Sistema inteligente de mapeo y visualización geoespacial de áreas verdes en ESPOL

Abstract

The sustainable management of the Gustavo Galindo Velasco campus at ESPOL requires precise information regarding its green areas. Currently, these lack an up-to-date digital inventory for monitoring. This project develops an intelligent system for geospatial mapping and visualization to support territorial planning. To this end, an interactive web platform integrating Geographic Information Systems and deep learning was implemented. Drone images were processed using the U-Net network for semantic segmentation, and the system was validated through functional tests. The results showed that the model achieved an accuracy of 85.91%. Furthermore, functional tests demonstrated that users completed tasks in less than 5 minutes with a success rate exceeding 90%. Thus, the system effectively integrates artificial intelligence and geoinformatics, providing a robust tool to support evidence-based decision-making. Keywords: Semantic Segmentation, Deep Learning, Environmental Management, Neural Networks.