This research presents a novel approach for deforestation detection using convolution neural network with satellite imagery. The proposed method leverages deep learning techniques to reconstruct satellite images and detect patterns indicative of deforestation. Through extensive experimentation and evaluation, the effectiveness of the approach is demonstrated on a dataset comprising both forested and deforested areas. The trained model achieves promising results in accurately distinguishing between forested and deforested regions, showcasing its potential for environmental monitoring and conservation efforts. This study contributes to the advancement of automated deforestation detection methods, offering a valuable tool for addressing critical environmental challenges.