Sistema de Detección y Alerta de Incendios Forestales Basado en Procesamiento de Imágenes e IA

The objective of this work focuses on the design and implementation of a low-cost intelligent system for the early detection of forest fires, using IoT (Internet of Things) and AI (Artificial Intelligence) technologies. The designed methodology involved the development and comparative evaluation of multiple algorithms, divided into two approaches: techniques based on the analysis of color spaces (RGB, YCbCr, HSI, HSV, and PJF) and AI models (CNN, YOLOv8, and Haar Cascade). From this analysis, a hybrid architecture was selected that integrates the two highest-performing methods: an object detector based on YOLOv8 (Method 9) and a chromatic algorithm that fuses the PJF, RGB, and YCbCr spaces (Method 12). This visual system is complemented by a PM2.5 particle sensor to validate the presence of smoke and GPS/4G modules to issue georeferenced alerts. As a key result, the final prototype validated under controlled conditions achieved outstanding metrics such as 99.82% accuracy, 99.64% sensitivity, and 100.00% specificity under high illumination conditions. It also demonstrated energy efficiency and thermal stability through continuous monitoring of CPU, RAM, and current consumption. The main contribution of this work consists of a validated field solution, whose hybrid architecture proves to be accurate, efficient, and adaptable, confirming its feasibility for implementation in emergency contexts.