Comparative analysis of Image classification capabilities of Support Vector Machine (SVM) and Random Forest (RF) with Google Earth Engine (GEE) platform: A case study of Sangamner, Maharashtra

Abstract

Support Vector Machine (SVM) and Random Forest are supervised machine learning
algorithms known for their ability to precisely classify complex landscapes on earth surface.
These advancements have been very productive for Geographical Information System domain
to monitor natural and anthropogenic transformation using remotely sensed datasets. In the
present study, Google Earth Engine (GEE) platform has been utilized to identify different land
use land cover zones of Sangamner tehsil of Maharashtra. Sentinel MSI satellite images of
January 2019 have been accessed and classified over GEE with both SVM and RF classifier. The
classification results demonstrate that the SVM classifier performs better than RF over study
area with 94.50% and 78.38% overall accuracy. The results obtained from the study illustrate
that the major area is utilized for agricultural and urban practices.