Computer Vision

What is computer vision?

You glance at a photo and instantly know it's a cat — no effort at all. Computer vision (CV) tries to give machines that same trick: read an image or video, recognize objects, understand the scene, track movement, and even grasp what's happening.

How does it work?

Images are just numbers
To a machine, a photo is a giant grid of pixels, each holding red, green and blue values. The job of computer vision is to pull "cat", "car" or "face" out of all those numbers.
From features to meaning
Early systems used hand-crafted features. Today most rely on convolutional neural networks (CNNs) that learn on their own. Layer by layer they move from edges and textures to eyes and ears, and finally assemble the idea of a "cat".

Where is it already working?

Self-driving cars
Cameras spot lane lines, pedestrians, vehicles and signs in real time so the car can "see" the road.
Security and surveillance
Systems flag anomalies in hours of footage and can recognize faces and license plates.
Medical imaging
AI helps doctors find lesions in CT scans and X-rays, catching problems earlier.
Your phone camera
Portrait mode, night shots and QR scanning all run on computer vision.

Why is it hard?

Things that are "obvious" to us are brutal for machines. Occlusion, changing light, odd angles — any of these can trip it up, which is why CV needs oceans of labeled data to train.

Bottom line: computer vision gives machines a pair of eyes that can actually make sense of the world.