Currently, agriculture faces major challenges related to the efficient utilization of natural resources, the optimization of crop productivity, and the maintenance of sustainable practices under increasingly unpredictable climatic conditions. Traditional monitoring techniques are often limited in spatial and temporal coverage, making it difficult to achieve timely and accurate assessments of land surface conditions. This study proposes the creation of an intelligent system for monitoring land surface coverage, encompassing agricultural crops, grasslands, and various land types, utilizing open-source satellite data and integrating machine learning models. The research is applied in the Banat region of Romania, a representative area for lowland agricultural ecosystems. The system acquires 20 meter resolution RGB imagery from the Copernicus browser, which is then processed to compute vegetation indices such as ExGR (Excess Green minus Red Index). These indices are essential indicators for evaluating vegetation health, density, and soil moisture. The processed data are subsequently analyzed using supervised and unsupervised learning algorithms to automatically classify land cover types with a high degree of accuracy. The results demonstrate that the proposed system effectively identifies and monitors diverse land surfaces, supporting improved agricultural management and precision farming strategies. This research highlights the potential of combining remote sensing, open-source satellite imagery, and artificial intelligence to create scalable, cost-efficient tools for sustainable agricultural and environmental monitoring.

Banat, Neural Network, open-source, RGB

geodesy engineering

Presentation: poster