NVIDIA-AI-IOT/live-vlm-webui

Real-time Vision Language Model interaction via webcam - WebRTC-based web interface

What it solves

Live VLM WebUI provides a universal web interface for interacting with Vision Language Models (VLMs) in real-time. It solves the difficulty of streaming live video feeds (like webcams or IP cameras) into various VLM backends for instant AI-powered analysis, benchmarking, and performance testing across different hardware platforms.

How it works

The tool acts as a interface between a video source and a VLM backend. It captures a live stream via WebRTC or RTSP and sends frames to a configured VLM API (such as Ollama, vLLM, or NVIDIA NIM). The system uses asynchronous processing to ensure the video feed remains smooth while the AI processes frames in the background. It also includes built-in system monitoring to track GPU/CPU usage and inference latency.

Who it’s for

It is designed for developers, AI researchers, and engineers working with vision AI, robotics, and industrial automation who need a way to quickly test and benchmark different VLMs on various hardware, including NVIDIA Jetson, DGX, and standard PCs.

Highlights

  • Universal API Support: Compatible with OpenAI-compatible APIs, including Ollama, vLLM, SGLang, and NVIDIA NIM.
  • Real-time Monitoring: Integrated live tracking of GPU utilization, VRAM, CPU, and RAM with sparkline charts.
  • Multi-source Input: Supports WebRTC webcam streaming and RTSP IP cameras.
  • Cross-Platform: Works across x86_64 and ARM64 architectures, including specialized hardware like Jetson Orin and Thor.
  • Interactive Prompting: Includes a prompt editor with over 10 preset prompts for tasks like OCR and object detection.

Related

  • Project
    ParisNeo/lollms-webui

    A unified local web interface for accessing hundreds of LLMs and multimodal AI models for text, image, video, and music generation.

  • Project
    mlc-ai/web-llm

    A high-performance in-browser LLM inference engine that uses WebGPU to run language models locally without a server.

  • Project
    gokayfem/ComfyUI_VLM_nodes

    A production-oriented suite of ComfyUI nodes for vision-language models, providing structured detection, segmentation, and adaptive video reasoning with VRAM optimizations.

  • Project
    wandb/openui

    An open-source tool that lets you describe UI components in plain language and see them rendered live, with the ability to convert them to React, Svelte, or Web Components.

  • Project
    open-webui/open-webui

    A self-hosted, extensible AI platform that provides a unified interface for local and cloud-based models, featuring built-in RAG, agent creation, and enterprise-grade user management.