Natural means of preventing forest fires exist: clearing, installing firebreaks, transplanting less flammable species... However, it is impossible to eradicate all these fires often caused by human negligence. Early detection is therefore crucial, which is what FireTracking, a young startup incubated within the startup studio Kaukana Ventures of Emerton Data, a research, innovation, and AI consulting company, proposes. Both have partnered with Axione, a central player in connectivity, and have won a project call launched by the Indre-et-Loire Departmental Council, aimed at enabling the Departmental Fire and Rescue Service (SDIS) to respond quickly to these fires.
Climate change, which causes water stress in trees sometimes combined with strong winds, promotes the rapid spread of fires. The department of Indre-et-Loire is not immune, with more than 300 fire outbreaks per year and an alarming increase. To face this challenge, it has decided to trust the consortium formed by FireTracking, Emerton Data, and Axione to detect and locate these fires through the installation of a network of intelligent cameras on twelve strategic sites and low-latency transmission devices.
The result of two years of joint development with Kaukana Ventures, FireTracking's solution relies on deep learning algorithms that can detect fire outbreaks in less than three minutes, while maintaining a false alarm rate of less than 10%. Initially tested during a pilot project in New Caledonia, this innovation has proven its efficiency and reliability, even in extreme conditions.
The cameras are installed on existing high points, such as pylons or antennas, with the ultimate goal of continuously monitoring nearly 95% of the Indre-et-Loire forest massifs.
Axione, with its expertise in digital infrastructure and network management, having deployed over 400,000 kilometers of networks across France, plays an essential role in the implementation of this project. Its connectivity expertise ensures real-time transmission of alerts generated by FireTracking's AI, allowing firefighters to respond more quickly and effectively.
Deployment in Two Phases
The consortium will implement its project in two stages. The installation and validation of the solution on six sites began last January to cover the most sensitive areas within the department and will be completed next June. From January to June 2026, the solution will be extended to six additional sites, thus covering almost all the department's forest massifs.
To better understand
What is deep learning and how is it used in fire detection?
Deep learning is a subfield of artificial intelligence that uses artificial neural networks to model complex data. In fire detection, it enables systems to understand and analyze images captured by cameras to quickly identify fire outbreaks based on pre-trained models that recognize visual signs characteristic of a fire.
How do low-latency connectivity infrastructures impact the detection and management of forest fires?
Low-latency connectivity infrastructures enable fast, real-time transmission of data from sensors and cameras to processing centers. This ensures that alerts about fire outbreaks reach emergency services almost instantaneously, reducing reaction time and allowing for quicker and more effective interventions to prevent the spread of fires.