Home - orthoseg/orthoseg GitHub Wiki

Welcome to the orthoseg wiki!

As the name implies, the purpose of the project is to help you to apply image segmentation on orthophotos. Image segmentation is quite a popular computer vision topic, eg. for use in self driving cars. The same technique can also be used on orthophotos: images in the viewpoint as available in eg. google maps.

The platform the photos were captured on (airplane, satellite, drone,...) obviously doesn't matter, but at the moment only 3 (possibly 4) bands/channels are supported.

Especially if the images you want to process are accessible by a standard WMS server, you'll be up-and-running very quickly.

Orthoseg uses deep neural networks, so access to a decent GPU is VERY recommended, otherwise you'll need to be (very) patient, or it might not work at all.