Wildfire detector & Risk Analysis

• 

  INPUT for Wildfire Detection

• 

  OUTPUT for Wildfire Detection

• 

  Data Visualization of Wildfire Risk Analysis

• 

  Model and it's performance - We trained a Unet model on the data and it's prducing >95% accuracy.

• 

  Model prediction on the test data for risk analysis

• 

  Pipeline of the project

• 

  Workflow of the project in real World

Inspiration

We were inspired from the Nasa space apps challenge to create something that contributes to the environment. Forest fire or wildfires are the ever growing problems that contributes most to climate change.

What it does

The project has two functionalities: a. Detection Is there a fire or not b. Risk analysis 1. Analyze surrounding vegetation If dry or wet 2. Wind strength if strong or weak 3. Power Lines if present or absent

How we built it

It was build in the following manner: -

1. Understanding the problem
2. Data Preparation
3. Pre-processing: - Data Augmentation -> Normalization -> Model Architecture
4. Data Modelling
5. Deploying: - Desktop->NGINX->Users->AWS

Challenges we ran into

1. Time Management
2. Setting up the GPU
3. Getting a perfect accuracy
4. Data Preparation

Accomplishments that we're proud of

1. Successful Deployment of wildfire detector
2. And a great accuracy

What we learned

1. How to deploy using AWS
2. How to manage things in a short span of time

What's next for Wildfire detector & Risk Analysis

We want to add some more functionalities like a recommendation system to recommend ways of how to control wildfire. Also recommend what all products can help in stopping wildfire in general.

Built With

• amazon-web-services
• deep-learning
• machine-learning
• python