Forest Fire Smoke Detection

Forest fires are one of the most devastating environmental disasters, causing severe damage to ecosystems, air pollution, and endangering human lives. This project leverages Computer Vision to classify forest imagery into three categories: fire, smoke, and neutral (no fire/smoke).

Target User

[ Local governments ]
/ Fire Departments

View Website

[ Huggingface ] [ Huggingface ]

Time Line

[ 2025 ]

View Project

[ Github ] [ Github ]

Objective

Automatically detect forest fires and smoke using Convolutional Neural Networks (CNN).

Data source: Forest Fire Smoke Dataset

Process

Data preparation: Cleaning data, Data splitting, Encoding categorical columns, and Scaling numeric features.

EDA (Exploratory Data Analysis): Identify important features such as age distribution, call duration, previous campaign results, and job status.

Modeling: Develop 5 machine learning models, which will later be optimized with hyperparameter tuning and validated using recall and cross-validation.

Overview

The model is based on a Convolutional Neural Network (CNN).
Image preprocessing and data augmentation are applied.
Evaluation includes accuracy metrics, confusion matrix, and prediction visualization.