Wildfire Detection
Overview
Wildfire detection studies how to identify active flames, thermal anomalies, and smoke from satellites, aircraft, and UAVs, often before fires appear in routine reports. Methods range from thresholding and gradient-boosted classifiers to CNN, YOLO, and U-Net models, pretrained representations, and weakly supervised learning; they use thermal, multispectral, and RGB data, with compact versions designed to run onboard satellites or drones. Benchmarks test transfer across incidents and regions, severe class imbalance, nighttime performance, and RGB-thermal reasoning, while robotic and satellite systems explore coordination under power, bandwidth, and latency limits. Reliable early detection remains difficult in smoke, unfamiliar regions, and rare-fire conditions: larger models and synthetic data do not guarantee generalization, so careful validation and operational reliability are central concerns.
Latest papers
21 papers0 papers in the last four weeks, down 100% on the four weeks before. 0.0% of all new papers.