skills.rest
NVIDIA-AI-Blueprints
NVIDIA-AI-BlueprintsOFFICIAL
Explore all skills by this author →
WebsiteGitHub

video-search

Search stored videos by natural language queries for objects, actions, or events.

1.8k|370|Updated Oct 22, 2024
Search Tags:#automation#search#video#semantic#actions#video analysis#objects
One-click install
npx skills add https://github.com/NVIDIA-AI-Blueprints/video-search-and-summarization --skill video-search-nvidia-ai-blueprints
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: video-search
Source: https://github.com/NVIDIA-AI-Blueprints/video-search-and-summarization/tree/main/skills/video-search
Command: npx skills add https://github.com/NVIDIA-AI-Blueprints/video-search-and-summarization --skill video-search-nvidia-ai-blueprints

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Enables users to find specific events, objects, actions, or people within video recordings using natural language queries, simplifying video analysis workflows.

Core Features & Use Cases

  • Semantic Video Search: Uses embedded vector similarity to locate relevant video segments based on user queries.
  • Attribute and Action Filtering: Combines appearance attributes with activities or locations for precise retrieval.
  • Use Case: For example, locate all instances of forklifts in warehouse footage during a specific time frame.
  • Verification: Supports detailed inspection and screenshot download for result validation and review.

Quick Start

Use the video search skill to find all clips where a person wears a red jacket or a vehicle appears near the loading dock.

Frequently Asked Questions about video-search

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I search video archives for specific objects and actions using natural language?▼

Semantic video search uses embedded vector similarity to locate relevant video segments based on natural language queries. You can find specific events, objects, actions, or people by combining appearance attributes with activities or locations for precise retrieval.

Can I filter video search results by time, attributes, and camera metadata?▼

Yes, video search supports detailed filtering by appearance attributes, time, and camera metadata. This allows you to combine appearance attributes with activities or locations, such as locating all forklifts in warehouse footage during a specific time frame.

Do I need pre-ingested videos and a deployed VSS search profile to perform semantic video search?▼

Yes, semantic video search requires pre-ingested videos and a deployed VSS search profile with Elasticsearch and Cosmos Embed1. These components enable the embedded vector similarity processing needed to locate relevant video segments.

What is the best way to verify video search results and inspect located events?▼

Video search supports detailed inspection and screenshot download for result validation. After locating specified objects, actions, or events using natural language queries, you can verify the identified video segments through screenshot inspection and review.

How does semantic video search handle complex queries with multiple attributes?▼

Semantic video search uses embedded vector similarity to interpret natural language queries combining multiple attributes. It processes requests combining appearance attributes with activities or locations, such as finding clips where a person wears a red jacket near a loading dock.

Related Skills

Explore More Content & Communication Skills→
NVIDIANVIDIA
49

vss-search-archive

Search recorded video archives for events, objects, actions, and people using natural-language queries.

Official
Intermediate
ahostbrahostbr

ls-youtube-transcript

Extract YouTube subtitles into timestamped markdown transcripts via yt-dlp.

Community
Intermediate
kevinsweetkevinsweet

vector-hybrid-search

Plan and implement end-to-end vector and hybrid search in Elasticsearch.

Community
Advanced
steipetesteipete
385.7k

qmd

Index local files and search them with BM25, vector, and hybrid queries.

Community
Advanced