Advanced Computing in the Age of AI | Thursday, April 22, 2021

California’s Sonoma County Adds AI to Wildfire Alert Network to Detect Destructive Blazes 

Source: Alchera Inc.

California continues to prepare for another round of wildfires by deploying AI-based early-warning systems designed to spot outbreaks before they become infernos.

Sonoma County is the latest to adopt AI technology in the Golden State’s ongoing efforts to monitor and prevent the spread of deadly wildfires that have previously scorched entire towns. County officials announced this week they have awarded a contract to Alchera Inc. of Seoul, South Korea, to develop a fire-detection platform based on the company’s machine learning technology.

Alchera’s visual anomaly detection system scans standard video to spot forest fires. The company (KOSDAQ: 347860) used millions of video images collected by remote cameras to train its fire-detection algorithm, developing what the startup claims is the first commercial software service for detecting wildfires.

In fact, Alchera is among a growing number of AI startups developing wildfire detection systems as California’s forests burst into flames during its annual dry season. For example, Silicon Valley startup Buzz Solutions introduced an AI-based system last year used to analyze millions of power line images to detect grid flaws linked to many of the state’s conflagrations.

Similarly, AI leaders such as IBM, Microsoft and Nvidia have collaborated on cloud-base AI pipelines that process field data collected by drones.

Alchera said it would begin implementing its real-time wildfire detection system this month. The project is scheduled to run though March 2023. Sonoma County’s wildfire alert system includes a network of video cameras, many attached to cell towers, that scan mountains and ridges in the heart of California’s Wine Country.

Alchera’s claims its proprietary AI technology using time-series data gleaned from video cameras reduces the number of false positives associated with conventional detection systems. The platform searches each image via a convolutional neural network to spot potential anomalies. Long short-term memory models are then applied to analyze patterns in video footage to pinpoint anomalies.

Alchera's AI-based detection system scans video images to spot the ignition point of potential wildfires. (Source: Alchera Inc.)

Lastly, times-series analysis is used to spot patterns over time to confirm an actual anomaly.

The framework is used, for example, to discern the diffusion pattern of distant smoke as a way of pinpointing the ignition point. That location data can be passed along to firefighters before a wildfire erupts.

Alchera said it has been developing its visual anomaly detection algorithms since 2018, collecting images from previous wildfires to train its machine learning model. South Korea’s largest utility, Korean Electric Power Corp., was the first customer to deploy the AI-based system.

With a claimed accuracy rate of 97 percent, Sonoma County awarded a contract to Alchera reportedly worth $300,000.

“This early detection technology will provide emergency managers and first responders with round-the-clock monitoring, a sophisticated addition we are excited to add to our alert and warning toolkit,” Lynda Hopkins, chair of the Sonoma County Board of Supervisors, told the Associated Press.

County officials said the system would be used to alert first responders and emergency managers to potential wildfires via text or email.

About the author: George Leopold

George Leopold has written about science and technology for more than 30 years, focusing on electronics and aerospace technology. He previously served as executive editor of Electronic Engineering Times. Leopold is the author of "Calculated Risk: The Supersonic Life and Times of Gus Grissom" (Purdue University Press, 2016).

Add a Comment

EnterpriseAI