Скачать или смотреть Satellite Cloud Generator with Mikolaj Czerkawski

In this video, Robin catches up with Mikolaj Czerkawski to discuss the Satellite Cloud Generator. This is a tool on Github which makes it straightforward to generate realistic clouds and apply them to imagery. This can be used as an augmentation during model training to make models more robust to the presence of clouds.

https://github.com/cidcom/SatelliteCl...
https://mikonvergence.github.io/

To learn more about deep learning applied to satellite & aerial imagery head to https://www.satellite-image-deep-lear...

Bio: Mikolaj is a final-year PhD student at the University of Strathclyde in Glasgow. His work extends over a range of topics, including radar signal analysis, remote sensing and general computer vision problems; such as generative modelling, image inverse problems, and learning in low-data regimes.
In the last year, he has been heavily focused on the task of processing clouds in satellite images, which includes detection, segmentation, and finally, removal of the clouds from affected images. He is the author of two open-source solutions related to this problem, which he published this year on GitHub, one for the removal of clouds with no pre-training, and the other for simulating the presence of clouds in clear images.