ABSTRACT: Environmental and vegetation monitoring traditionally relies on raw satellite datasets that are difficult for nonspecialist users to access, process, and interpret. This paper presents SatPredict AI, an Android mobile application that fetches satellite and environmental data for a user-selected location, processes it through a Python backend, and delivers both a current environmental condition assessment and a future trend prediction using Machine Learning. The system combines a Flutter-based mobile front end, a FastAPI backend, satellite data services such as Sentinel-2 and Landsat, and a PostgreSQL/PostGIS spatial database. Two learning algorithms form the analytical core: a Random Forest classifier evaluates the current environmental condition from indicators such as the Normalized Difference Vegetation Index (NDVI), temperature, and rainfall, while a Long Short-Term Memory (LSTM) network forecasts future vegetation and environmental trends from historical time-series data. The findings indicate that this hybrid classification-and-forecasting approach converts complex, multi-source satellite data into a simple, understandable mobile experience, with clear potential for extension into agriculture, drought-risk, and land-use monitoring.
KEYWORDS: Satellite Data, NDVI, Machine Learning, Random Forest, LSTM, FastAPI, Flutter, PostGIS, Environmental Monitoring, Mobile Application
SATPREDICT AI: SATELLITE DATA FETCHING AND ENVIRONMENTAL CONDITION PREDICTION USING MACHINE LEARNING AND BACKEND APIS
MS. SHRUTIKA DAVE