Physical Informality Through Earth Observation And Deep Learning
1. Capture physical/morphological dimension of deprived areas by combining satellite imagery data and deep learning technique.
2. Rationale: Deprivation is assumed to be manifested by physical/morphological appearance.
3. The U-Net is used as a benchmark model.
4. The model scripts along with dependencies are to be encapsulated into standalone environment by using the Docker, so that reproducibility of the entire workflow can be ensured under equal experimental configuration.