This chapter starts by reinforcing the essential building blocks of Convolutional Neural Networks (CNNs). We assume familiarity with basic concepts but will briefly review components like convolutional layers, pooling operations (e.g., $max-pooling$, $average-pooling$), and activation functions (e.g., $ReLU(x) = max(0, x)$).

Building on this foundation, we examine the progression of CNN architectures. We trace the development from early influential models towards contemporary designs such as ResNet, Inception, DenseNet, and EfficientNet. Key innovations will be analyzed, including:

• Residual connections, often represented as $y = \mathcal{F}(x) + x$
• Network-in-Network principles
• Dense connectivity patterns
• Compound scaling strategies

You will gain an understanding of the design choices behind these architectures and the trade-offs involved, considering factors like computational cost and parameter efficiency. The chapter concludes with practical guidance and exercises on implementing these models using standard deep learning frameworks.