildings, roads, bare earth, water, impervious surfaces and more.

Finally, I will discuss the reliability of this approach and the resulting spatial data layers and what this could mean for organisations maintaining and managing geographic information systems.

Held Friday 27 March 2020