Deep Learning for classification of building materials using hyperspectral imaging


Supervisory Chair

External Member

Dr Nariman Habili


Instead of the three bands of an RGB image, hyperspectral images contain 10s or even 100s of bands, enabling significantly more information to be derived from a scene. The project aims to develop deep learning methods to classify building materials (such as bricks, concrete, glass etc) using hyperspectral images of building facades. Our current building dataset consists of 531 fully annotated hyperspectral and RGB images. 


Updated:  10 August 2021/Responsible Officer:  Dean, CECS/Page Contact:  CECS Marketing