Abstract: Wildfire is one of the serious natural disasters. Scientific and effective rescue of wildfires and post-disaster loss assessment are extremely important. This research combines PIE secondary development SDK, SegNet and UNet deep learning models, remote sensing and other technologies to design and implement a deep learning based GIS platform for mountain fire rescue and evaluation, which includes functions such as wildfire spreading simulation, automatic extraction of buildings before and after disasters, and scientific assessment of post-disaster losses. This platform can provide more intelligent solutions for the emergency treatment and emergency rescue of wildfire disasters.