An early fire notification system based on intelligent video technology successfully passed comprehensive testing in Miloshevichi forestry station (Lelchits district, Gomel region, Belarus).
Based on the Axxon Intellect PSIM platform with smoke detectors powered by deep learning, the sample system includes three PTZ video/thermal combos located on 35 meter-high lookout towers. The customer, Belgosles state enterprise, is completely satisfied with the test results.
Cameras automatically pan 360 degrees to scan the forest for rising smoke. The neural network was specifically trained in similar environments to distinguish smoke from clouds, fog, and dust. The detection distance ranges up to 50 km in clear weather.
Video and thermal images are displayed on an operator's monitor along with an interactive map where current camera FOVs are indicated. Upon detection of smoke, an alarm sounds while a set of targeted emails containing a map screenshot and the alarm image are sent. A hot spot can be localized with an accuracy of several hundred meters, which is more than satisfactory regarding the scope of forest lands.
AxxonSoft software will replace the legacy system, which is functional but non-neural, limited in detection range, less accurate, and prone to false alarms.
Miloshevichi forestry station is the first site to deploy the new system. The customer strategy is to expand the solution over the whole national forestry service. This will increase overall awareness while lowering risks related to human error. Moreover, the intelligent video system requires less operator attention and effort due to its fully automated detection and notification.
It's worth noting that surveillance cameras are used on 80 percent of 580 forest lookout towers and masts in Belarus. This forms a solid base for further deployment of the early forest fire notification system.