Deep Convolutional Neural Network Architecture for Enhanced Image Classification

Abstract

The classification of natural images is a fundamental problem in computer vision and machine learning. In this work, we present a convolutional neural network (CNN) architecture optimized for the CIFAR-10 dataset, comprising 60,000 images across 10 categories. Our approach integrates batch normalization, dropout regularization, and adaptive learning rate scheduling to improve generalization performance. Experiments demonstrate that the proposed model achieves an accuracy of 94.2%, outperforming baseline models such as LeNet-5 and a standard VGG-like architecture. Comparative analysis, ablation studies, and statistical tests confirm the robustness and efficiency of the proposed method. The model and training code are made publicly available to support reproducibility.