When the skies of California turned a haunting shade of orange during the devastating wildfires, one startup took up the challenge by combining computer vision and generative AI.
Emrah Gultekin, the CEO of Chooch, a prominent Silicon Valley-based computer vision company, felt a personal connection to the 2020 wildfires and approached fire officials to offer assistance. They discovered that existing wildfire detection systems were inundated with up to 2,000 false positives per week. These erroneous predictions were triggered by factors such as fog, rain, and smudges on camera lenses.
In a groundbreaking pilot project, Chooch integrated their fire detection software with the camera network. Every 15 minutes, the software analyzed snapshots, searching for any indications of smoke or fire. However, it was Hakan Gultekin, Chooch’s CTO and Emrah’s brother, who had an ingenious idea.
They developed a generative AI tool that automatically generated descriptions for each image, aiding reviewers in accurately identifying the presence of smoke. As a result, the number of false positives plummeted from 2,000 per week to a mere eight.
“Fire chiefs expressed great enthusiasm about implementing the technology within their monitoring centers and the potential it holds,” revealed Michael Liou, the president of Chooch, during a recent webinar where he discussed the project.
Chooch’s innovative generative AI tool equips firefighters in Kern County, California with a real-time dashboard accessible on their smartphones and PCs. This dashboard provides instant alerts, enabling them to swiftly detect wildfires.
In 2020 alone, California witnessed a staggering 9,900 wildfires, scorching 4.3 million acres of forests and resulting in $19 billion in losses. Chooch estimates that preventing even a single fire from spiraling out of control would generate enough savings to sustain the wildfire detection system for 50 years.
