Classification of an image is a critical task in computer vision with many uses, including medical diagnostics and autonomous driving. Because of its capacity to automatically extract pertinent features from unprocessed pixel data, convolutional neural networks have become an extremely effective tool for image categorization. Using its deep architecture, we extract hierarchical features from input photos and offer a convolutional neural network model for image classification in this study. Using benchmark datasets like CIFAR-10, CIFAR-100, and ImageNet, we investigate various CNN architectures, including well-known ones like VGG, ResNet, and Inception. To improve our model’s capacity to work and its relevance, we also go over data pre-treatment approaches, regularization techniques, and hyper-parameter tuning tactics. Our technique achieves state-of-the-art accuracy on various benchmark datasets, demonstrating its usefulness in experimental outcomes. In addition to offering insights into creating effective and precise deep learning systems for practical uses, our work advances the field of picture categorization using convolutional neural network models.