Explore the intersection of machine learning and geospatial image analysis in this 33-minute conference talk. Discover how cutting-edge machine learning techniques can be applied to georeferenced images, enabling not only object recognition but also precise location identification. Learn about Norkart's approach to combining geographical data expertise with deep learning methods, including the implementation of convolutional neural networks for segmentation on multispectral and RGB geospatial images. Examine real-world applications such as calculating solar panel effectiveness, detecting illegal buildings, and improving risk analysis for insurance purposes. Gain insights into the revolutionary potential of machine learning in extracting valuable geographical information from aerial and satellite imagery.