Development of a Wildfire Detection and Monitoring Solution Using Computer Vision

In this work, a platform for the detection, monitoring, and alerting of wildfires is developed, with the aim of demonstrating that the integration of artificial intelligence and other current technologies allows for the development of robust solutions to address various challenges. For this purpose, different versions of YOLOv8 were trained for the detection of fire and smoke. The result of this yielded the best model, the large version of YOLOv8, with metrics exceeding 80% precision and mean Average Precision. This model is integrated into a platform developed to keep a record of all generated alerts and to take actions on them. The platform features functionalities to record the location, date, time, and risk level of each alert. Additionally, it allows for viewing these alerts on a map and includes visualization elements to track the status of all alerts.