vss-deploy-dense-captioning

Name: vss-deploy-dense-captioning
Availability: InStock
Author: rblake2320

Deploy and test a standalone RT-VLM dense-captioning REST API with Docker.

Updated May 29, 2026

Search Tags:#deployment #docker #kafka #api-testing #standalone #rt-vlm #dense-captioning

Ready to run

npx skills add https://github.com/rblake2320/vigil --skill vss-deploy-dense-captioning-rblake2320

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: vss-deploy-dense-captioning
Source: https://github.com/rblake2320/vigil/tree/main/.claude/skills/vss-deploy-dense-captioning
Command: npx skills add https://github.com/rblake2320/vigil --skill vss-deploy-dense-captioning-rblake2320

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Deploys and tests the standalone RT-VLM dense-captioning service and its REST API (uploads, captions, streams, chat-completions, Kafka) without requiring a full VSS profile.

Core Features & Use Cases

Standalone deployment: stand up the RT-VLM service using a local docker-compose copy and validate health, model loading, and API surface.
API surface validation: exercise endpoints such as /v1/files, /v1/generate_captions, /v1/streams, /v1/chat/completions, and Kafka topics (when enabled).
Testing and experimentation: preflight RTSP streams with provided sample URLs, run smoke tests, and verify Kafka wiring for end-to-end validation.

Quick Start

Follow the standalone RT-VLM deployment flow: copy the deploy/docker/services/rtvi/rtvi-vlm/rtvi-vlm-docker-compose.yml into a writable directory, derive the image tag from the copy, strip the standalone-only dangling depends_on block, create a .env with required values, pull the exact image, bring up the service, and run a quick health check.

Frequently Asked Questions about vss-deploy-dense-captioning

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

FAQPage Schema

How do I deploy a standalone RT-VLM dense captioning service without a full VSS profile?▼

To deploy a standalone RT-VLM dense captioning service, copy the docker-compose file to a writable directory, strip the dangling depends_on block, create a .env file, pull the exact image, and bring up the service for rapid readiness checks.

Do I need NVIDIA Container Toolkit and GPU visibility to run RT-VLM dense captioning?▼

Yes, running the RT-VLM dense captioning service requires Docker with the NVIDIA Container Toolkit installed, proper GPU visibility, and access to NGC keys for pulling required container images.

Can I test RT-VLM REST API endpoints for file uploads and chat completions locally?▼

Yes, you can test the standalone RT-VLM REST API locally by exercising endpoints such as /v1/files, /v1/generate_captions, /v1/streams, and /v1/chat/completions for API surface validation.

How do I validate Kafka wiring and RTSP streams for a dense captioning deployment?▼

You can validate Kafka wiring by enabling Kafka outputs in the .env configuration, and preflight RTSP streams by running precheck commands against provided sample URLs before starting the dense captioning service.

What is a standalone RT-VLM deployment used for in dense captioning workflows?▼

A standalone RT-VLM deployment is used for rapid readiness checks, API surface validation, model loading verification, and optional Kafka wiring without deploying a full VSS profile.

Why should I strip the depends_on block from the docker-compose file for standalone RT-VLM testing?▼

Stripping the dangling depends_on block from the local docker-compose copy allows the RT-VLM service to run independently for standalone testing without requiring the full VSS dependency chain to be active.