Artificial Intelligence to calculate the probability of a fire before it happens.

During the summer months, forest fires are frequent throughout the Iberian Peninsula. In Spain, on average, during the last ten years, 28,000 hectares have been burned annually, according to data from the Ministry of Agriculture, Fisheries, and Food. But technology can play a crucial role in improving the response and management of fires.

Artificial Intelligence to calculate the probability of a fire before it happens.
Artificial Intelligence to calculate the probability of a fire before it happens.

In Spain between January 1 and June 16, there have been a total of 6,220 forest fires, almost 4,000 more than in 2018, according to government data. So far this year, nearly 40,000 hectares have been burned, a third more than the average since 2009.

   At European level, June 2019 has been the warmest month of June since records. A situation that facilitates the appearance and spread of fires, as indicated by IBM in a statement.

    To detect and combat fires, the Portuguese company Compta has incorporated Watson, IBM's artificial intelligence, into its Bee2FireDetection solution. This tool can calculate the probability of a fire occurring before it happens.

Bee2FireDetection uses Artificial Intelligence, spectrometers, different digital cameras for video surveillance and thermography (a technique that allows analyzing temperatures at a distance) to detect forest fires at distances of up to 15 kilometers without interruption, 365 days a year.

     The collected images analyze with IBM Watson Visual Recognition, which uses different deep learning algorithms to allow the system to detect any sign of fire, such as a smoke column, a temperature increase or a change of light, and to give the alert, practically in real-time.

This system can also calculate the level of fire risk, combining Watson with a series of meteorological and environmental terrain data, such as relative humidity, wind direction or temperature.

    By collecting and analyzing this massive amount of data, Bee2FireDetection can provide a very accurate forecast of the likelihood of a fire occurring before it occurs. Knowing this risk index allows you to establish an appropriate alert level, and that fire fighting teams can be prepared to act on time.

The artificial intelligence of IBM also allows that Bee2FireDetection can be a handy tool in those cases in which a fire is already active. Taking advantage of Watson's capabilities and the analysis of different meteorological and environmental data, Bee2FireDetection can estimate how the light will evolve on the ground.

    The system can calculate, for example, the speed and direction that the flames could take. Having this valuable and detailed information can help firefighters and other civil protection bodies to plan fire control in a very agile and efficient way.