← All glossary terms Glossary

Edge AI (Edge artificial intelligence)

Edge AI means running artificial intelligence models on the device that captures the data, such as a camera or a roadside processing unit, instead of sending raw video to a central server or cloud for analysis.

What it means

An edge system sends events, evidence images and short clips instead of continuous high-resolution video. That cuts network bandwidth and makes the response time independent of the link to the data centre. It also keeps working when the link drops: events are stored locally and sent when the connection returns. The trade-off is that model updates must be pushed to many devices, and each device needs enough processing power.

How it applies in India

Long highway corridors and many city junctions have limited or unreliable connectivity. With edge processing, a camera on a remote stretch can still detect incidents and violations and keep its evidence. Server-based designs suit sites with a strong fibre network and a need for heavy, central analysis. Many projects use both. For a buyer's view of the choice, see the guide on edge vs server ITMS architecture. Stellarview runs its analytics at the edge, with vision language model search on the Recon camera's processor.

Questions about Edge AI on your project?

Request a demo All glossary terms