Wildfire Detection System

Demo

Machine learning Module

The folder contains a working machine learning project that is able to detect "fire", "no-fire" and "start-fire" in a mp4 video. The module is being used as backend service on a Django server to process multiple video at the same time. The module can also be used for Live-video feed.

Setup

Mac Setup below but it should also work in Linux, Ubuntu.

Requirements & Setup:

Machine learning module

• Download the project
• Python 3.6.12 is required
• Setup Python virtual environment Link
• Go into the machine_learning module directory
• Install the all the requirements: pip install -r requirements.txt
• h5py<3.0.0 is required:
  pip install 'h5py<3.0.0'
• If there tkinter error, please use tkinter as backend setup with Python as the project requires Python as a framework. IF you have problems with tkinter, run the commands in the link: Link
• The model file is not included in the project and can be downloaded here. Place model under the folder: 'machinelearning/model-saves/Inception_based/'

Django Server

• install the latest version of Django.
• Go back to the root folder and install all the requirement: mysqlclient ->

pip install mysqlclient

Run the Project:

5 input videos are displayed in the index.html file. These videos are used for wildfire detection and the URLs can of the videos can be viewed under view.py file

input videos.

• The project already contains sample videos under WildfireDetectionApp/static folder. You can replace these videos with your own input videos and update the same videos for display in index.html
• change the videos links to your directory in view.py file under the WildfireDetectionApp folder.

Start the server

Under the root folder, use the command line to start the project.

python manage.py runserver

Once the server start, It will launch the home (index.html). In the command-line, you can view the multiple videos being processed and the model giving out prediction for each frame.