Map-Segmentation

Click on the image to watch the video

Introduction

Everyday Geospatial Data Storages are deluged with millions of optical overhead imagery captured from airborne or space-borne platforms. Manual data interpretation on such a large amount of data becomes an intractable task, hence machine vision techniques must be employed if we want to make any use of the available data. In this project, we have made an application which would deal with semantic segmentation of high-resolution (aerial) images where a semantic class label is assigned to each pixel via supervised classification. Deep learning techniques have shown impressive performance, particularly for image processing. However, a major drawback of using deep learning techniques is that they are extremely data-hungry and using them leads us in exacerbating the limitations of supervised learning, to get enough annotated training data. But on the bright side, these techniques are immune to noise so instead of taking and annotating large datasets we can use publicly available datasets. Such publicly available datasets might contain errors but in return, a very high amount of data is available to us on which we can train our model. As deep learning models would require high computational power and in order to maintain high scalability we have tried to exploit the benefits of Cloud Computing and RESTful services (Representational state transfer). We have successfully built a REST API and deployed it on the institute's server. Our application segments each map image into various segments through which we can perform time series analysis on images and track development in various parts of our country. Moreover, our time series analysis would help us in tracking various environmental changes such as deforestation, afforestation, urbanization, etc and infrastructural changes such as Rural development. Also, we would provide additional support files to load all the detected classes as the layers in QGIS and improve segmentation parts through LabelMe.

HOW TO USE

Install and setup python 3.6

Run this piece of code only if you have any other version of python installed One can check his/her python version by typing

python3 --version

on your terminal and if you get

Python 3.6.0

as your response you have python 3.6 installed as default. or else install python 3.6 as follows

sudo apt install build-essential checkinstall
sudo apt install libreadline-gplv2-dev libncursesw5-dev libssl-dev libsqlite3-dev tk-dev libgdbm-dev libc6-dev libbz2-dev
wget https://www.python.org/ftp/python/3.6.0/Python-3.6.0.tar.xz
tar xvf Python-3.6.0.tar.xz
cd Python-3.6.0/
./configure
sudo make altinstall

Install pip and python virtual environment

sudo apt install python3-pip
sudo apt install python3-venv

Create virtual enviorment naming Map-Seg

python3.6 -m venv Map-Seg
cd Map-Seg
source bin/activate

Clone the repository

git clone https://github.com/parshwa1999/Map-Segmentation.git

browse to directory by typing

cd Map-Segmentation

Install all requirements from requirements.txt

pip install -r requirements.txt
pip freeze > requirements.txt

Migrate Database

cd Achilles
python3 manage.py migrate
python3 manage.py makemigrations
python3 manage.py migrate

Run Django Server

python3 manage.py runserver

let the entire server run.

To visit application page click here and to visit admin page click here

Username: root

Password: 11to1or11

For further details one can refer my Report and Presentation

parshwa1999/Map-Segmentation