Enhancing Wildfire Detection and Prediction with Deep Learning and Quantum Machine Learning

Wildland fires in Canada pose significant risks, causing extensive damage to ecosystems, property, and human life. The record-breaking 2023 wildfire season, which devastated 18 million hectares, underscores the urgent need for improved detection and mitigation strategies. Recent advancements in deep learning have demonstrated strong potential in early wildfire detection and fire spread prediction, providing critical support for mitigation efforts.
Building on this progress, Quantum Machine Learning (QML) offers a promising avenue to further enhance wildfire management. By leveraging quantum computing’s ability to process complex data structures and optimize algorithms, QML can complement deep learning approaches, improving model performance in challenging scenarios. Synthetic datasets like SWIFT, have already enhanced training for real-world applications, and QML can further improve the efficiency of these synthetic data-based approaches.
This project aims to integrate QML with deep learning to develop cutting-edge solutions for wildfire management. Objectives include generating synthetic visible and infrared data, developing real-time detection systems, predicting wildfire spread through spatiotemporal analysis, identifying potential ignition hotspots, and creating comprehensive wildfire risk maps. The integration of QML is expected to significantly improve detection accuracy, prediction capabilities, and overall wildfire management strategies.

Faculty Supervisor:

Moulay Akhloufi

Student:

Partner:

Federal University of Parana

Discipline:

Computer science

Sector:

Artificial Intelligence; Environmental Science and Technology; Information and Communications Technology; Quantum Science

University:

Université de Moncton

Program: