NVIDIA DeepStream SDK

DeepStream is NVIDIA's toolkit for building GPU-accelerated pipelines that run vision AI models on live video and other sensor streams. Developers use it for things like traffic monitoring, retail analytics and factory inspection, from Jetson devices up to cloud servers.

Pipelines are assembled from GStreamer plugins that handle decoding, batching, inference with TensorRT, object tracking and output, so many camera streams can be processed on one GPU. The same application can be deployed at the edge, on-premises or in the cloud.

The GitHub monorepo holds the SDK together with reference applications that show complete pipelines and serve as starting points. It fits teams already working in the NVIDIA stack who need low-latency analytics rather than offline batch processing.