Computer vision — Glossary

Computer vision builds systems that interpret visual data for tasks including classification, detection, segmentation, tracking, and OCR. Deep neural networks currently dominate many benchmarks, often pretrained on large image datasets and fine-tuned for specific domains.

Deployment success depends on data that reflects real lighting, cameras, and edge cases. Metrics like accuracy or mean average precision should be paired with operational testing, especially where errors affect safety or rights.