Convolutional neural networks (CNNs) have revolutionized the field of image classification by automating feature extraction through convolutional and pooling layers. This chapter provides a comprehensive overview of CNNs, focusing on their core components, training processes, and applications in image classification. The chapter begins by detailing the fundamental building blocks of CNNs, including the convolution layer, which learns local features through convolutional kernels; the pooling layer, which reduces dimensionality and computational overhead; and the fully connected layer, which compresses spatial features for final classification. Activation functions such as sigmoid, ReLU, and tanh are also discussed, highlighting their role in introducing nonlinearity into the network. The training process of CNNs is explored, covering loss functions, regularization techniques, and the Softmax classification layer, which outputs probability distributions for multi-class classification. The chapter then delves into prominent CNN architectures, including AlexNet, VGGNet, ResNet, and DenseNet, emphasizing their contributions to improving classification accuracy through innovations like residual connections and deeper network structures. Finally, the chapter extends the discussion to image segmentation, introducing frameworks such as FCN, U-Net, DeepLab, and PSPNet, which leverage CNNs for pixel-level classification tasks. By providing a thorough understanding of CNNs and their applications, this chapter serves as a foundational resource for researchers and practitioners in the field of deep learning.