Forest Fire Detection

The aim of this project is to create neural network models for forest Fire detection using Flask and TensorFlow, and integrate them into a website for convenient use.

Note: This code is not optimized for production enviroment.

For documentation of the code in this repository, please see the Wiki. See below for instructions to install and run the web server.

Requirements

• Python => 3.7 and =< 3.10
• pip sudo apt install python3-pip
• git sudo apt install git

How to install the code

git clone [email protected]:akdasUAF/ForestFireDetection.git

If you do not have access to the repository from your command line, but your github account does have access, then follow the instructions at this link to set up ssh keys to provide access. Then, run the git clone command again.

Then, download the three files in this google drive folder (these are too large for github to accept, and one contains another git repo). Move yolov5.tar.gz and dataset.tar.gz into the repo directory and extract them using

gunzip yolov5.tar.gz
gunzip dataset.tar.gz
tar -xf yolov5.tar
tar -xf dataset.tar

Then, move dbn_pipeline_model.joblib.gz to the Models/weights directory inside of the repo, and extract with

gunzip dbn_pipeline_model.joblib.gz

How to install dependecies

pip install -r requirements.txt

If you get warnings, you may need to add the following line to your .bashrc file (the path being added to PATH may vary depending on your operating system):

export PATH="/home/$USER/.local/bin:$PATH"

To do this, run

echo 'export PATH="/home/$USER/.local/bin:$PATH"' >> ~/.bashrc

Then, close and re-open your terminal or server connection, or just run

source ~/.bashrc

How to run

To run the web server, just do:

python3 app.py

If you want it to run persistantly (staying up after you close your terminal connection) do:

nohup python3 app.py &