Opening the book…
We have seen how electric utilities are currently using and have future plans for using artificial intelligence in a variety of applications. One of the most pressing applications is for wildfire detection and management.
On a smaller scale, some communities in areas prone to wildfires have begun to investigate making use of wildfire detection equipment, some of which makes use of artificial intelligence.
For example, The Mercury News published a story from Phil Diehl of The San Diego Union-Tribune that described how the city of Carlsbad, California is looking to deploy cameras whose feeds would be analyzed by artificial intelligence to detect potential wildfires and notify the appropriate authorities when a fire is detected. [1]
The University of California, San Diego, one of the partners cited by the large California utilities in the previous sections, developed the wildfire monitoring system called AlertCalifornia in 2023. On its website[3], one can view feeds from a multitude of cameras across the state.
"ALERTCalifornia has more than 1,144 high-definition, pan-tilt-zoom cameras deployed across California (as of January 2025), providing a 24-hour backcountry network with near-infrared night vision to monitor disasters such as active wildfires. ALERTCalifornia cameras can perform 360-degree sweeps approximately every two minutes and can view as far as 60 miles on a clear day and 120 miles on a clear night."[3] The cameras use Forward Looking Infrared (FLIR) technology.[3]
“During its first season, the program detected more than 1,200 fires across the state, beating 911 calls to report the fires in 30% of the incidents, the AlertCalifornia website states.”[1]
ALERTCalifornia is a partnership between UCSD, the wireless internet company DigitalPath and CalFire. It makes us of artificial intelligence to analyze camera feeds and notify emergency responders when fires are detected.[1]
"A machine-learned model has been created to accurately differentiate between smoke and other particles in the air."[1]
Use of Artificial Intelligence
The ALERTCalifornia website explains its use of Artificial Intelligence as follows:
"ALERTCalifornia collaborated with the California Department of Forestry and Fire Protection (CAL FIRE) and industry partner Digital Path to create a fire detection AI tool with the goal of improving firefighting capabilities and response times. The value of this public-private partnership is the development of AI to aid firefighters, mitigate watchstander fatigue, reduce false positives, and confirm fire incidents in the incipient phase.
When the AI spots a potential fire on ALERTCalifornia’s network of cameras, the system alerts firefighters and provides a percentage of certainty and estimated location for the incident. If the incident is vetted and confirmed by trained watchstanders, firefighters respond quickly to extinguish the fire at the incipient phase. The camera network is also used to provide actionable real-time data to quickly scale fire resources, help evacuations through enhanced situational awareness, and monitor fire behavior.
Early detection and rapid response allow firefighters to combat fires before they grow. The AI tool became available to all 21 CAL FIRE 911 Dispatch Centers in September 2023. This new tool is especially effective in spotting anomalies in remote locations and is proven effective at night, even alerting firefighters before 911 calls."[3]
CalFire has invested around $24 million in the program.[1]
A second The Mercury News article “AI-boosted smoke-sniffing wildfire sensors approved for Santa Clara County” is of interest for a few reasons. De Anza College is located in Santa Clara County. Additionally, the article makes brief mention of how Artificial Intelligence is used.
“Made in the U.S. by Maryland company N5, the solar-powered devices detect gases and particulates that smoke produces. Artificial intelligence 'machine learning' software analyzes the detections to determine the likelihood that a wildfire is present.”[2]
“When a possible fire is identified, a device transmits a warning to local fire authorities, using whatever cell service has the strongest signal. If the chemical signatures of fire are strong enough to cross a threshold, the message to authorities is an alert, denoting a higher priority for response. If cell service is unavailable, as can occur when fires damage infrastructure, a backup Starlink satellite link is used.”[2]
Each N5 sensor costs about $5,000 to $7,000, with pricing dependent on whether a wind meter is included, and the status of tariffs.[2]
We will look at the N5 sensor offering in the next section.
A third The Mercury News article[4] updated the number of camera in use to 1,211 and mentioned the cameras were using artificial intelligence software as of 2023 developed by DigitalPath. "To address privacy concerns, UC San Diego blurs homes, parking lots and other nearby features that could track people in real time."[4] The cameras "send a message to dispatch centers across the state, which is then verified by humans."[4] In 2024 "there were 7,553 wildfires in Cal Fire’s jurisdiction. Of those, 1,668 were picked up by the cameras...and 38%, or 636, were detected by the cameras before any person called 911 to report them."[4]
DigitalPath
The DigitalPath-developed software is powered by AI trained on NVIDIA GPUs. "DigitalPath has refined a convolutional neural network to spot signs of fire in real time."[5] "The team began by training a convolutional neural network on a cloud-based system running an NVIDIA A100 Tensor Core GPU and later transitioned to a system running on eight A100 GPUs." [5]
A convolutional neural network (CNN) is a type of feedforward neural network that learns features via filter (or kernel) optimization. This type of deep learning network has been applied to process and make predictions from many different types of data including text, images and audio. Convolution-based networks are the de-facto standard in deep learning-based approaches to computer vision and image processing.[6]
"One of DigitalPath’s breakthroughs is its system’s ability to identify the same fire captured from diverse camera angles. DigitalPath’s system efficiently filters imagery down to a human-digestible level. The system filters 8 million daily images down to just 100 alerts, or 1.25 thousandths of one percent of total images captured."[5]
Architectural Considerations
The wildfire cameras and sensors need to be placed where they can be operationally effective. Cameras are often placed at mountain-top observatories, buildings, and stand-alone towers with good line of sight. They also need sources of power that might be available at these structures, or may require solar power with battery back-up if they are located in more remote locations.
An important consideration is where the bulk of processing is going to occur. If the bulk of the processing is going to occur close to the point of data collection, the computational power and electrical power budgets are going to be high, but the amount of data to be transferred to a central facility will be reduced. If the bulk of the processing is going to occur at a central processing facility, the local computational and power budgets are going to be lighter, but the data transmission requirements will be higher. In remote locations, having a reliable data transmission mechanism might be difficult, and likely would require consideration of having a back-up communications mechanism (e.g., a satellite connection).
Another consideration is when an alert should be sent to the appropriate authorities and by whom. Should the sensor or camera send that alert? Should the alert be sent after processing in a more centralized processing center, perhaps with human oversight?
In the next section, we will take a look at some commercially available products and see what design choices they have made.
References
- Diehl, Phil. “City Looks into camera that detect wildfires”, The Mercury News, Tuesday July 29, 2025, Section B, Page 5. Note: Phil Diehl is at The San Diego Union-Tribune.
- Baron, Ethan. “AI-boosted smoke-sniffing wildfire sensors approved for Santa Clara County”. The Mercury News. May 23, 2025.
- ALERTCalifornia. ALERTCalifornia.
- Rogers, Paul. The Mercury News (online version), September 19, 2025. AI cameras spot wildfires across California, often before humans call 911
- Caufield, Brian. How AI Helps Fight Wildfires in California. October 4, 2023. How AI Helps Fight Wildfires in California | NVIDIA Blog
- Convolutional neural network - Wikipedia. Text is available under the Creative Commons Attribution-ShareAlike 4.0 License.
7.5: Applying Artificial Intelligence to Wildfire Detection and Management is shared under a CC BY 4.0 license and was authored, remixed, and/or curated by LibreTexts.
The book could not be loaded just now.