Video Search and Summarization
Reference architecture for AI agents that search and summarize video
What it is
A GPU-based reference stack from NVIDIA for building agents that work over live camera streams and recorded footage. It combines vision language models, embeddings, object detection and tracking, and RAG to answer questions about video, search archives in plain language, summarize long recordings, and check live alerts before they fire, exposing its tools over MCP. Developers and ML engineers building video analytics for warehouses, smart spaces, or procedure checks deploy and customize its microservices and web UI.