AI Model Generates Synthetic Wildfire Satellite Imagery to Address Data Shortage

Researchers demonstrate that EarthSynth, a diffusion-based AI model, can create realistic post-fire satellite images from burn masks without task-specific training, potentially solving the scarcity of labeled wildfire data for detection systems. Inpainting-based approaches outperformed full-image generation across quality metrics, offering a path to augment training datasets for improved wildfire monitoring.