Skip to main content

Check out the latest Austrian AI podcast where Stefan Habenschuss, our Head of Computer Vision and Machine Learning, explains our massive scalable aerial image segmentation system. Our system creates photorealistic 3D digital maps containing more than 1.5 Billion 3D building reconstructions at an astonishing level of detail.

Running inference on imagery from around the world poses its own unique challenges. Challenges such as achieving the same quality across all continents, cultures, and image sources. Stefan discusses various aspects of our pipeline including ML-assisted assessment, enhancement of image quality, inference, vectorization, and 3D reconstruction. The results of’s work are used to generate virtual environments well-suited for a variety of applications. Applications as diverse as sensor training for autonomous driving, enabling new services in government, insurance, and AAA gaming, just to name a few.