DOI: 10.15393/j2.art.2022.6523
Vycherova Nataliya Romanovna | Ukhta State Technical University, nvycherova@ugtu.net |
Budevich Evgeny Arturovich | Ukhta State Technical University, ebudevich@ugtu.net |
Belyaev Andrey Eduardovich | Gazprom Nedra PF "Vuktylgazgeofizika" LLC, belandre@yandex.ru |
Key words: drone computer vision pattern recognition by the method of support vectors (Support Vector Ma-chines SVM) |
Summary: For most of the world, wildfires continue to bea serious problem. There are now many ways to deal with them, and they are all mainly aimed at reducing the damage caused by fires using early detec-tion methods. According to the estimates of the Federal Forestry Agency, on average, the amount of damage from forest fires per year is about 20 billion rubles, of which from 3 to 7 billion is damage to forestry (wood loss). The article discusses the use of unmanned aerial vehicles (UAVs) for regular pa-trolling of potentially dangerous areas of fire distribution, using the advantages of artificial intelligence (AI) and the ability to self-process the information received. This allows UAVs to use computer vision techniques to detect smoke or fire, based on information received from their video cameras |