Fire Detection at the Edge

The purpose of this tool, complementary to Fire Detection from IoT Devices is to detect fire and smoke in videos and photos taken from UAVs and UGVs. These photos or videos are sent to the ground station via Mesh-in-the-Sky or downloaded upon arrival. The tool analyses, using computer vision algorithms, the received images detecting fire and smoke. The result of this analysis, with the addition of geolocation of the images, is sent and used in the Dashboard. The tool is able to process video footage in near real time.

The algorithm used for detecting fire and smoke has been trained using a synthetic dataset of more than 30000 images of fire and smoke in forest areas generated for the project.

The analysis uses the edge capabilities of SILVANUS. The analysis is made on the fire monitoring station or control station.

Capabilities and Benefits:

• Fire and smoke detection in near real time: The tool is able to detect fire and smoke in real time footage from any available source able to provide video or images (cameras, UAVs, etc).
• Support to decision: The tool provides additional data that the operator can use to detect new fires and monitor its evolution. The tool does not make any decision by itself but helps the operator to make informed decisions adding visual information and the automatic detection of the fire.
• Liberate operator from monotonous tasks: This tool eliminates the necessity of the operator to be visually checking multiple monitors which can be a monotonous task prone to distractions. The tool can autonomously detect fire and smoke, sending then a signal to the operator and showing the fire event in the image.