An early detection system for wildfires is a key component when it comes to an efficient global fire fighting system. Smoke is a clear signal of a fire and also makes a significant signature on a satellite image. Clouds could be a challenge, but wildfires are more likely to start on cloud-free days. Within the Copernicus program EFFIS, there are methods for
early detection of wildfires, but based on temperature and not smoke detection.

In AI VISIONS we think that a smoke detection system, using AI, could be a good supplement to the methods used today. And such a system will also have the ability to give a more exact positioning of the fire spot. Tests we have completed show that training of a convolutional neural network to fulfill this task is within range.

Images from the Copernicus satellites have the quality needed to detect these smoke patterns, and with technology that can easily scan hundreds of thousands of images per hour this will work perfectly in combination with a warning system for high risk areas like EFFIS.

#AIforGood #Computervision #Machinelearning #AI #AIVISIONS #neuralnetworks #Landfills #Copernicus #Satelliteimages #Deeplearning