Surface image Classification using CNN

Automated Classification of Planetary Surface Images Using Convolutional Neural Networks explains a system that automatically identifies different types of land areas from satellite images. The project uses deep learning models such as Inception-v3 and DenseNet121 with transfer learning to classify land types like residential, industrial, commercial, and agricultural areas. The models were trained using the UC-Merced Land Use dataset and achieved about 92% accuracy. The system also uses Grad-CAM heatmaps to show which parts of the image the model focuses on when making predictions. This approach helps improve applications such as urban planning, agriculture monitoring, and environmental management by providing accurate land-use classification.