Scene Segmentation

Scene Segmentation: Rethinking Infrastructure Mapping

 

Over the past year, Sobolt worked hard on improving our capabilities to analyse imagery. Our passion combines machine learning and remote sensing data. One natural combination of these two topics is scene segmentation.

Scene segmentation is a process that divides an image into multiple segments, makes raw data more meaningful and easier to analyze. It is used in a vast amount of applications, from self driving vehicles to medical imaging and facial recognition. We use this technique to locate objects on remote sensing imagery. This enables the visibility of even the smallest differences in the image.

 

We combine general data processing techniques with machine learning to analyse the sources, obtaining valuable results in a short amount of time. Thanks to our agile approach, we have been able to participate in several projects, where a precise quantification of the available infrastructure was required. We have correctly identified asphalt and used the segmented data to calculate the surface of asphalt with an error of less than 1%.

Whether you want to differentiate between land used for crops and forests, or you want to identify relevant infrastructure in a sea of data, we have the expertise, tools, and a huge training dataset to realize your ideas.

Would you like learn more about our capabilities or learn what artificial intelligence can do for you?

Comments are closed.