An automatic detection system with AI makes it possible to fight fires early and effectively

The French image processing solution thus makes a decisive contribution to limiting the damage. Thanks to the early detection of fires and the precise localization of the sources of fire, it considerably reduces the risk of propagation and can minimize the resulting damage to people, the environment and the economy.

Heat waves caused by climate change are currently on the rise across Europe and the resulting risk of forest fires is increasing immensely. How to detect and locate fires in time in order to minimize, or even avoid, damage with serious consequences? Image processing and artificial intelligence make it possible to meet such challenges. Neural networks and deep learning algorithms allow an image processing system to see, recognize and verify objects – in this case, smoke.

The French company Paratronic has looked into the matter. Within the scope of its business area natural hazard monitoring, the solution provider successfully focuses among other things on the development of intelligent products for fire monitoring. Their automatic forest fire detection system ADELIE (Alert Detection Localization of Forest Fire), whose key components are industrial image processing and artificial intelligence, has proven itself in the field. Four industrial cameras from IDS Imaging Development Systems GmbH are integrated into each system.

360° monitoring

They constantly observe a specific forest area within a radius of up to 20 kilometres. Depending on the installation, it takes them a maximum of two minutes to check a 360 degree radius. Thanks to the algorithms developed by Paratronic, the system is able to detect and locate the sources of fire from the images taken and provide real-time information for the corresponding possibilities of action. The ADELIE system thus guarantees effective planning and guidance of firefighter intervention forces for the protection of our living environment, not to mention the protection of buildings, power lines, telecommunication lines, road infrastructure. or rail.

The ADELIE system consists of at least two interconnected monitoring points. Each surveillance point includes two detection cameras and an additional camera which is used to remove doubts. Four IDS ethernet cameras are integrated in each ADELIE detection camera. A total of eight IDS cameras are thus used per monitoring point. These points allow 360° surveillance, each azimuth being visualized approximately every two minutes. Automatic monitoring of the observed natural area is carried out 24 hours a day, seven days a week.

Real-time visualization of the event

The system is connected to a processing unit whose software contains image processing algorithms based on artificial intelligence. The program developed by Paratronic records, compares and analyzes the images provided by the cameras. By comparing images and based on learned characteristics, the system detects rising smoke. As soon as this smoke is visible from the monitoring station, ADELIE triggers the alarm. This phase is the automatic detection of fires and forest fires. The service operator then controls the camera remotely to remove the doubt and checks the nature of the detection. It locates the source of the fire by triangulation on a map and informs the intervention center, which launches fire-fighting measures.

At the same time, all the information, images and knowledge acquired by the AI ​​are transmitted without delay to the fire alarm center or the fire control center. Thanks to the real-time visualization of the event, the localization of the source of fire on a digital map and various augmented reality functions, it is possible to immediately visualize the context, the extent and the evolution of the fire. fire, and to take appropriate measures to fight it. A remote-controlled video camera completes the system. This is used to check and monitor the fire until the arrival of the first extinguishing unit and allows the fire to be followed without interruption from its start until its extinction.

When choosing the right model for the automatic forest fire detection system, the decision was made in favor of an ethernet camera from the uEye SE series from IDS. “Our system uses the UI-5240SE-NIR-GL model,” explains Loïs Carrié, development engineer at Paratronic. This particularly powerful industrial camera is equipped with a 1.3 megapixel CMOS sensor from e2v. The high-sensitivity sensor is used by Paratronic in the near-infrared NIR variant (EV76C661ABT). In addition to its exceptional light sensitivity, the sensor offers two shutter variants (global and rolling), switchable during operation. This allows maximum flexibility in case of changing environmental requirements and conditions, as in this case, caused by different times of day and weather conditions. In addition, four areas of interest (AOI) are available. This allows multiple features to be monitored simultaneously or AOIs to be captured in a series of exposures with different settings.

The camera thus meets all the requirements, confirms Loïs Carrié. “We chose this model for three main reasons. On the one hand, it convinces with its spectral sensitivity. The sensor captures all visible color wavelengths, with particularly good sensitivity in the near infrared. Additionally, we need to be able to place a wavelength filter close to the sensor in the C-mount. Finally, the camera allows four images to be taken sequentially with increasing exposure time. Continuous shooting achieves a very high dynamic range.”

For image capture, the system uses the uEye software development kit. “This is where our own image processing system comes into play,” explains Edouard Bouillot. The ADELIE software is then responsible for analyzing the images in order to detect the presence of smoke on the canopy. The analysis takes place by comparing two images taken in the same orientation in order to detect possible smoke. This is made possible by several exclusive algorithms developed by Paratronic, which make it possible to compare a very large number of factors invisible to the naked eye.

This analysis takes place in three phases. In the first phase, the images to be compared are recorded to the nearest 50th of a degree. The second phase consists of comparing the images with each other in order to highlight any changes, such as the movement or displacement of objects or the appearance of smoke. The third phase consists of an advanced analysis based on the use of different algorithms: The highlighted differences are not only examined from the point of view of their shape, size, distance, etc. in order to eliminate as much as possible of all elements other than smoke. Other algorithms using automatic classifiers and working with parameters extracted from one or more images complete this analysis.

13,500 images are taken in 24 hours

The data is then transmitted to the computer control system via a digital network, for example fiber optic cables. The respective datasets contain both a JPEG file of the image for on-screen display and a file containing the camera number, viewing angle, date and time of shooting. sight as well as the azimuth. The integration of a weather station also makes it possible to enter and take into account weather data such as wind strength or precipitation. If an image and its associated file indicate a fire, an automatic check is performed: the system estimates the location of the smoke and then compares it to known locations where other types of smoke are present. In this way, it is ensured that an alarm is triggered only if the detection has not taken place in an exclusion zone, i.e. in an area where it is known that smoke is constantly present, as in a factory chimney. If only one tower detected the smoke, the range specified in the telemetry is used. If at least two towers have detected it, the precise localization of the source of the fire takes place at the level of the plant by triangulation.

As with any automatic system, human validation of the transmitted alarms is also essential for ADELIE. Control center personnel use a high-resolution camera with a powerful optical zoom (30x, with a wide-angle lens) to determine if this is really the start of a fire. Thanks to these cameras, called doubt detectors, the people in charge of surveillance can observe the situation from a distance without interrupting the detection system. The ADELIE notification system therefore remains fully active in order to be able to deal with the possible appearance of several fires.

The system is extremely efficient. For each site, 13,500 images are taken in 24 hours, transmitted and stored for 30 days, whether or not they contain a detection. In addition to the captured images, the system also records video from the camera used to observe and validate the incident, allowing for full documentation. Based on all the data collected, ADELIE can establish statistics from which the intervention forces can adjust and optimize their actions. “Thanks to the recorded images, we can analyze the course of the fire and the subsequent firefighting. The amount of data continually increases with each event that enters the database. This then increases the reliability of the statistics, which are necessary for the continuous improvement of prevention and control measures,” summarizes Edouard Bouillot.

A truly integrated fire information monitoring and management system

The system is particularly used successfully in Sarthe, the most wooded department in the northwest quarter of France. A total of 48 cameras now monitor the forests above the treetops at twelve points spread across the department, near particularly vulnerable areas.

ADELIE is therefore more than an automatic fire (and forest fire) detection system, it is a real integrated monitoring and information management system. The French image processing solution thus makes a decisive contribution to limiting the damage. Thanks to the early detection of fires and the precise localization of the sources of fire, it considerably reduces the risk of propagation and can minimize the resulting damage to people, the environment and the economy.

Learn more about monitoring and managing fire information with IDS Imaging: HERE

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *