This document outlines a project aimed at automating the classification of satellite imagery of the Amazon rainforests using the VGG16 deep learning model. The authors achieve high accuracy in identifying land cover types from a dataset of over 40,000 images, with training and testing accuracies of 97.35% and 96.71%, respectively. The automated labeling system can assist in tracking deforestation and human encroachment, thus aiding governmental and local responses.