I implement Deep Learning techniques to solve segmentation and object detection problems in the observation of Earth, and of other planets. Spe- cifically, my work whilst visiting the Φ-lab has fo- cused on designing novel convolutional models that are able to handle images with arbitrary spectral responses as input. This sensor-independence is a means of achieving greater inter-operability between cloud masks for satellites to gain more value from existing labelled datasets that use images from a wide range of sensors. To this end, I am also interested in dataset harmonisation, and have developed a machine learning pipeline that are often time-consuming and tedious. toolbox to help restructure multiple segmentation datasets into a single shared format.
My background is in physics and Space science. I obtained my BSc in physics from the University of Oxford, and have a MSc in Space Science from UCL. I am now pursuing a PhD in the Imaging Group at Mullard Space Science Laboratory, UCL.
Back to Our People page
