Automated detection of green land cover regions in satellite images can play a vital role in the effective estimation of vegetation area in order to monitor ecological balance for ensuring environmental safety in the wake of deforestation. This study aims to contribute in terms of meeting the demand of deploying an automated tool as an assistive ecosystem service provider by developing an innovative two-stage green land cover detection framework analyzing satellite images. The framework is specifically designed to tackle the challenges related to the deployment of automated computer vision-based frameworks for detecting green land cover over large areas using satellite imagery. In the initial stage, a clustering strategy is applied for the segmentation of various types of chromatically homogeneous land cover regions in the satellite image. For the second stage, an autoencoder model is built with varied green land cover samples which is later employed to distinguish only green land cover clusters from other clusters representing water bodies, barren fields, built-up areas, etc. The proposed framework showcases an impressive detection accuracy of over 96%, even with a small experimental dataset which is found to be comparable to the results of widely used Mask R-CNN-based segmentation tools.