Master Thesis: GAN post-processing for Mosaics - Linköping, Sverige - Arboair

Beskrivning

At the same time, forest management could be conducted both more profitably and sustainably if we had high-quality digital decision-making data for the prevailing state of the forest.

To this end Arboair is leading the transition to a digitized, sustainable precision forestry with high precision image analysis driven by AI algorithms.

Currently Arboair creates mosaics using low overlap high resolution images, resulting in mosaics with irregularities, stitches that went a bit wrong and so on.

These mosaics are in turn used internally for analysis but also for visualizing the order for the customer, which would require some post-processing done to increase the visual fidelity.

Therefore the aim of this research would be to try to smooth and fix our mosaic using GAN before visualizing them for the customers.