Advancements in astronomical research heavily rely on the accurate classification of celestial objects, driving the need for robust machine learning solutions. In this study, we present an extensive approach for classifying celestial objects using Convolutional Neural Networks (CNNs), leveraging the Residual Network (ResNet) architecture within the PyTorch framework. We commence with meticulous data collection and preprocessing, curating a diverse dataset of celestial object images. Subsequently, we employ ResNet-based CNN models and PyTorch’s powerful functionalities to train and fine-tune our classifiers for superior performance. Our methodology encompasses rigorous testing and evaluation, benchmarking the CNN classifiers against diverse datasets to assess accuracy and scalability. Furthermore, we develop an intuitive web interface utilizing Flask, enabling seamless interaction with the CNN classifiers. Our study underscores the effectiveness of employing ResNet-based CNNs in celestial object classification tasks, empowered by the versatility and efficiency of the PyTorch framework. Through this approach, astronomers gain a robust tool for precise analysis and understanding of celestial phenomena, facilitating advancements in astronomical research.