Image categorization has emerged as a key application of machine learning, enabling systems to classify images into predefined categories. Recent advancements in deep learning, particularly Convolutional Neural Networks (CNNs), have enhanced the ability to handle complex image classification tasks. This study aims to develop an efficient image categorization model by optimizing CNN architecture and training parameters. The primary objectives include improving classification accuracy, speed, and noise tolerance during the model training phase. We propose a CNN model that uses three convolutional layers, filters per layer, dense layer etc. The model is trained on different image formats (JPG, PNG) and evaluated based on key performance metrics: classification speed, accuracy, and noise tolerance. Results indicate that the proposed model achieves an accuracy of over 88%, outperforming existing methods in terms of both speed and robustness to irrelevant data. These findings highlight the effectiveness of the proposed CNN-based approach for accurate and efficient image categorization.