I am interested in the spectral signatures of landslides and other catastrophic events such as fires and floods.
I investigate applying deep learning techniques to remote sensing images, to gauge what datasets and techniques perform better for image segmentation and classification.
I am interested in applying novel AI technologies to the detection of natural disaster in real time.
MPhys Physics (Hons), University of Edinburgh, 2014 – 2019
-Masters Dissertation: "Impact of East Asian sulphate aerosol emissions on climate and extremes".
Supervised by Dr Massimo Bollasina
PhD in Geology and Geophysics, University of Edinburgh, 2019-present
Learning to recognize landslides and catastrophic landscape change with deep neural networks.
Efficiently identifying landscape change caused by natural disasters such as landslides, floods and fires can be challenging, especially in a time constrained scenario.
However, the increasing availability of high resolution satellite imagery and open source AI and deep learning frameworks have brought new opportunities to improve research in this field. The aim of this project is to combine satellite data with topographic and geospatial information to train deep neural networks in classification and segmentation tasks focused on natural disaster imagery. An analysis of images in multiple spectral bands will be performed with different satellite types and resolutions, using multiple neural network architectures and techniques to explore the optimal set up. As a result we will be able to efficiently identify natural disaster satellite images to allow for real time response and to improve on current mapping techniques.
N/A