Detección de cambios en imágenes satelitales mediante aprendizaje profundo: caso amazonía ecuatoriana

Abstract

The Ecuadorian Amazon, a vast region of exceptional biodiversity and ecological significance, faces severe challenges due to human activities such as deforestation, urban and agricultural expansion, and mining. Given its extensive area, timely monitoring to mitigate these issues becomes a complex task. The lack of adequate tools has hindered environmental monitoring and management in the Ecuadorian Amazon, necessitating the development of advanced techniques to address these issues. This work focuses on the need to implement precise and efficient methods for detecting and classifying changes in land cover in this region. To address this need, a method was developed that combines the use of satellite imagery and deep learning techniques. Historical images were collected and preprocessed, using convolutional neural networks that have demonstrated high accuracy in detecting changes in land cover. The results show that these techniques are effective not only for detecting changes but also for classifying the types of affected coverage, providing valuable information for the implementation of conservation policies and environmental management in the Ecuadorian Amazon.