AI-Driven Temporal-Spatial Wildfire Monitoring: A Hybrid Approach Using Edge Computing and Cloud-Based Analytics

Abstract

The growing severity and frequency of wildfires require sophisticated real-time detection techniques to minimize damage to the environment and economy. In this poster, data sources including temporal data from ground sensors (e.g., temperature, humidity, gas/smoke concentration), spatial data from satellite, and contextual data such as weather conditions and historical fire records has been used. The framework is developed on TensorFlow/PyTorch for model development and Stable-Baselines3 for reinforcement learning. The experimental results have shown effective improvements in wildfire detection accuracy, reduced false positives, and faster response times. This research contributes to scalable and autonomous wildfire monitoring systems by offering a transformative solution for real-time detection and mitigation.