BigHugger
GH Repository · NVIDIA-AI-Blueprints

video-search-and-summarization

NVIDIA AI Blueprint for video search and summarization (VSS) is a GPU-accelerated reference architecture for building video analytics agents with real-time verified alerts, visual Q&A, and automated reporting. The VSS Blueprint uses vision language models (VLMs) such as NVIDIA Cosmos, LLMs such as NVIDIA Nemotron, RAG, and NVIDIA NIMs.

stars
1,867
30-day movement
+134/day
Related entries
61
Connections
3
video-understandingvideo-summarizationPythonvision-agentvideo-agentreal-time-video-analyticsnatural-language-searchvideo-analyticsvideo-searchvlmlong-video-understandingmodel-context-protocolvision-language-modelraggenerative-aivideo-ragskillscomputer-visionmultimodal-ainvidia-nimretrieval-augmented-generation

An NVIDIA AI Blueprint providing a GPU-accelerated reference architecture for building video analytics agents. It combines vision language models (NVIDIA Cosmos), LLMs (Nemotron), RAG, and NVIDIA NIMs to enable video search, summarization, visual Q&A, real-time verified alerts, and automated reporting.

It gives you a reference implementation for video RAG and natural-language video search built on NVIDIA's model stack, so you don't assemble the pipeline from scratch.

Use it to

  • Build a video search agent using natural language queries
  • Generate summaries of long videos
  • Answer visual questions about video content
  • Set up real-time alerts on live video streams
  • Automate video reporting workflows

For Developers building GPU-based video analytics and video RAG applications

Role
rag
Language
Python
Licence
custom (licence file present)
Forks
393
Open issues
9
Last push
2026-09-17
Latest release
v1.0 · 2025-03-15
Skills shipped
1
topicsvideo-ragvideo-analyticsvision-language-modelnvidia-nimvideo-summarizationcomputer-vision