Water Recognition

Introduction

Bachelor Thesis Project - AI Model that recognises bodies of water using satellite imagery. This project is being developed as part of an experiment conducted for Software Engineering and Management Bachelor Thesis at University of Gothenburg. The goal of this project is to train an AI model that recognizes bodies of water using LANDSAT satellite imagery.

Team

The research group consists of the following members:

• Ivica Crnkovic (Academic Supervisor)
• Arvin Esfahani Zadeh (Student)
• Sebastian Baszczynski (Student)

Prerequisites

In order to use this project it is required to have access to and install the following:

• Python
• Conda
• Jupyter Notebook
• GDAL Library
• Google Earth Engine Library
• GEEMAP
• TensorFlow-GPU
• Keras
• Numpy
• Matplotlib

Installation

Python and Conda

Install python from the following link and use the this guide to install conda, a python package and environment manager. After successful installation of Python and Conda, Conda will be used to create a new environment and install the dependencies.

Create and use a new environment

In order to create a new environment by entering the following text in your command line interface:

conda create -n water-recognition-env vs2015_runtime=14 python=3.6

Activate the conda environment:

conda activate water-recognition-env

Install Libraries

Use the following commands in your preferred command line interface. Ensure you're in the correct conda environment by activating your preferred conda environment.

Jupyter Notebook

conda install -c conda-forge jupyterlab

GDAL

conda install -c conda-forge gdal

Google Earth Engine

conda install -c conda-forge earthengine-api

GEEMAP

conda install -c conda-forge geemap

TensorFlow-GPU

conda install -c anaconda tensorflow-gpu

TensorFlow-GPU

conda install -c anaconda tensorflow-gpu

Keras

conda install -c anaconda keras

Numpy

conda install -c anaconda numpy

Matplotlib

conda install -c anaconda matplotlib

OpenCV

conda install -c conda-forge opencv

TQDM

conda install -c conda-forge tqdm

TFLearn

conda install -c contango tflearn

Using your environment in the Jupyter Notebook

In order to use your environment in the Jupyter Notebook you need to run the following command:

python -m ipykernel install --user --name=water-recognition-env

Now you can run Jupyter Notebook and you will have water-recognition-env as one of its kernel. It is highly recommended to use this kernel as it has all of the required dependecies installed.