A Feature Augmented Deep Learning Based Model for Satellite Object Detection

Machine Learning (ML) and Deep Learning (DL) have revolutionized various fields, and satellite object detection is no exception. By enabling the automatic identification and classification of objects in satellite imagery, such techniques have significantly improved the efficiency and accuracy of remote sensing applications. From monitoring environmental changes to urban planning and disaster management, machine learning enabled satellite object detection plays a critical role in modern geospatial analysis. This paper proposes an approach comprising of segmentation, feature extraction and neural network based pattern recognition for satellite object detection. The deconvolution algorithm has been employed for noise removal followed by maximal gradient based segmentation prior to pattern recognition using a deep neural network structure. The performance of the proposed approach is the classification accuracy which is found to be higher compared to exiting work in the domain.