Abstract
Soil erosion is very hazardous to the global ecosystem. Government aided soil erosion control schemes happen dilatorily with minimal resources. Recognition and identifying the scale and the area of eroded land can be extremely time-consuming and difficult as well. To overcome this problem, a real-time Soil erosion detection system is introduced. The real-time part has been implemented using satellite imagery with the use of RUSLE modelling considering various factors. This was generated with the help of Google Earth Engine (GEE) interface. The RUSLE model offers a straightforward approach to assess soil erosion. By using remote sensing data and GIS, RUSLE effectively evaluates erosion. Researchers have developed various equations to model the five factors of the RUSLE model, considering the diverse variations in the soil erosion process. The system also includes the analysis of satellite imagery with a mapped view of soil erosion. Here, the Unet (EfficientNetb3) model is used giving optimal accuracy for the detection of soil erosion.