Deep Learning-based Classification of Geographical Land Structure using Satellite Images

Detecting the generic land structure in satellite images is typically done manually, requiring significant time and effort. Thus, the project proposes an approach to reduce human efforts, reducing the cost and time for identifying the land structure. In this project, a deep learning-based approach was developed to automate the process of classifying geographical land structures. The study compared the performance of three architectures, namely CNN, ResNet-50, and Inception-v3. The performance of the proposed model, CNN has achieved an accuracy of 94.8%. The results highlight the potential of deep learning models in scene understanding and their significance in efficiently identifying and categorizing land structures from satellite imagery.