On May 27, 2025, the first public conference of the Karst Firewall 5.0 Interreg Italy–Slovenia project was held in Sesljan (Sistiana), marking an important milestone in cross-border cooperation for wildfire prevention in the Karst region. The event brought together institutional partners, research institutes, and companies, including Infordata Sistemi, which as project partner for the digitalisation is developing an innovative AI-based predictive algorithm to predict wildfire risk.
The conference received wide media coverage, demonstrating the significance of the project:
- Koper-Capodistria Television aired two segments (in Italian and Slovenian) focused on the event – [Italian video link] – [Slovenian video link]
- The newspaper of the slovenian community in Italy, Primorski Dnevnik, published a detailed article – [article link]
- Coverage also appeared in Primorske novice, a major newspaper in Slovenia’s Littoral region – [article link]
Key updates included the deployment of thermal drones, the installation of 1,000 IoT “electronic noses” to detect smoke at early stages, mapping and restoring dry-stone walls as natural fire barriers, and experimenting with replacing black pine forests.
Here is the English translation of the improved text:
Infordata’s team rappresented by Marko Petelin, Nicola Zuccato and Samuel Zidarich, as the technical partner of the Karst Firewall 5.0 project, presented how they conducted an in-depth analysis of historical data dating back to 1993, processing over 8 million meteorological records from ARPA (Italy) and ARSO (Slovenia) stations in the cross-border Karst area, along with approximately 38,000 wildfire events. These data were used to develop a predictive model based on Machine Learning algorithms (specifically XGBoost), currently capable of achieving 85% accuracy in classifying fire-risk days and 81% recall for actual fire events.
The algorithm is continuously evolving and will be further optimized throughout the project duration, as predicting wildfires requires evaluating numerous interrelated variables (weather, geomorphology, land use, anthropogenic factors). The predictions serve as decision-support tools and do not replace expert human judgment, in full compliance with ethical standards and the EU AI Act principles.
The Karst Firewall 5.0 project, led by the Iuav University of Venice, is a cross-border cooperation initiative involving Infordata Sistemi, the Slovenian research center ZRC SAZU, and the municipalities of Duino Aurisina (Italy) and Miren-Kostanjevica (Slovenia). The project aims to strengthen the Karst region’s resilience to wildfires by promoting an integrated approach based on advanced technologies – including drones, satellite imagery, electronic noses, and IoT sensors – and AI-based predictive models, to support an innovative early warning and fire management system.