Isabelle Marguin Efremovski

We provide powerful insights into the world by extracting information from satellite imagery fully automatically. A scalable artificial intelligence builds the core of the blackshark platform, detecting features with incredible precision and speed. A novel machine learning approach was developed to train a neural network for new features rapidly, enabling blackshark to serve various use cases in many different industries.

Input sources supported are:

satellite imagery
2D photographs (aerial, street level, …) point clouds (photogrammetry or lidar) HD map data OSM & more Examples for feature detection:

buildings (footprint, height, type, roof, orientation, facade) vegetation streets land classification flood detection & more We can answer questions such as:

Can my drone land in this backyard?
How many solar panels are installed in India?
How many tents does a certain refugee camp have?
How many cars are parked in a specific parking lot?
We can also can also fully automatically and semantically reconstruct detected features in 3D, using a patented technology for efficiently storing and streaming petabytes of data, as showcased in our work on the Microsoft Flight Simulator.

Our team of 50+ data scientists, geospatial engineers, 3D rendering programmers and developers, self-funded and based in San Francisco, US, and Graz, Austria, Europe’s computer vision hub.