DLGeo

This project aims to calculate the Photovoltaic potential of Providence, RI using Image Segmentation via a U-Net based model. This is an attempt to re-implement the paper Urban solar utilization potential mapping via deep learning technology: A case study of Wuhan, China by Z. Huang et al, with a specific focus on Providence RI.

References

• Urban solar utilization potential mapping via deep learning technology: A case study of Wuhan, China by Z. Huang et al.

• Semantic Segmentation of GEE High Resolution Imagery by Mort Canty

• Understanding U-NETs

• Loading large Datasets in Keras

Folder Structure

• code-try-1 : Contains some U-Net code from our first attempt at implementing this paper.
• utilites : Some utility code we built to pre-process and tile data
• segmentation_notebook.ipynb : Our final logic/model in the form of a Colab notebook. This can also be found in its latest format here. This borrows heavily from Mort Canty's segmentation algorithm, as listed above.