Convolutional Neural Networks (CNNs) [19] transformed image recognition by learning patterns such as edges, textures, and shapes directly from raw pixels. Their layered structure, which combines convolution and pooling, enables efficient feature extraction and scales well to complex image data. CNNs remove the need for handcrafted features and learn from raw inputs in a data-driven way. This shift led to major gains in accuracy and robustness across visual recognition tasks.