Celestial Insights: An Innovative Convolutional Neural Network for High-Resolution Satellite Remote Sensing Image Classification

The utilization of satellite remote sensing has become an essential instrument for monitoring and understanding the ever-changing landscapes of the Earth. Accurate categorization of satellite imagery plays a crucial role in fully leveraging its diverse range of applications in the fields of environment and geography. This study classifies satellite images into four unique categories: cloudy, water, desert, and green region. This research introduces a novel Convolutional Neural Network architecture specifically designed to effectively identify and differentiate the distinct features of various terrains in satellite data. To assess the robustness of our methodology, the performance measures of our model are compared with those of the highly regarded VGG16 model. The Satellite Remote Sensing Image dataset has been utilized for the training and testing of both models. This dataset contains 5631 images belonging to four classes. The findings are highly persuasive: the customized convolutional neural network model demonstrated a remarkable accuracy rate of 98%, surpassing the VGG16 model's accuracy of 92%. This highlights the potential advantages of developing tailored models for distinct categorization tasks in the field of satellite remote sensing. This study contributes to the advancement of approaches used in satellite image classification and serves as a catalyst for future research in the development of customized models for complex classification challenges.