The Challenge
The community of Austin, Texas, lies at the edge of a wildland–urban interface, where fast-growing suburbs meet the fire-prone brush of the Hill Country. Local stakeholders recognized the need for a long-range wildfire monitoring system to protect the area, but faced a unique set of challenges. No dedicated fire-surveillance infrastructure existed in the region, nor were there resources to establish a dedicated dispatch center. Moreover, the system had to be operable by fire-emergency services and accessible to the public at the same time, as part of the community outreach initiative proposed by the stakeholders. It needed to cover a wide geographic area, detect smoke instantly, and alert authorities without delay – all while keeping false alarms to an absolute minimum. Such project requirements emphasized the need of a fully remote solution as well as tailor-made modifications to our existing software that we were happy to implement to provide early warning about wildfires in an area to a wider audience.
The Solutions
SmokeD provided a turnkey AI-powered wildfire detection setup by utilizing highly elevated locations including commercial buildings, strategically located residential buildings and telecom tower mast. To maximize reliability, the design uses redundant detection devices: pan-tilt-zoom (PTZ) cameras, complemented by additional fixed SmokeD DuoVue and V1.0 detectors. This dual system means the cameras can cross-verify any potential smoke sighting – if a detector catches on smoke, by utilizing PTZ cameras we can verify the fire and provide live situational awareness and other way around if a smoke was caught by PTZ a detector can be used to estimate fire location. Every installation was configured for remote access and diagnostics over the Internet. This allows engineers to monitor system health, adjust settings, or troubleshoot issues without needing to access the sites, an essential feature given level of criticality of such system. In addition, entire detection system is cloud-managed: there is no need for a local monitoring room on-site, as the SmokeD platform and user interface are hosted in the cloud and accessible via web browser. Live video feeds from all cameras stream in real time at Full HD resolution to SmokeD’s cloud, where AI analysis runs 24/7 on the imagery. To handle alerts, SmokeD implemented a multilayered workflow: whenever the AI detects a possible smoke plume, an expert at SmokeD’s central Data Analysis Center immediately reviews the images. Only verified fire alerts are then pushed out as notifications to users’ phones, complete with the location of the threat and a snapshot of the scene. Local fire responders and authorized personnel use the SmokeD mobile app to receive these alerts and can also view the live camera streams on-demand for situational awareness. By leveraging an existing infrastructure, cloud technology, and double confirmation, the Austin Wildfire Detection project delivered a robust wildfire detection system finely tuned for remote management and high reliability.
The Effects
Since its deployment, the Austin Wildfire Detection system has provided continuous, worry-free wildfire surveillance over the area. Establishing double-check system eliminated false alarms significantly. This means that when an alert does hit the fire department’s app, responders can trust that it’s real and can act immediately, rather than wasting time on glitches or false sightings. The remote diagnostics capability has kept the system running smoothly; most issues can be identified and resolved off-site, greatly reducing the need for on-site maintenance visits. The community now benefits from a cutting-edge early warning network: residents and officials receive prompt, verified notifications of fires in their vicinity, often within minutes of ignition. They can also tap into live high-definition video from the wildfire cameras at any time, giving them peace of mind and real-time insight during high fire danger days. Overall, the Austin Wildfire Detetion project demonstrates how a cloud-based, redundant AI detection system can overcome access challenges and protect a growing community at the wildland’s edge with minimal oversight and maximum confidence.
The Challenge
The Chmielnik Forest District in south-central Poland (Świętokrzyskie region) needed to modernize and expand its aging wildfire lookout network to guard a vast woodland area. The district encompasses roughly 335,000 acres of forest, including remote and rugged terrain where fire outbreaks could go undetected for too long. Historically, only two manual lookout towers were in place, leaving parts of the district vulnerable. After a series of regional fire incidents, the forestry management recognized the need for a comprehensive, automated detection system covering the entire district. The challenge was significant: it required building new fire towers in hard-to-access locations, upgrading old towers with modern equipment, and linking all sites to a central monitoring hub. The system had to provide real-time surveillance over hundreds of thousands of acres, withstand power outages and harsh weather, and allow technicians to troubleshoot issues remotely due to the difficult access to tower sites. In short, Chmielnik needed a state-of-the-art wildfire detection network with full coverage, high reliability, and integration into their forestry operations.
The Solutions
To address these needs, SmokeD and the Chmielnik Forest District implemented a sweeping upgrade of the fire monitoring infrastructure. Three new fire watch towers were constructed using durable pre-tensioned concrete, strategically sited to fill coverage gaps, and the two existing towers were thoroughly modernized. Each of the five tower sites was outfitted with a SmokeD long-range PTZ camera for 360° wildfire observation, as well as a secondary fixed CCTV camera acting as a backup and providing security (anti-theft monitoring) for the site. All video feeds from these towers are transmitted back to the district’s headquarters via a network of four wireless radio links in the 10.5 GHz band, ensuring a high-bandwidth, low-latency connection even across remote forest terrain. The video transmission is designed to be lossless and real-time, so operators see exactly what the field cameras see with no delay. To guarantee around-the-clock operation, each tower was equipped with an uninterruptible power supply (UPS) and backup batteries, providing at least 24 hours of power in case of an outage. The SmokeD automatic smoke detection system (AI software running on the camera feeds) was deployed on all towers, analyzing the video 24/7 to spot any sign of smoke or fire within seconds of its appearance. At the central monitoring station in Chmielnik, a new control room was set up with five large display screens – one for each tower’s live feed – and an integrated console to control any of the PTZ cameras as needed. The monitoring center’s computer systems run specialized SmokeD software featuring interactive forest maps and real-time visualization of camera coverage, which helps operators and forest guards orient themselves quickly when an alert comes in. Another key feature of the solution is its multi-level alert verification: whenever the AI detects smoke on a camera, an alert is instantly sent to both the local monitoring center and to SmokeD’s off-site Data Analysis Center, where a staff member double-checks the images. Confirmed alarms are then forwarded to on-duty forestry officers’ phones and the central command screen, complete with the fire location and images, ensuring that only credible, verified alerts trigger an emergency response. This tiered approach greatly reduces false alarms and builds trust in the system’s warnings.
The Effects
The Chmielnik Forest District now operates one of the most advanced wildfire detection networks in the country, with full 24/7 coverage of over 1.5 million acres of forests and surrounding land. This expanded network of automated towers has dramatically improved early detection capabilities even a small smoke plume in a remote corner of the district will be spotted within minutes by the AI cameras, prompting an immediate alert to rangers. By upgrading and adding towers, the previous blind spots have been eliminated, meaning fires can no longer start unnoticed in those areas. The use of multiple interconnected towers and backup systems also provides strong redundancy: if one tower’s equipment goes offline due to maintenance or an incident, the remaining cameras can often cover its area until it’s restored, and the UPS units keep the detectors running through power failures.
The new central command center allows for streamlined coordination; operators can assess a situation at a glance with real-time feeds and maps, then dispatch the nearest fire response units with precise location data.
Thanks to the verified alert workflow, the forest district staff have confidence that when the alarm sounds, it’s a real fire – this has reduced costly false callouts and ensures that firefighting resources are mobilized only when truly needed. Overall, the modernization and expansion project in Chmielnik has not only increased wildfire detection speed and accuracy, but also integrated the system smoothly into daily forestry operations. It serves as a model for how legacy fire lookout infrastructure can be transformed with AI detection, robust communications, and smart power backups to protect natural resources and communities from devastating wildfires.
The Challenge
As utility-scale solar farms continue to expand across the northwestern United States, ensuring their safe and uninterrupted operation becomes a top priority. One such site – spanning an area larger than 1,200 football fields – is located in a semi-arid environment surrounded by natural shrubland. While the vegetation here is adapted to drought, it’s also highly flammable. Sagebrush, rabbitbrush, and native bunchgrasses create a landscape that, despite active fuel management, remains at risk of wildfire. In this kind of setting, early detection is key to protecting infrastructure and maintaining operational continuity.
The Solutions
To manage this risk, SmokeD provided a complete wildfire detection solution, including the PTZ camera, DuoVue detectors, and the Alarm and Dispatch Center – all fully integrated and tailored to the site. Each surveillance point is backed by an uninterruptible power supply (UPS), providing more than five hours of autonomous operation in the event of a power outage. The two observation posts are also connected via a radio link, ensuring continuous internet transmission between them – even in remote terrain. Together, these components enable 24/7 monitoring and rapid alerts in case of smoke or fire, giving operators the critical lead time needed to respond effectively and prevent serious damage.
The Effects
While the system was installed only recently, its presence has already increased confidence among site operators by adding a critical layer of readiness. The site is now continuously monitored with automated smoke detection and real-time alerting — capabilities that did not exist before. Even though no fire threats have occurred since deployment, the infrastructure is now in place to respond immediately, reducing the risk of damage and downtime. As the fire season progresses, the system stands as a key component in the site’s broader risk mitigation strategy.
Context and Objectives
Albania and North Macedonia both face significant risks from wildfires, particularly during the summer months when dry conditions and high temperatures are common. Both countries contain large forested areas, some of which are located in remote and mountainous regions, making early fire detection and rapid response a challenge. Climate change is exacerbating these risks, as temperatures continue to rise, leading to more frequent and severe wildfire events. To address these concerns, both countries have sought to modernize their wildfire management systems by deploying advanced monitoring and early detection technologies.
The SmokeD system was selected for this purpose due to its capabilities in early fire detection using AI-powered wildfire detectors combined with pan-tilt-zoom (PTZ) cameras. The goal of the deployment was to enhance the countries’ ability to monitor remote forested areas, detect wildfires in their early stages, and provide timely alerts to firefighting teams and local authorities to improve response times.
This case study describes the deployment and implementation of the SmokeD wildfire detection system in Albania and North Macedonia, which took place between August 2022 (first phase) and October 2023 (second phase). The project aimed to improve wildfire surveillance, strengthen local fire management capabilities, and ensure better response through automated alerts.
Deployment Overview
Albania and North Macedonia Deployment Details
The deployment covered 4 telecom towers located in strategic, fire-prone regions across both Albania and North Macedonia. The objective was to install a camera-based system capable of providing 360-degree coverage of each site. The system installation included:
The system design ensures that the cameras and detectors are fully integrated with the monitoring centers, which receive real-time images and alerts. The information is made available through SmokeD Web and the SmokeD Alerts mobile app, allowing for remote monitoring and immediate action when a potential fire is detected.
Challenges Encountered
One of the primary challenges during the installation was the logistical difficulty of accessing remote sites. The telecom towers were located in high, rugged terrain, which required manual transport of equipment, including cameras, solar panels, and batteries. This was physically demanding, and the project team had to work around weather challenges such as strong winds and variable weather conditions.
In addition to the physical installation challenges, the systems were designed to operate autonomously, which required careful integration of solar power systems to ensure reliability and uninterrupted power supply (UPS) at each site. These measures were critical to maintain system functionality in remote areas where power outages could otherwise disrupt operations.
System Use and Early Observations
Since installation, the SmokeD system has been actively monitoring the forested areas around the four tower locations. The system operates in 24/7 mode, with the detectors analyzing incoming video footage for signs of smoke or fire. Once a potential fire is detected, the system automatically triggers alerts, sending notifications with the exact location and a snapshot of the suspected fire to the monitoring center. Additionally, personnel at the monitoring centers can access the system’s live video feed via SmokeD Web and review alerts through the SmokeD Alerts mobile app.
Initial observations suggest that the system has successfully detected several wildfire incidents since deployment. The real-time alerts generated by the system have allowed local firefighting teams to respond quickly, preventing larger-scale fires. The combination of fixed detectors (for continuous scanning) and PTZ cameras (for zooming in on specific areas) has proven effective for monitoring expansive, often hard-to-reach landscapes.
At the dispatch centers, the system has enhanced the operational efficiency of fire management personnel. The combination of automated alerts, real-time camera views, and geo-location data helps operators respond with a clear understanding of where fires have occurred and the scale of the threat.
User Training and Maintenance
As part of the project, training sessions were conducted for fire management teams on how to use the SmokeD system. This included training on the software interfaces, how to interpret alerts, and how to manage emergency responses. Additionally, the SmokeD team has provided ongoing system maintenance and technical support, ensuring that the system remains operational and any potential issues are addressed promptly.
Conclusion
The SmokeD wildfire detection system has been deployed in Albania and North Macedonia to enhance wildfire management capabilities.
By equipping four strategically located towers with advanced camera-based monitoring and AI-driven wildfire detection, the project has enhanced the ability to detect fires early and respond in a timely manner.
Despite challenges related to the terrain and weather conditions, the project has been successfully completed with the system operating reliably across both countries. Initial feedback indicates that the system is functioning as intended, detecting fires early and improving coordination between firefighting teams. Both countries now have a modernized surveillance system in place, with the potential to improve wildfire response in the coming years.
The integration of solar power, real-time alert systems, and remote monitoring tools has proven crucial for maintaining system uptime and ensuring continuous coverage, even in the most remote regions. The project’s impact is expected to grow over time as more data is collected and analyzed, contributing to better wildfire prevention and management strategies in the future.
Key Takeaways:
This project is a clear example of how innovative technologies like SmokeD can be leveraged to improve wildfire detection and response, helping to protect lives, property, and the environment.
The detectors use AI algorithms to detect signs of smoke and flame, sending real-time alerts to local authorities.
The communication infrastructure for the deployment was provided by Firebreak, using LTE SIM cards to ensure reliable data transmission between the detectors and the monitoring stations.
Results
Increased detection speed: the SmokeD system successfully detected signs of smoke in the pilot areas during the first few months of operation. The system’s AI-powered smoke detection capabilities allowed for rapid identification of potential fire sources, giving local firefighting teams a head start in responding to the fires.
Effective use of surveillance technology: the detectors offered detailed, almost real-time monitoring of high-risk areas. SmokeD’s panoramic coverage ensured that large sections of the terrain were monitored continuously, with minimal reliance on human observation.
Positive feedback from local fire agencies: local firefighting units and government agencies expressed strong interest in the system’s capabilities. The integration of real-time alerts and automated detection provided firefighters with accurate data, allowing them to respond more effectively. Firebreak, as the local partner, also reported that the system performed well, meeting the project’s expectations.
Scalable and future-ready solution: SmokeD’s system is designed to be scalable, allowing for future expansion. The detectors and cameras can be added to additional areas in the region as the need arises. This is crucial for addressing the ongoing wildfire threat in the south of France.
Conclusion
The SmokeD Wildfire Detection System deployment near Nice in June 2023 has been a successful demonstration of its capabilities in wildfire-prone regions. The system provided real-time detection and automated alerts, enabling local firefighting teams to respond faster to potential fires. By partnering with Firebreak, SmokeD has introduced a modern, AI-powered solution to help protect fire-prone areas in southern France.
This pilot installation has demonstrated the potential of SmokeD as a scalable solution for wildfire detection, and has generated interest among local agencies in expanding its use for future prevention and monitoring.
The Challenge
A large photovoltaic (PV) farm in Europe faced the critical challenge of fire risk management. Solar farms are vulnerable to fires caused by electrical faults, hot spots, or external wildfires, yet such incidents often go undetected until significant damage has already occurred. The remote, wide-area layout of the solar installation made continuous human monitoring impractical. Any uncontrolled fire could lead to severe financial losses, prolonged operational downtime, and safety hazards for the renewable energy facility. The client required an early-warning solution that could detect the first signs of smoke or fire around the solar panels – enabling intervention before a minor spark turns into a significant incident. Ensuring 24/7 surveillance of the expansive site, with rapid alerting capabilities, was paramount to protect the investment and maintain uninterrupted power generation.
The Solution
To address these risks, the project was executed in two phases: an initial on-site deployment of the SmokeD early fire detection system, followed by integration of that system into the client’s own operations platform. This phased approach allowed immediate coverage of the fire risk, and later streamlined how alerts were handled by the operations team.
Phase 1: On-Site Deployment (May 2025)
In May 2025, the client deployed a SmokeD Early Wildfire Detection System on the solar farm premises. This included one Manta FPS61 HD PTZ camera and three SmokeD DuoVue detectors, strategically mounted to achieve maximum coverage of the PV installation. Together, these devices provide comprehensive 360° visual surveillance and automated smoke detection across the entire facility. The three fixed DuoVue detectors (each equipped with optical and infrared imaging and on-board processing) continuously scan the horizon in all directions, covering the solar farm and its surroundings within a radius of up to 15 kilometers. The Manta PTZ camera, a high-definition pan-tilt-zoom unit, complements the detectors by automatically zooming in (up to 30× optical zoom) to verify potential smoke sightings and by patrolling the area when no alarms are present. This hardware configuration ensures there are no blind spots in monitoring, and even distant smoke plumes can be detected early.
Key capabilities of the deployed SmokeD system included:
By the end of Phase 1, the solar farm had a fully operational, automated early warning network for fires. The detectors and PTZ camera worked in tandem to surveil the solar arrays and surrounding environment in real time, providing the client’s safety team with immediate alerts of any anomalous smoke. This deployment significantly reduced the likelihood that a fire could start and spread undetected on the premises.
Phase 2: Platform Integration (September 2025)
The second phase focused on integrating the SmokeD system’s outputs into the client’s existing operations platform. Initially, in Phase 1, alerts and camera feeds were accessed through SmokeD’s own interfaces (a web dashboard and mobile application). While effective, the client sought to streamline user experience by incorporating fire detection monitoring into their central control system, the same system used daily for supervising power production and site operations.
In Phase 2, the SmokeD platform was tightly integrated with the client’s operations software. Practically, this meant that any smoke or fire alert triggered by the SmokeD detectors would pop up within the client’s regular operational console, accompanied by the live video feed from the PTZ camera. Operators no longer needed to switch to a separate app; all fire surveillance data is now available in one unified dashboard SmokeD’s API and customizable web interface were used to integrate the system, and the SmokeD team embedded camera view directly into the client’s dashboard. Permissions and alarm settings were configured so that the right personnel on the client’s team received the alerts through their normal notification channels (on-screen alerts, emails, etc.), just as they did for other operational events.
This seamless integration simplified usage and response: the solar farm staff could acknowledge alarms and view incident footage in the same workflow used for their other monitoring tasks. Training requirements were minimal, as the alerts appeared in a familiar format. By the end of Phase 2, the early fire detection system was not just an add-on tool but a fully incorporated element of the site’s operations management.
The Effects
The two-phase deployment of the SmokeD system has markedly strengthened fire safety at the PV farm. Fire and smoke detection is now automated and immediate, giving the operators precious early warning to take action. Any incipient fire in or around the solar plant is likely to be detected within minutes of ignition, with the system typically identifying smoke well under the 10-minute mark from its first appearance. This level of responsiveness means that even a small spark or smoldering event can be addressed by maintenance crews or emergency responders before it escalates, dramatically reducing the risk of severe damage. In a domain where a single unchecked fire could destroy thousands of solar panels and halt generation, the value of this early warning capability is enormous.
Moreover, the comprehensive 360° coverage of the SmokeD detectors and PTZ camera ensures that the entire solar farm and its surrounding environment are under observation at all times. The farm is now protected not only against internally caused fires (such as electrical faults in panels or inverters) but also against external threats like grassfires or wildfires approaching the site. This wide-area surveillance provides ongoing protection for the critical infrastructure, which is especially important given the growing wildfire risks in many regions.
Another significant outcome is the improved operational efficiency due to the Phase 2 integration. By embedding SmokeD alerts into the existing operations platform, the client’s team can respond faster and with more confidence. When an alert is raised, operators see real-time camera visuals and location data on their primary screen, enabling them to immediately assess the situation and decide on next steps. This integration has eliminated the need for parallel systems and has streamlined the decision-making process during potential fire events. In daily practice, the staff reports that the fire detection system “blends into” their routine monitoring – meaning there’s no added complexity, just added safety.
Overall, the deployment has transformed the farm’s fire risk management from a reactive stance to a proactive, high-tech surveillance model. It demonstrates how AI-powered early detection can safeguard renewable energy assets: the solar farm remains continuously guarded against fire hazards with minimal human oversight required. The client now benefits from enhanced protection of their investment, reduced downtime risk, and reassurance that they are prepared for one of the worst-case scenarios in the industry. This case also provides a blueprint for other utility and renewable energy operators looking to bolster the resilience of their installations with automated wildfire detection systems – all while integrating smoothly with existing operational workflows.
The Challenge
The Calabria region in southern Italy faces a growing wildfire threat as climate change drives hotter, drier conditions. In recent years, wildfires have become more frequent and severe across the Mediterranean; for instance, a massive 2021 blaze in Calabria scorched over 15,000 hectares of forest and scrubland and tragically claimed five lives. Such extreme events, exacerbated by rising temperatures and prolonged summer droughts, underscore the urgent need for enhanced early-warning systems. The University of Calabria, situated in this high-risk Mediterranean environment, recognized that traditional fire surveillance methods were no longer sufficient to protect the region’s natural heritage and communities from fast-spreading wildfires.
Compounding the wildfire risk are Calabria’s challenging geography and logistics. The SmokeD system was to be installed at two strategic mountaintop sites – Monte Botte Donato and Monte Curcio – which offer commanding views over vast forested areas. Botte Donato, the highest peak in the Sila plateau at 1,928 m elevation, and Monte Curcio (~1,768 m, accessible by cable car) provide ideal vantage points for smoke monitoring. However, their remote, mountainous terrain posed significant challenges. Transporting towers, cameras, and other heavy equipment up winding alpine roads and ski lift infrastructure required careful planning and rugged vehicles. Moreover, the mountaintops are exposed to strong winds and harsh weather. During initial tests, high winds caused camera shake – a serious issue for an optical detection system. This meant the team had to design and install a dedicated camera-stabilizing pole and reinforced mounts to ensure the sensors and pan-tilt-zoom cameras remained steady and operational even in gusty conditions. In short, the University’s challenge was twofold: environmental, in addressing a rising wildfire threat in a changing climate, and technical, in deploying advanced detection hardware in a rough, high-altitude landscape.
The Solution
On request from the University of Calabria, SmokeD deployed a cutting-edge early wildfire detection system, tailored to the region’s needs, to modernize wildfire surveillance. At each of the two mountain installation points: Botte Donato and Monte Curcio – the project team mounted a Manta FPS61 HD pan-tilt-zoom (PTZ) camera paired with a cluster of SmokeD wildfire detectors. Each PTZ camera provides 360° visual coverage of the surrounding landscape, while the SmokeD detectors continuously scans all directions for the slightest hint of smoke. These smart AI-powered detectors can identify a rising smoke plume up to approximately 15 km away under good visibility conditions, giving the system a vast surveillance radius from each peak.
The equipment was securely installed on reinforced poles, including the special stabilizing mast on the windy Botte Donato summit, to minimize vibration and ensure clear imaging. The fiber internet connection with a static IP address is provided by the local service provider to transmit the real-time data down to the University’s monitoring center. Despite the logistic hurdles, the installation was completed successfully by May 2025, resulting in two high-elevation automated lookout points guarding the Calabrian forests day and night.
In parallel with the field installations, SmokeD team set up an indoor wildfire monitoring center at the University campus to operate and oversee the system. This monitoring station is equipped with two large wall-mounted TV displays that show live panoramic video feeds from the mountaintop PTZ cameras, and a computer workstation running the dedicated SmokeD Desktop application. Through this software, operators can remotely control each PTZ camera – panning across the forests, tilting, and zooming in up to 30× – to investigate any suspected smoke spotted by the detectors. The SmokeD Desktop console also presents incoming alerts on an interactive map, with real-time overlays that pinpoint the exact geolocation of any detected fire. In addition to the desktop setup, the University’s team has been given access to SmokeD’s web and mobile tools for flexible, distributed monitoring. SmokeD Web, an online dashboard, allows authorized users to manage the detectors and check system status from anywhere. Complementing it, the SmokeD Alerts Pro mobile app delivers instant push notifications to team members’ smartphones whenever the system’s AI algorithms identify potential smoke or flame. This means that even if staff are not in the control room, they will immediately receive an alert with the location and a snapshot of the event if a fire is detected. All alerts and live video feeds are also integrated into the University’s own systems – enabling, for example, live stream viewing and geo-tagged alarm data to be shared with researchers or regional emergency officials in real time. By combining advanced hardware on the peaks with a robust software ecosystem (desktop, web, and mobile), the solution provides the University of Calabria with a comprehensive wildfire detection network tightly integrated into their operational and research infrastructure.
The Effects
Though only recently commissioned (the system went live in May 2025), the SmokeD deployment at the University of Calabria has already proven its worth by catching several real wildfire ignitions in their early stages. In the weeks following installation, a few small fire incidents occurred in the region’s woodlands – and in each case the SmokeD DuoVue detectors spotted the telltale wisps of smoke within minutes, triggering an immediate alert. The moment the AI detection algorithms recognize a potential fire, the system automatically sends out an alarm with precise GPS coordinates of the site and visual evidence. At the monitoring center, the SmokeD Desktop application flashes a notification on the map, and simultaneously the on-call staff receive an alert on the SmokeD mobile app with a snapshot of the smoke plume. This early-warning workflow enabled the University’s team to promptly verify each incident via the live camera feed and notify local firefighting authorities to intervene. The system detects fires within 5-10 minutes of ignition, as designed. (The system is tuned to detect fires generally within ten minutes of smoke appearing, greatly increasing the probability that a fire can be contained while still small.) In one case, a grassfire near a rural village was detected by the SmokeD sensor long before any 911 calls were received. Thanks to the geo-localized alert sent to officials, fire crews were mobilized and arrived while the blaze was still small, preventing a disaster. This incident is illustrated in Figures 2 and 3, which show the fire as marked by the SmokeD system and its precise location on the map.
These initial successes illustrate how the SmokeD system is fundamentally improving wildfire preparedness: by providing rapid, reliable detection of nascent fires, it buys precious time for response and thereby helps limit damage to forests and communities.
Beyond the immediate practical benefits of faster detection and response, the system is yielding broader positive effects for research and environmental monitoring at the University. The continuous 24/7 surveillance provided by the mountaintop cameras and detectors closes critical gaps that existed in the previous monitoring approach. Rather than depending just on human patrols or towers far away, the area is now continuously monitored by an automated system – even during the night, holidays, or when visibility is poor. This around-the-clock monitoring fills the off-hour gaps when wildfires might have previously smoldered unnoticed. The data stream from SmokeD is also a valuable resource for the University’s scientists and students who study wildland fires, climate change, and land management. They can analyze the timings and locations of detections to better understand wildfire patterns, or even use the live camera feeds to observe fire behavior and smoke dispersion in real time as part of their research. Additionally, the University’s integration of SmokeD’s alert feed into its systems means that campus safety officers and environmental researchers alike have a new level of situational awareness regarding wildfire threats in the surrounding territory.
In summary, the early impact of the SmokeD System in the Calabria region goes beyond saving several forests through rapid intervention – it has also enhanced the region’s readiness and reinforced the University’s commitment to safeguarding the environment. The successful detections to date have demonstrated the system’s life-saving potential, validating the University’s decision to invest in high-tech wildfire surveillance for this Mediterranean region on the frontline of climate impacts.
The Challenge
Salt River Project (SRP), one of Arizona’s largest electric utilities, faced an escalating wildfire threat across its vast service territory. In the state’s arid climate, wildfires can ignite and spread rapidly, endangering high-voltage transmission lines and nearby communities. SRP’s remote power lines traverse rugged forests where early fire detection and real-time situational awareness are difficult with traditional patrols. The utility initially deployed SmokeD optical wildfire cameras to spot smoke, but it sought to enhance early detection and even anticipate fires by monitoring environmental conditions in real time. The overarching challenge was twofold: (1) detect wildfires near critical infrastructure as soon as they start, and (2) monitor the power line structures themselves for any physical hazards – all in remote, hard-to-access locations. Furthermore, SRP needed a solution that could integrate advanced weather sensors with the AI smoke detection on six towers, creating a unified system that would remain reliable under harsh field conditions. In summary, SRP required a proactive approach to wildfire management that not only provided immediate alerts of fires and equipment issues, but also forewarned of high-risk weather conditions that precede wildfire outbreaks.
The Solutions
Phase 1 (April 2023) – AI wildfire surveillance on transmission towers: SRP partnered with SmokeD in 2023 to pilot an AI-powered wildfire detection and grid monitoring system on six strategic transmission towers situated in the Tonto and Apache-Sitgreaves National Forests. Each tower was equipped with a SmokeD optical camera unit that continuously scans 360° for smoke plumes day and night, using AI-driven video analytics to distinguish wildfire smoke from other visuals (such as dust or cloud haze). The cameras can detect faint smoke up to ~ 15 kilometres (10 miles) away with a wide 82° field of view, completing a full panorama every few minutes. When the system’s machine-learning algorithms identify a potential fire, it immediately triggers an alert to SRP’s operators with precise GPS coordinates, distance, and a snapshot of the suspect smoke plume. This alert is delivered via a cloud-based dashboard and mobile app, allowing dispatchers to verify the imagery and pinpoint the location on a map within seconds. By providing automated, geolocated fire notifications 24/7, Phase 1 established a virtual wildfire lookout network that vastly improves on human spotters or routine aerial patrols in both speed and coverage. Notably, the SmokeD system “learns” the normal background of each camera’s environment over time, which helps it ignore known features (like steam from a mill or morning fog) and focus on truly anomalous smoke signatures, reducing false alarms.
In addition to wildfire surveillance, the Phase 1 deployment introduced a dual-purpose monitoring capability for SRP’s grid infrastructure. The high-resolution SmokeD cameras periodically capture images of the tower structure and power lines they are mounted on to perform remote asset inspections. The AI analyzes these images to identify potential hazards such as a leaning or damaged tower, a downed conductor, or overgrown vegetation encroaching on the lines. If any anomaly is detected, for example, a broken line or fallen tree after a storm – the system flags it and immediately notifies SRP maintenance teams with the location and images of the issue. This early warning for infrastructure problems enables crews to respond quickly and fix issues before they escalate into outages or fire ignition sources. All Phase 1 devices are networked through SRP’s secure communication backbone and are powered by reliable solar supplies, ensuring uninterrupted operation even on remote mountaintop towers. The camera feeds and alerts stream in real time over this network to SmokeD’s cloud servers, where advanced machine-learning algorithms process the data and continuously refine detection accuracy. SRP’s operators can access the live video and alarm interface through any web browser or smartphone, enabling full situational awareness from anywhere. By the end of Phase 1, SRP had a robust 24/7 automated wildfire detection and asset monitoring system of remote transmission corridors. This AI-driven network started providing SRP with immediate notifications of fires or equipment issues, dramatically improving the utility’s ability to safeguard its grid and the surrounding environment.
Phase 2 (July 2024) – integration of advanced weather stations: building on the success of the pilot, SRP and SmokeD launched Phase 2 in mid-2024 to add a crucial predictive dimension to the system. In July 2024, each of the same six towers was retrofitted with an array of smart weather sensors mounted alongside the SmokeD cameras. These compact weather stations continuously measure temperature, humidity, wind speed, wind direction, rainfall, and even fuel moisture (the dryness of vegetation) in real time. The environmental sensors feed their data into the existing tower communication links, streaming live readings to the SmokeD cloud platform over SRP’s network. The SmokeD software was enhanced to synchronize meteorological data with the video feeds: if a camera detects smoke, the system automatically displays the local weather conditions (e.g. current heat, wind, and humidity levels) at that tower on the same dashboard view. This means operators can instantly correlate what the camera sees with critical weather variables on-site, for instance, knowing that a detected smoke plume is occurring during red-flag (hot, dry, windy) conditions. Phase 2 effectively unified AI smoke detection with on-the-ground weather intelligence, turning the system into a comprehensive early-warning network. The upgraded interface allowed SRP’s team to remotely monitor both live video and weather data 24/7 from any web browser or mobile device, with automated alerts not only for smoke but also for any abnormal weather readings that could signal elevated fire risk. Importantly, this enhanced solution was deployed without adding new communications infrastructure – it leveraged the existing tower network and SmokeD cloud, demonstrating how additional sensor technology can be seamlessly layered onto an AI camera system. By integrating advanced weather stations into the SmokeD platform, SRP gained the ability to not just react to wildfires, but to anticipate them based on real-time local climate conditions at the towers. This fulfills SRP’s goal of a predictive wildfire management approach, all while using the same six sites as a backbone.
The Effects
Faster detection and response: with the two-phase SmokeD system in place, SRP now has a proactive wildfire defense that detects fires faster and coordinates response more effectively than ever before. Smoke from a wildfire near SRP’s lines, as captured by a SmokeD camera soon after the system went live. In one early example, shortly after the Phase 1 cameras were installed, one of the SmokeD units spotted smoke from a prescribed burn roughly 4 miles away and automatically alerted fire officials within 90 minutes of ignition. In many cases the AI system has proven even faster – often recognizing a new smoke plume within minutes of a wildfire’s start, long before any 911 calls or lookout reports. Every minute gained is critical. These earlier warnings have given SRP’s emergency coordinators and local fire agencies a head start, enabling them to attack emerging fires sooner and contain them while still small. In practice, the SmokeD alerts (which include exact GPS coordinates of the fire) allow fire crews to be dispatched to the precise location without wasting time on scouting, significantly reducing response times compared to previous protocols. SRP reports that any sign of smoke now triggers an immediate alarm, and this speed, combined with the system’s accuracy, has tangibly improved wildfire response outcomes in the covered areas.
Improved forecasting and situational awareness: the addition of real-time weather monitoring in Phase 2 has greatly sharpened SRP’s predictive capabilities and situational awareness for wildfire management. By continuously tracking local temperature, wind, and humidity at the tower sites, SRP’s team can anticipate high-risk fire conditions before a fire even ignites. For instance, if the sensors detect a sudden drop in humidity and gusty winds during a heat wave, SRP can proactively put crews on standby or adjust line operations, knowing the likelihood of a wildfire is spiking. This data fusion of weather and visual intel also means that when a SmokeD camera does catch smoke, operators immediately know the context, such as which direction the wind is blowing the fire and how fast, or how dry the vegetation is, which helps in predicting the fire’s behavior and spread. As a result, SRP and local firefighters can make more informed decisions on resource deployment: they can allocate firefighting crews, aerial support, and outage repair teams more efficiently, targeting the areas of greatest need with better intel on the fire’s potential growth. The unified SmokeD dashboard provides emergency managers with a complete live picture of wildfire activity and weather at once, all on a single screen. This level of situational awareness, knowing when and where a fire is likely to erupt, and seeing it in real time with on-site weather context, is a game-changer for SRP’s wildfire preparedness. It has improved prediction accuracy for wildfires and given the utility a more predictive, rather than purely reactive, stance in the face of fast-changing conditions.
Enhanced infrastructure safety and reliability: beyond fire detection, the project’s dual monitoring approach has measurably strengthened grid resilience for SRP. The SmokeD system’s ability to catch equipment problems on the transmission lines has led to earlier interventions for grid issues that could have caused outages or even sparked fires. For example, if a windstorm damages a distant transmission tower or knocks down a line, SRP operators no longer have to wait for an outage alarm or a field patrol to discover it, the nearest SmokeD camera will typically spot the change and send a visual alert within minutes, even if the area is 200 miles from the operations center. This rapid awareness allows maintenance crews to be dispatched immediately to the exact location of the trouble. By mitigating power line faults before they escalate, SRP reduces the risk of those faults igniting a wildfire and improves overall service continuity. In practice, SRP field crews now have a powerful tool for rapid infrastructure inspection after events like lightning strikes or high winds – they can simply check the camera images in real time to decide if a line trip was due to visible damage, enabling faster restoration or preventive shutdowns as needed. The integrated camera and sensor network essentially acts as an automated sentinel for both fire and asset health, which boosts safety for the community and the grid. SRP’s emergency managers benefit from this comprehensive view: they can see wildfire threats and grid status side-by-side, improving coordination (for instance, knowing if a fire is approaching a line or if a line failure might have caused a fire). Overall, the system provides peace of mind that even in remote, difficult terrain, SRP’s critical infrastructure is being watched in real time and protected by intelligent automation.
Outcome and impact: since implementing Phase 1 and 2, SRP’s wildfire management has become both faster and smarter, yielding significant benefits. The utility now receives timely, location-specific alerts whenever a wildfire threat emerges near its lines, often providing extra minutes or even hours of lead time for officials and residents to prepare or evacuate compared to prior methods. Fire ignitions that might previously have grown large before detection are now caught in their infancy, and several potential crises have been averted through early intervention. Importantly, this two-phase project demonstrates how combining AI wildfire detection with smart weather stations can significantly strengthen grid resilience and wildfire preparedness for utility companies. SRP’s pilot has become a model for other power companies facing similar wildfire challenges. The approach shows that leveraging existing utility infrastructure (like transmission towers) for new purposes, mounting cameras and sensors, can create an innovative early warning network without requiring dedicated surveillance infrastructure. By focusing on factual performance and effectiveness, SRP can report that the SmokeD system delivered on its promises: faster wildfire detection, improved forecasting accuracy, quicker response times, and enhanced infrastructure safety. Ultimately, the project has helped protect both the electric grid and the community by providing SRP and first responders with the critical intelligence they need to stay one step ahead of wildfires, rather than one step behind.
Introduction
A SmokeD optical wildfire detection camera installed on a forest observation tower, scanning for early signs of smoke. Kampinos National Park is a vast woodland expanse just outside Warsaw and holds status as a UNESCO Biosphere Reserve. It is one of Poland’s largest and most frequented national parks, home to rich biodiversity (over 16,500 species) and attracting more than a million visitors each year. These natural and human factors have heightened the park’s vulnerability to wildfires – prolonged droughts linked to climate change and the heavy tourist footfall especially in summer have increased the risk of fires. Park authorities recognized the need for advanced fire detection technology to protect this national treasure. In mid-2025, they introduced the SmokeD AI-powered wildfire detection system as part of a broader fire safety modernization, aiming to enable faster response to potential fires in this high-value, high-traffic park.
Challenges
Before the SmokeD system’s installation, Kampinos National Park faced several fire monitoring and operational challenges:
Solution
To address these challenges, Kampinos National Park implemented the SmokeD early wildfire detection system as the core of its new fire monitoring strategy. The SmokeD solution consists of a network of high-precision smoke detection cameras coupled with pan-tilt-zoom (PTZ) functionality and artificial intelligence for image analysis. As part of the project, SmokeD installed 5 Manta PTZ cameras and 15 Duo Vue detectors (a set of three on each tower) across the park’s 5 observation towers. These detectors and cameras were mounted atop the park’s existing forest observation towers to provide a panoramic view of the surrounding woodlands. The devices operate around the clock, continuously scanning the horizon. The live video feeds from the cameras are analyzed in real time by SmokeD’s AI algorithms, which are trained to recognize telltale signs of a wildfire. During daylight hours, the system looks for rising smoke plumes, while at night it can detect the glow of flames, effectively providing 24/7 coverage in all lighting conditions. According to the system’s specifications, each detector can monitor up to a 15 km radius, giving the park wide surveillance coverage from each tower.
SmokeD also provided communications infrastructure in the 10.5GHz licensed band, ensuring reliable data transmission between the towers and the central fire monitoring center. This ensures that all devices stay connected even in remote parts of the park.
When the SmokeD AI identifies a potential fire (for example, a distant wisp of smoke), the system automatically determines the precise location of the threat and triggers an immediate alarm. It sends a digital alert to the designated park personnel and firefighting authorities, complete with the GPS coordinates or a map of the suspected fire location. The alerting is integrated via both a web dashboard at the park’s fire control center and through mobile applications, ensuring rangers and responders in the field receive instant notifications on their devices. This automation ensures the system can quickly relay information to those who can act, much faster than would be possible via human observation and radio call-ins.
It is important to note that SmokeD’s responsibility in this project was focused solely on the fire detection system – namely the detectors and cameras, detection software, and alerting platform. The broader infrastructure supporting it was handled separately by the park. As part of the park’s modernization program, the observation towers were upgraded and equipped with SmokeD’s advanced camera units. SmokeD played a key role by providing the detection system, including the cameras and AI technology for real-time analysis, while the park handled the necessary infrastructure upgrades such as the physical towers and communication network to support the system. This clear division of scope meant that SmokeD’s implementation could integrate seamlessly into Kampinos’s existing infrastructure. The Monitoring Centre at the park’s headquarters was equipped with the SmokeD monitoring software, and field staff were given access to the SmokeD Alerts mobile app, but all of this rides on the park’s own towers and upgraded radio networks connecting the system together.
Results
The introduction of the SmokeD system has markedly improved Kampinos National Park’s ability to detect and respond to wildfires. The most immediate benefit has been a significant reduction in detection time. Fires that once might have gone unnoticed until a lookout spotted visible flames or a passerby reported smoke can now be identified within minutes of ignition. In fact, the new automated system often spots smoke in a few minutes of its appearance, enabling park staff to spring into action at the earliest stage of an incident. This speed is critical – by catching a wildfire at its inception, the park’s responders can contain it while it is still small. According to reports, the rapid alerts provided by SmokeD are crucial for limiting the spread of fire and minimizing environmental damage in the park’s valuable ecosystems. In practical terms, the window of reaction has shifted from what could have been tens of minutes (or longer if occurring at night) down to almost real-time awareness, vastly improving the odds of dousing a fire before it grows.
Park authorities report that safety has improved since SmokeD’s deployment. They expect that this technology-driven approach will bolster the protection of the Kampinos forest and make firefighting interventions “more effective and faster than before”. Early detection means not only smaller and more easily controlled fires, but also reduced risk to firefighters and visitors. The park’s management has expressed satisfaction with the system’s performance. In a public statement, the Kampinos Park directorate highlighted how the new AI system addresses their peak-season challenges: “In the summer period, with increased tourist traffic and high fire risk, every minute has significance. Thanks to this system we can react faster to threats and protect nature as well as visitors”. This endorsement from the park leadership underscores that the SmokeD implementation is meeting its key objectives – providing the quick alerts and actionable information that were urgently needed.
Beyond these immediate results, Kampinos National Park has taken a place at the forefront of modern wildfire prevention through this project. It is now one of the few national parks in Europe to have deployed an AI-driven wildfire detection system at scale, blending traditional conservation stewardship with cutting-edge technology. The successful integration of SmokeD’s detectors and AI analytics with the park’s operations serves as a model for how protected areas can leverage technology to adapt to growing wildfire risks. In sum, the SmokeD system’s implementation in Kampinos has achieved faster fire detection, improved overall safety, and earned the confidence of park authorities, a significant step forward in safeguarding both the natural heritage and the public in Poland’s beloved “green lungs” near its capital.
The Challenge
The State Forest Service of Latvia (Valsts Meža Dienests) oversees millions of hectares of forests, and in recent years it has faced a surge in wildfire incidents. For example, the 2023 fire season saw 643 forest fires scorch 636 hectares of land – more burned area than in any of the previous three years. Some years have been even worse; in 2019 nearly 1,000 wildfires charred over 800 hectares. This upward trend in fires, driven by drier summers and human factors, underscored an urgent need for faster and more comprehensive detection of wildfires.
Traditionally, Latvia has relied on a network of about 180 fire lookout towers staffed by seasonal observers to spot smoke. However, maintaining full coverage with human lookouts has grown difficult. The State Forest Service struggles to recruit and retain enough tower watchers (the work is demanding and wages are modest), leaving gaps in surveillance. Many remote or rugged forested areas remain at risk of fires starting unnoticed, especially outside of peak hours or when visibility is poor. The challenge was clear: how to modernize wildfire monitoring to provide 24/7, automated coverage over vast forest areas and detect fires in their infancy – without the need for additional manpower. The Service sought a state-of-the-art solution that could quickly pinpoint smoke over long distances, endure off-grid conditions, and integrate with firefighting operations to enable rapid response.
The Solution
SmokeD DuoVue wildfire detectors and a PTZ camera mounted on a forest tower, powered by solar panels. The State Forest Service partnered with SmokeD to pilot an AI-powered early wildfire detection system on two existing fire lookout towers located in forested regions. Each tower was outfitted with a Manta FPS61 HD pan-tilt-zoom (PTZ) camera for 360° visual coverage, along with three SmokeD DuoVue detectors covering all directions. The DuoVue detectors are dual-optics smoke sensors that provide a 164° field of view each, meaning just three units can monitor the full 360° panorama from a tower. These smart detectors continuously scan for any hint of smoke on the horizon and are capable of identifying a rising plume up to 15 km away under good conditions. Once installed, the combined camera-and-detector setup acts as an automated lookout, watching over a broad radius around each tower (covering an estimated 142 hectares of forests) at all times. Because these tower sites had no grid electricity, SmokeD implemented an off-grid power solution: solar panels and high-capacity batteries were installed at each location to sustain the cameras and detectors 24/7, with enough stored energy to run through nights and cloudy days. The real-time video feeds and detectors data from the towers are transmitted securely to the command center via wireless networking (utilizing point-to-point radio links for video transmission from PTZ cameras), with the main internet connection provided by LTE ATL antennas with a SIM card, ensuring that the system remains connected even in remote locations.
In tandem with the field installations, SmokeD set up a central monitoring station for the pilot in the local fire station’s dispatch room. This monitoring center is equipped with two display screens that show live video panoramas from the PTZ cameras on each tower, as well as a computer workstation running the dedicated SmokeD Desktop application. The dispatcher on duty can use this software to remotely control the PTZ cameras – panning, tilting, and zooming in to verify potential smoke sightings – and to review incoming alerts flagged by the system. The SmokeD platform features interactive maps and real-time camera overlays, which help operators quickly orient themselves and pinpoint the reported fire location when an alarm comes in.
In addition to the desktop console, the State Forest Service team has access to SmokeD Web, an online application to manage the detectors and view status remotely, and the SmokeD Alerts mobile app, which pushes instant notifications to smartphones whenever a suspect smoke is detected. The entire system is backed by SmokeD’s AI algorithms running on the tower cameras and detectors, analyzing video frames continuously to spot any telltale wisps of smoke or fire glow. Crucially, SmokeD employs a multi-level alert verification process to minimize false alarms: the moment the AI detects possible smoke on a camera, an alert is automatically sent to both the local monitoring center and to SmokeD’s off-site analysis desk, where a trained specialist quickly reviews the images for confirmation. Only verified alerts are then forwarded as emergency warnings – complete with precise GPS coordinates of the fire and a snapshot of the smoke – to the phones of on-duty forest officers and to the central command screens. This tiered workflow ensures that the fire station dispatcher and forest rangers are only notified when a real fire threat is present, greatly reducing needless call-outs and building trust in the system’s accuracy.
The Effects
A SmokeD camera’s view identifying a distant forest fire moments after ignition – early detection like this is what the Latvian pilot aims to achieve.
Although this pilot implementation is still in its testing phase, it has already begun to demonstrate the transformative benefits of automated wildfire surveillance. The two AI-equipped towers now provide around-the-clock vigilance over approximately 141 hectares of high-risk forests – a level of coverage and consistency that would be impossible to maintain with human observers alone. The moment a thin column of smoke rises anywhere within roughly a 15 km radius of these towers, the system will detect it, often in under 10 minutes of ignition. This speed of detection means fire crews can be alerted and dispatched while a wildfire is still in its infancy, dramatically increasing the chances of extinguishing the flames before they grow out of control. The SmokeD detectors operate continuously, scanning the horizon day and night, even during dawn, dusk, and low-visibility conditions. This 24/7 automated monitoring plugs the critical gaps during off-hours or when human lookouts might miss a small distant smoke plume. As a result, the forest service now has continuous monitoring of the landscape, allowing for near-real-time detection of even small fires, such as a tree igniting or a grassfire spark.
Beyond early detection, the pilot is fostering better coordination and confidence in wildfire response. By integrating the SmokeD system directly into the local fire station’s workflow, any verified fire alarm comes through with actionable information – the exact location on the map, camera visuals of the incident, and details relayed to firefighters’ mobile devices. This allows the dispatchers and firefighting teams to assess the situation at a glance and respond to the right location without delay or confusion. The ability to remotely confirm a smoke sighting via the PTZ camera’s zoom and the backup verification by SmokeD’s analysts has virtually eliminated false alarms, so the staff can act decisively when the alarm sounds. Although final metrics are not yet available (since the pilot is ongoing), early use of the system has been smooth, and the State Forest Service reports that the technology is fitting well into their operations. Front-line personnel have expressed optimism that AI-driven monitoring will reduce their workload during fire patrols and enable faster containment of actual fire incidents. Additionally, they have expressed being truly impressed and satisfied with the performance of the SmokeD system.
While it is too soon for definitive results, the expected outcomes of this pilot are very promising. The SmokeD implementation is on track to significantly cut detection times and improve situational awareness during the critical initial moments of a wildfire. By catching fires sooner and providing reliable alerts, the system should help limit the average burn size and prevent small ignitions from exploding into large conflagrations. Importantly, this project is also serving as a proof of concept for Latvia – showing how existing lookout infrastructure can be enhanced with modern AI, cameras, and renewable power. If the pilot continues to perform well, it will pave the way for a broader rollout of automated wildfire detection towers across Latvia in the coming years. In summary, the Latvian State Forest Service’s collaboration with SmokeD has created a cutting-edge early warning network on a pilot scale, one that blends technology with forestry expertise. Even during this testing phase, the system is offering valuable reassurance: forest managers can rely on continuous monitoring, ready to alert them at the first sign of smoke. This initiative stands to greatly bolster Latvia’s wildfire preparedness and protect its natural resources and communities from the growing threat of wildfires.
