Optimizing Astronomical Image Classification with EfficientNetB0: A Comprehensive Study of Celestial Object Identification

Advancing the knowledge of the universe, helping to discover and identify celestial bodies, and enabling study in astrophysics and cosmology depend on astronomical picture classification, which also supports Cataloging large volumes of data produced by telescopes depends on proper classification of astronomical images, which also helps astronomers to effectively spot trends and anomalies that can lead to new discoveries, therefore supporting sustainable innovation. In this work we categorize a hierarchically structured astronomical image dataset using EfficientNetB0, a deep learning model renowned for its accuracy and efficiency. There are 4,099 high-quality photos in the dataset; 3,279 for training and 820 for testing split. Using its capacity to scale both resolution and depth, the model was modified to execute macro- and fine-grained classification challenges. We obtained a classification accuracy of 89% following model training, therefore proving its efficiency in managing challenging astronomical data. This result emphasizes the possibilities of using modern convolutional neural networks for dependable and efficient classification in astronomical research, so supporting the development of automated systems for celestial object detection, which is absolutely vital for scientific advancement, quality education, and the encouragement of infrastructure development.