In recent years, Sweden has been hit by a series of extensive forest fires that have caused major material damage. In 2014, during the big fire around Surahammar in Västmanland, people and animals had to be evacuated. Early detection of forest fires with the help of drones could lead to major economic and social benefits.
Foggy problems
But for this to become a reality, drones need to be smart enough to find suspicious fires and make accurate assessments. One of the difficulties is that drones today have difficulty distinguishing between fire smoke and fog.
"In this way, you can tell at an early stage if there is a real fire or if it is a false alarm. You can quickly determine how big the fire is and what resources are needed to fight it," says Julia Åhlén.
By analyzing the shape of the smoke formations that the drones find in the forest, the algorithm today identifies 90 percent of the images correctly – a figure that Julia Åhlén believes will be even better in the future when AI is used to train the algorithm on a larger image material.
Can you see a future where artificial intelligence in combination with drones means that we completely avoid major forest fires?
"Of course, that's my great hope.