Computer Vision Algorithms for fast information extraction of remote sensing data
We are using Computer Vision in the form of Convolution Neural Network (CNN) in the area of remote sensing. The traditional methods depend on the intensity of pixel level interpretation while the modern techniques are focused in the semantic understanding of the images. With the help of artificial intelligence algorythms it is possible to recognize objects automatically. We can provide knowledge how to handle large amounts of data and with our hardware it is possible to exploitate spatial information from images in a rapid and cost efficient way.
Technically and methodologically up to date
In addition to conventional use of satellite data, machine learning methods are used to identify certain patterns. Artificial intelligence is helpful, for example, to recognize objects that appear in combination with certain other objects. In view of the diversity of landscape forms, however, the use of such techniques in geoinformatics is more difficult than, for example, in face recognition, which is already well established.
Our approach also uses unsupervised AI. With this method the system learns to independently identify patterns and connections in the data in an explorative manner. The algorithm is fed with input data without being given a goal – a certain type of pattern that it should look for.
With this approach our customers can save time, money and lifes.