There exist several applications for satellite images, such as environmental monitoring, law enforcement, and disaster response. For these applications, manual identification within such imagery is essential. Automation is necessary, though, because of the large geographic areas and scarcity of human resources for analysis. For such jobs, traditional object recognition algorithms frequently suffer from poor accuracy and dependability. CNN in particular, are part of deep learning and have shown promise in automating image interpretation tasks. In this study,. In particular, we offer a deep learning method that is intended to categorize dataset items and facilities into 60 distinct types. Our system comprises an ensemble of CNNs augmented Using additional neural networks that combine visual characteristics with satellite information. Python is used for implementation, and the deep learning libraries Keras and Tensor Flow are utilised. attaining an F1 score of 0.797 and an overall accuracy of 83%.