Video analytics

The classic toolkit

Motion detection, line crossing, zone intrusion, object counting, left-luggage rules. All of it works on pixels and shapes: the system knows something moved, roughly where and roughly how big, and that's the extent of its vocabulary. It's useful, mature and famously fond of false alarms from headlights and foxes. Motion detection still has its place; it just can't tell you what it saw.

What vision models changed

The newer generation describes scenes in language instead of firing on pixel changes. A frame stops being “motion in zone 3” and becomes “a man in a grey hoodie putting a bottle in his coat”. Once that description exists as text, analysis becomes reading, and finding a moment becomes natural language video search. Svid works this way: a vision model describes what's happening in every frame of an existing CCTV feed, with no training on your site and no tagging, and search results come back in seconds however many hours sit behind them.

Video analytics vs video intelligence

Analytics is the established industry term and the narrower one, rooted in detection techniques. Video intelligence is the umbrella covering the shift from detecting movement to understanding events. If you're weighing up tools, the label matters less than the question each generation can answer: did something move, or what happened.