SDSC Develops Deep Learning Model to Detect Wildfires

New technique could result in earlier warning for impending danger

The WIFIRE lab, at the San Diego Supercomputer Center (SDSC), has developed a deep learning model to improve the detection of wildfires. The model uses real-time camera footage, satellite-based fire detection information, and weather data for a multi-modal approach to fire detection enhanced with AI techniques. The goal is to detect patterns that could indicate the first stages of a wildfire.

According to the report on the SDSC website, the model resulted in a 13.6% improvement in time to detect a fire, which could translate to valuable minutes saved in notifying rangers and residents of the affected region.

See the full study in the journal Remote Sensing.