Advanced Deep Learning Models for Satellite Image Analysis and Classification

Satellite images are primary data in weather prediction modeling. Large quantities of annotated data with a variety of features are needed for training in the deep learningbased technique, which is a promising option for autonomous image processing. With applications in environmental monitoring, disaster relief, and land use studies, accurately classifying satellite images is an essential task. This research provides a Convolutional Neural Network (CNN)-based deep learning solution to satellite image categorization. We create and assess our model using the Sentinel-2 satellite photos from the Eurosat dataset. The classification accuracy of our CNN model was 71 %. However, an EfficientNet model that had been trained beforehand had a far higher accuracy of 96.77 %. The results demonstrate the effectiveness of advanced deep learning algorithms in enhancing the degree of precision of satellite picture classification, and provide a powerful tool for using satellite data in a range of applications.