Semantic-Segmentation-on-Sand-Dunes-and-Solar-Farms

Semantic segmentation on satellite images of solar farms and sand dunes

Introduction

This repository consists of a summary of my research based project work which was done at Regional Remote Sensing Centre (NRSC, ISRO) 2020. The main objective of this project was to perform Semantic Segmentation on high resolution GeoSpatial Images. There were 3 parts in the project for detecting 3 different things. The first part of the project was for detecting Buildings, the second part for detecting Solar Farms and the third part for detecting Sand Dunes in Rajasthan. The project was aimed at the promotion of autonomous detection of target objects so that they could be precisely tracked, monitored, and it also served as a key indicator for managing disasters as well and for promoting sustainable growth as well.

The Platforms, Libraries & Frameworks which were used-

1. Google Earth Engine
2. Google Colaboratory
3. Geo-Processing Tools: Solaris, GeoPandas, Rasterio, Supermercado, Rio-Tiler
4. Augmentor
5. UNet (FastAI)
6. DeepLab V3+ (TensorFlow 1)