AI Image Wildfire Detection

This research focuses on early wildfire detection using AI-based image classification. The research explores the use of Convolutional Neural Networks (CNNs) to automatically distinguish between images of fire, smoke, and non-fire scenes. The goal is to enable rapid, accurate detection of wildfires from images, supporting real-time monitoring and emergency response.

Project Structure

• Research Paper:
  The full research and methodology are documented in Early Wildfire Detection Using AI-Based Image Classification.pdf.
  This PDF details the problem statement, dataset preparation, model architectures, experimental results, and conclusions. It is the primary source for understanding the scientific background and findings of this project.

• Code Implementation:
  The main code for data processing, model training, evaluation, and visualization is provided in classifier.ipynb.
  This Jupyter notebook walks through:

  • Downloading and preparing the dataset
  • Removing duplicate images and cleaning data
  • Encoding labels and splitting data into train/validation/test sets
  • Training and evaluating baseline models and CNN architectures
  • Visualizing results and analyzing model performance

How to Use

1. Read the Research:
   Start with the PDF file to understand the motivation, approach, and results.

2. Run the Code:
   Open classifier.ipynb to explore the code, reproduce experiments, and visualize results. The notebook is organized in logical sections matching the research workflow.

Summary

• Research PDF:
  Comprehensive explanation of the wildfire detection approach, experiments, and findings.

• Classifier Notebook:
  End-to-end code for dataset handling, model training, and evaluation, supporting the research with practical implementation.

For any questions or further information, please refer to the respective files.