Automatic Feature Extraction from Satellite Imagery for Remote Sensing

Abstract

Feature extraction from satellite images has steadily but progressively grown in recent years. Numerous feature extraction techniques have been created as a result of the increase. The loss of certain features is one of the various obstacles that must be overcome while selecting the best path for feature extraction. This study suggests utilizing a Convolutional Neural Network (CNN) approach to tackle these issues and extract attributes from satellite images supplied by an open source dataset called MLRSNet. Images with labels explaining each image's characteristics are displayed in the output. This makes it simpler to recognize and understand different components in satellite images.