One-Sentence Definition
Computer vision is the field of AI that enables machines to interpret and extract meaningful information from images, video, and other visual inputs.
How It Works
Computer vision gives software the ability to "see." At its simplest, a computer vision system takes pixel data as input and outputs a structured interpretation: this image contains a dog, this video frame shows a pedestrian crossing the street, this medical scan has an anomaly in the upper-left quadrant.
Early approaches relied on hand-engineered features -- edge detectors, color histograms, template matching. The deep-learning revolution changed everything. Convolutional neural networks (CNNs) like ResNet learn to extract visual features automatically from labeled datasets. A CNN trained on ImageNet's 14 million images can classify objects across 1,000 categories with superhuman accuracy.
The field has expanded well beyond classification. Object detection (YOLO, Faster R-CNN) identifies and locates multiple objects in a single image. Semantic segmentation labels every pixel with a category. Pose estimation tracks human body joints in real time. Optical character recognition (OCR) reads text from photos and documents. More recently, vision transformers (ViTs) have shown that the same attention-based architecture powering LLMs also works for images, and multimodal models like GPT-4o and Claude can now reason about images and text together in a single conversation.
Why It Matters
Computer vision runs inside autonomous vehicles (Tesla, Waymo), manufacturing quality control systems, medical imaging tools that detect cancers and retinal diseases, satellite analytics platforms that monitor deforestation, and the face-unlock feature on your phone. The global computer vision market is projected to exceed $40 billion by 2027.
The convergence of vision and language models is the current frontier. Multimodal AI systems can describe images, answer questions about charts, and extract data from photographs -- capabilities that are reshaping document processing, accessibility tools, and content moderation.
Key Takeaway
Computer vision enables machines to understand visual data using deep neural networks, and its fusion with language models is creating a new class of multimodal AI systems.
Part of the AI Weekly Glossary.