mathew-endreson/NASA_wildfire_algeria

An end-to-end Python project that uses a TensorFlow CNN to detect wildfires in Algeria from NASA satellite imagery. # Automated Wildfire Detection from NASA Satellite Imagery ## Project Overview This project is an end-to-end machine learning pipeline that automatically detects wildfires from satellite imagery. It uses a Convolutional Neural Network (CNN) trained on publicly available data from NASA to identify fire and smoke patterns in satellite tiles of Northern Algeria. The goal is to demonstrate a real-world application of computer vision for environmental monitoring and disaster response. The entire workflow, from data acquisition to prediction, is automated with Python scripts. ## Key Features - **Automated Data Acquisition:** Downloads satellite images and corresponding fire data directly from NASA's public APIs (GIBS and FIRMS). - **Data Processing:** Automatically labels and sorts images into `fire` and `no_fire` classes to create a custom dataset. - **Deep Learning Model:** Implements a CNN using TensorFlow and Keras to perform binary image classification. - **Visualization:** Generates a "before-and-after" GIF to clearly visualize the model's predictions on a test image. ## Technology Stack - **Language:** Python 3 - **Core Libraries:** - TensorFlow / Keras: For building and training the deep learning model. - Pandas: For handling and processing the fire location data. - Requests: For interacting with NASA's APIs. - OpenCV-Python & Pillow: For image processing and manipulation. - Imageio: For creating the output GIF. - Matplotlib: For plotting the model's training history. ## Project Structure asa_wildfire_algeria/ │ ├── data/ # Folder for the image dataset (created automatically) ├── venv/ # Python virtual environment │ ├── download_data.py # Script 1: Downloads the image dataset ├── train_model.py # Script 2: Trains the machine learning model ├── predict.py # Script 3: Uses the model to create the final GIF │ ├── firms_data.csv # Input: Raw fire location data from NASA ├── requirements.txt # L …