DigitalPath is igniting change in the golden state — using computer vision, generative adversarial networks and a network of thousands of cameras to detect signs of fire in real time. In the latest episode of NVIDIA’s AI Podcast, host Noah Kravtiz spoke with DigitalPath system architect Ethan Higgins about the company’s role in the ALERTCalifornia initiative, a collaboration between California’s wildfire fighting agency CAL FIRE and the University of California, San Diego. DigitalPath built computer vision models to process images collected from network cameras — anywhere from eight to 16 million a day — intelligently identifying signs of fire like smoke. “One of the things we realized early on, though, is that it’s not necessarily a problem about just detecting a fire in a picture,” Higgins said. “It’s a process of making a manageable amount of data to handle.” That’s because, he explained, it’s unlikely that humans will be entirely out of the loop in the detection process for the foreseeable future. The company uses various AI algorithms to classify images based on whether they should be reviewed or acted upon — if so, an alert is sent out to a CAL FIRE command centers. There are some downsides to using computer vision to detect wildfires — namely, that extinguishing more fires means a greater buildup of natural fuel and the potential for larger wildfires in the long term. DigitalPath, along with UCSD, are exploring using high-resolution LIDAR data to identify where those fuels can be let out in the form of prescribed burns. Looking ahead, Higgins foresees the field tapping generative AI to accelerate new simulation tools — as well as using AI models to analyze the output of other models to doubly improve wildfire prediction and detection. “AI is not perfect, but when you couple multiple models together, it can get really close,” he said.