Paddy Segmentation Using Google Earth Engine: A Remote Sensing Approach Abstract

Paddy field segmentation using remote sensing is crucial for agricultural monitoring, yield prediction, and resource allocation. In this research, we employ Google Earth Engine (GEE) for paddy segmentation using Sentinel-2 satellite imagery. Our method leverages Normalized Difference Vegetation Index (NDVI) and Land Surface Water Index (LSWI) to mask paddy fields efficiently. We collected 2000 images (masked and unmasked), trained a ResNet model achieving 91% accuracy, and implemented a real-time mobile application. This paper details the dataset preparation, masking methodology, implementation pipeline, and mobile app integration.