Release Notes: Live Video Captioning#

Version 2026.1.0#

June 17, 2026

New

  • Live Video Captioning RAG as an additional feature in Live Video Captioning, enabling Retrieval-Augmented Generation (RAG) chat.

  • Integration of model-download service to simplify downloading before preparing the model.

  • Added support for using the host machine’s camera as an input source.

  • Dynamically selects the default pipeline based on detected hardware such as GPU.

  • Simplify the setup process of the application.

  • Deployment with Helm chart.

  • Enhancements in UI/UX for the Alert mode.

  • Documentation updates.

Known Issues

  • The sample application is not validated either on the Standalone or Developer Node versions of Edge Microvisor Toolkit.

Version 1.0.0#

April 01, 2026

Live Video Captioning is a new sample application, using DLStreamer and VLMs to produce captions on live camera feed. It enables you to configure the VLM model used, prompt, frame selection, the rate at which the captions are generated, frame resolution, and more. It also presents usage metrics of CPU, iGPU, and memory. A rich UI displays all the configuration options, live camera feed, and the generated text.

In addition, the sample application may generate alerts with custom prompts, with customizable generation delay. Docker-based deployment is supported currently.

New

  • Docker Compose stack integrating DLStreamer pipeline server, WebRTC signaling (mediamtx), TURN (coturn), and FastAPI dashboard.

  • Multi-model discovery from ov_models/.

  • Live captions via SSE and live metrics via WebSockets.

  • Support for the live metrics service.

  • A rich graphical user interface.

Known Issues

  • Helm support is not available in this version.

Upgrade Notes

  • If you change .env values (ports, HOST_IP, model paths), restart the stack: docker compose down && docker compose up -d.

On this page

This Page