Convolutional Neural Network (CNN) is a type of neural network designed to process images, using layers that detect visual features like edges, shapes, and textures. CNNs powered the breakthroughs in computer vision.
What it means in plain English
A CNN is built specifically for grid-like data such as images. Its “convolutional” layers scan across an image looking for small patterns — an edge here, a curve there — and deeper layers combine these into more complex features, from shapes up to whole objects. This design mirrors how visual recognition builds up from simple to complex, and it made machines dramatically better at understanding images.
CNNs were the workhorse of computer vision for years, though transformers are now also widely used for images.
A simple example
A CNN used for medical imaging scans an X-ray, its early layers spotting edges and textures and its later layers recognising patterns associated with a particular condition — flagging areas for a doctor to review.
Why it matters
CNNs drove the deep-learning revolution in computer vision, enabling reliable image recognition, face detection, and more. They remain a foundational architecture and a key part of how AI learned to see.
Related terms
- Computer Vision — the field CNNs transformed.
- Neural Network — the broader family CNNs belong to.
- Deep Learning — the approach CNNs helped advance.
Frequently asked questions
What are CNNs best at?
Convolutional neural networks (CNNs) excel at image and spatial data, powering image classification, object detection, and other computer-vision tasks by detecting local patterns like edges and shapes.
How do CNNs work at a high level?
They apply filters that scan across an image to detect features, building from simple patterns (edges) in early layers to complex ones (objects) in deeper layers.
