Case Studies
Surveillance video system: pipeline optimization for zero-copy hardware acceleration
About the service
The client develops global hardware and software for video surveillance systems. Their solution uses AI to analyze video flows from multiple IP cameras, which are later aggregated and encoded for real-time streaming in their client web application.
Their encoding/streaming system was based on FFmpeg, but as they switched to GStreamer, they required our help to perform a deep analysis and suggest possible improvements to their pipelines. Our expertise in this framework allowed us to optimize them for NVIDIA® hardware acceleration.
What We Achieved
Deep analysis of pipelines and source code. Proposal of a different architecture to ensure zero-copy, hardware acceleration, and interconnection with the WebRTC streaming protocol.
