The powerful combination of AI and Earth Observation Data:
01 February 2019
At DHI GRAS we routinely conduct coffee talks, where our experts present their latest research and development within advanced image analytics, data science or other trends within Earth Observation.
This week our AI expert Philip Graae demonstrated results of applying Deep Learning on Earth Observation Data.
By using principles of transfer learning, we have trained a state-of-the-art object detection neural network to detect and segment small and very detailed objects within earth observation image data.
With proper tuning and by using techniques such as data augmentation, it is possible to get good results with just a small amount of training data as input to the AI.
We rely on such deep learning techniques when traditional image analytics is not sufficient to extract the information from the imagery.