On Sept 11, 2019, Salar Ghaffarian successfully defended his thesis proposal. Salar’s proposal was titled “Automatic Mapping of Residential Rooftops using High-Resolution Thermal Imagery”. In order to meet the needs of our commercial partner (MyHEAT) this research proposes two main goals: (1) optimize a leading-edge Convolutional Neural Network (CNN) method for automatic and accurate rooftop delineation from MyHEAT’s existing H-Res TIR imagery and (2) define the optimal spatial resolution for CNN based rooftop delineation. By satisfying these goals, we expect to: (i) reduce MyHEAT’s ancillary data acquisition/processing costs as their (optimal resolution) TIR imagery will be the only required data source for HEAT loss metrics, (ii) speed up their entire analytical pipe-line, as there will be no need to acquire, correct, or negotiate for auxiliary data, (iii) be able to apply the optimized model to existing and future archives of MyHEAT TIR imagery.
All committee members noted how well written the proposal was. Well done Salar.