The project leverages U-Net–based architectures trained on the FLAME Dataset to detect wildfire regions from aerial imagery captured by drones. In addition to pixel-level fire segmentation, the system incorporates temporal consistency analysis and a rule-based alert mechanism to trigger wildfire warnings in video streams.
-
Updated
Jan 8, 2026 - Jupyter Notebook