项目作者: joshasgard

项目描述 :
Fire Boundary Prediction using Linescan Satellite Imagery of Australia.
高级语言: Jupyter Notebook
项目地址: git://github.com/joshasgard/EY_Data_Challenge_1.git
创建时间: 2021-05-12T00:09:27Z
项目社区:https://github.com/joshasgard/EY_Data_Challenge_1

开源协议:

下载


2021 Ernst & Young Better Working World Data Challenge

Introduction

Participation and submission for the 2021 Better Working World Data Challenge.
Find out more about the fire edge detection challenge in this github wiki

Installation

Refer to the link above for a detailed installation instruction.
Trust me, it takes a bit of effort to do. And you’ll be working in a pre-loaded Open Data Cube Environment on a Microsot Azure VM.
Be sure to take your time and enjoy the experience.
You can also choose to use your local machine, but you’ll need Docker and a lot of processing power.

File Description

There are 3 notebooks in the repo. One was created by Digital Earth Australia to describe the Basic Analysis steps involved when working with Satellite data (Geodataframes and linescans). The remaining 2 are step-by-step analysis followed to make a prediction of the boundaries of fire in a satellite linescan image.

The .csv file contains the first prediction.

Result

.csv file contains the first result for Challenge 1. The accuracy of the prediction was found to be approximately 0.11. I’m still exploring ways of improving this prediction accuracy.

Authors and Acknowledgements

Credits to:

  1. - **Ernst&Young**,<br>
  2. - **Open Data Cube**, <br>
  3. - **Digital Earth Australia**, and <br>
  4. - **Microsoft Azure** for providing free credits