Search
Menu

Surveillance Cameras Serve on the Front Line Against Wildfires

Facebook X LinkedIn Email
Manned lookout towers, once an icon in the fight against wildfires, are vanishing. At one time, California had 625 fire towers, according to the Forest Fire Lookout Association. Today, 50 or fewer are staffed, mostly with volunteers. Although some people want to rebuild the fire tower program, others have advocated for a different approach — one that uses the latest vision technology.

Cameras placed on new or existing infrastructure (left) help to spot wildfires and provide information that is used to combat them. Courtesy of Axis Communications.

 
  Cameras placed on new or existing infrastructure (left) help to spot wildfires and provide information that is used to combat them. Courtesy of Axis Communications.

The economic cost of wildfires varies, but recent estimates have put the price tag for California alone at $10 billion or more annually. As part of the state’s response, CAL FIRE, the agency charged with combating the problem, is funding 172 high-definition cameras within the ALERTCalifornia network, which is part of the six-state, 800-plus camera effort called ALERTWildfire. The money for the consortium comes from a variety of sources, including government agencies, private utilities, and companies, as well as grants and public contributions.

Located in remote areas, the cameras provide real-time images. Upon command, they pan, tilt, and zoom to better see areas of interest.

“We utilize the system for enhanced situational awareness and validation of an incident,” said Phillip SeLegue, CAL FIRE’s deputy chief of intel. “The cameras provide much-needed immediate intelligence to assist in determining augmentation of a wildland fire dispatch and location of the incident.”

A person reporting a wildfire often does not know exactly where it is located. CAL FIRE will dispatch trucks, crews, and aerial support, as determined by conditions, to the general location. Operations center staff members use multiple cameras to determine the fire’s location in cases where the exact location is unknown. The staff feeds this information, along with any other pertinent findings gleaned from the cameras, to the incident commander so that the right resources can get to the correct spot as quickly as possible, potentially saving lives and property.

Low-light sensitivity

ALERTWildfire places cameras on existing structures — such as fire towers, cell towers, and private communications infrastructure — whenever possible and builds new platforms as needed. Craig Binford — regional sales manager for Northern California at Axis Communications, which supplies the cameras — said the company’s Lightfinder technology enables the cameras’ unique capabilities.

The cameras were tested as part of an ALERTWildfire evaluation. “Our camera got down low enough in light detection to be able to see the flames inside the smoke, so that they could even spot a fire at nighttime,” Binford said.

The Thomas Fire burns in the hills above Los Padres National Forest in December 2017. The fire had burned about 273,000 acres and was 65% contained. Today, vision technology helps to combat wildfires. Courtesy of Stuart Palley/Forest Service.

 
  The Thomas Fire burns in the hills above Los Padres National Forest in December 2017. The fire had burned about 273,000 acres and was 65% contained. Today, vision technology helps to combat wildfires. Courtesy of Stuart Palley/Forest Service.


The cameras’ Lightfinder technology consists of quality optical components, a sensitive progressive-scan CMOS image sensor, and a system on a chip with embedded signal processing and a machine learning processing unit. In the latest camera models, this combination of optimized components, software, and algorithms provides color images down to 0.1 lux and black-and-white images down to 0.002 lux — significantly less illumination than provided by a quarter moon at night.

The image processing in the camera itself removes noise, recovers colors, and clarifies images. Minimal temporal filtering can also produce frozen frames with little blur.

The Lightfinder technique works best on black-and-white images, such as those used for wildfire detection, but the technology also can deliver high-resolution, full-color video in near total darkness, Binford said. The cameras also know where they point, which is vital when trying to use two or more to pin down a fire’s location.

While the current approach and technology work to spot fires, automated detection would be an improvement. Utilities, government agencies, and researchers are investigating ways to automate detection using AI and machine learning. The goal is to deploy a machine learning-based classifier to detect a tiny puff of smoke against a backdrop of trees, hills, or brush, which could indicate the start of a fire.

Axis Communications is not involved in the AI investigations, but the company has been increasing its cameras’ computing horsepower and deploying new models with more advanced technology, Binford said. Some of its cameras are already running edge AI for tasks such as license plate recognition and object analytics. Thus, the cameras may already be capable of running a smoke detection classifier, and, if not, they may be able to do so soon.

In the future, users may be able to combine the cameras’ visual information and the resulting analytics with other data. They could add lidar-derived rainfall estimates, wind speeds, humidity levels, and other meteorological readings, for instance — thereby enabling firefighters to estimate which direction a fire may take.

Automated vision-based classification also brings another benefit: It eliminates human fatigue.

People looking at a screen can only do so for a short time before their attention starts to drift. An automated system does not get tired or bored, making it ideally suited to a job that never ends.

Pacific Gas & Electric Co. has installed hundreds of its own cameras as part of ALERTCalifornia. The California utility has conducted trials with several software firms, putting AI to the test. The results have not been publicly released.

CAL FIRE’s Research Development and Innovation Program has investigated the use of automated detection technology as well, as part of an evaluation of several companies and technologies, SeLegue said. The results are promising, but the technology still needs work before it can be relied upon to dispatch trucks and firefighters or to lessen the burden on skilled personnel in an operations center.

“Although the automated detection has improved, there remain false positives in the report of smoke,” SeLegue said.

Published: August 2022
Glossary
machine vision
Machine vision, also known as computer vision or computer sight, refers to the technology that enables machines, typically computers, to interpret and understand visual information from the world, much like the human visual system. It involves the development and application of algorithms and systems that allow machines to acquire, process, analyze, and make decisions based on visual data. Key aspects of machine vision include: Image acquisition: Machine vision systems use various...
artificial intelligence
The ability of a machine to perform certain complex functions normally associated with human intelligence, such as judgment, pattern recognition, understanding, learning, planning, and problem solving.
Vision in Actionvision-based classificationmachine visionAIartificial intelligenceclassifier

We use cookies to improve user experience and analyze our website traffic as stated in our Privacy Policy. By using this website, you agree to the use of cookies unless you have disabled them.