Convolutional Neural Networks (CNNs) are a class of deep neural networks designed to process grid-like data, most commonly images. They are inspired by the structure of the visual cortex and are particularly effective at learning spatial hierarchies of features — from simple edges in early layers to complex shapes in deeper layers.

The core component of a CNN is the convolutional layer, which uses small filters (kernels) that slide across the input to extract local features. Unlike fully connected layers, convolutional layers dramatically reduce the number of parameters while preserving spatial relationships.

CNN architectures typically consist of:

• Convolutional layers for feature extraction,
• Pooling layers (e.g., max pooling) to downsample feature maps and introduce translation invariance,
• Fully connected layers at the end for classification or regression tasks.

CNNs are widely used in:

• Image classification (e.g., ResNet, VGG),
• Object detection and segmentation (e.g., YOLO, Mask R-CNN),
• Medical imaging, video analysis, and more.

Key advantages of CNNs include weight sharing, local connectivity, and parameter efficiency, making them the dominant architecture for computer vision tasks.