Comparative Analysis of Computer Vision AI Models for Fire Detection

This study compares two artificial intelligence models for fire detection in images, aiming for their integration into a security drone. Two approaches were evaluated: one using Google Teachable Machine for its ease of use and implementation, and another trained in Google Colab, offering greater flexibility and adjustable precision. Performance metrics such as accuracy, sensitivity, and confidence in classifying fire and non-fire images were analyzed. The results indicate that the Google Colab model has a higher fire detection rate, while the Teachable Machine model is easier to implement and less sensitive to variations in lighting. This research lays the foundation for the development of a modular autonomous monitoring system, with future improvements in smoke detection and security drone integration.