Fire Detection in Images

The main goal of this project is to build a pipeline to detect fire in images with high accuracy.

In addition, secondary objectives are to process images in Python to extract features that can be used by Machine Learning algorithms for tasks such as image classification.

Essentially three methods of feature extraction for images have been explored:

• Pixels

• Histogram of oriented gradients

• Convolutional Neural Networks

Image filtering and equalisation as an image processing tool is also studied.

This project could have a multitude of practical industrialised applications, from fire detection using drones to the implementation of fire alarms based on video surveillance.

• Images Processing
• Filtering Images
• Feature Extraction Methods for Images
• Applying inner evaluation
• Selecting the best pipeline
• Applying Outer Evaluation
• Predicting New Images