unblink

zapdos-labsAGPL-3.0更新放缓

用视觉语言模型做摄像头监控，能理解画面内容而不只是检测移动，适合替代传统 NVR 的简单告警。

Camera monitoring with VLM

main-build发布于 2026-03-02 15:08 UTC1,518star171fork同步于 2026-11-07 07:31 UTC

视觉语言模型摄像头监控视频监控多模态理解 NVR

对比 GitHub camera computer-vision monitoring nvr vlm

怎么上手

由 README 提炼，执行前请确认

原文

unblink-node

硬件要求按 README 说明

README 未说明显存或 GPU 要求；节点提供 Linux x86_64、Linux ARM64、Windows x86_64、Windows ARM64 预编译包，可本地运行

下载：源码 zip·GitHub Releases（安装包）本站不镜像安装包

项目说明

由 README 自动翻译提炼

没人工点评？写下你的经验

用视觉语言模型做摄像头监控，本地节点转发视频流，支持自然语言检索录像

用视觉语言模型做摄像头监控，能理解画面内容而不只是检测移动，适合替代传统 NVR 的简单告警。

核心能力

用 Qwen3-VL 分析视频帧并生成摘要
用自然语言与摄像头画面聊天交互
用自然语言搜索已录制的视频帧
节点以私有代理方式转发 RTSP/MJPEG 本地摄像头
用 go2rtc 处理 RTSP 与 WebRTC 视频流

容易踩的坑

首次运行需在浏览器打开显示的 URL 完成节点授权；从源码安装需把 Go 的 bin 目录加入 PATH；开发环境需先复制 .env.example 为 .env 并配置

内容由模型根据 README 生成，可能不完整或不准确，以官方文档为准。看原文 README

用户点评

登录用户的真实使用记录，可按版本和硬件筛选

共 0 条

还没有人点评。第一个说出真实体验的人最有价值。

README 原文

未翻译

在 GitHub 打开

Unblink V2

AI-powered camera monitoring application with relay-node architecture and real-time video processing.

Vision Language Model: Qwen3-VL for frame analysis and summarization
Chat Interface: Natural language interaction with your camera feeds
Video Search: Search through recorded frames using natural language

Getting Started

The node runs locally and connects to the relay:

Without Go Toolchain

Download the prebuilt unblink-node archive for your platform from GitHub Releases:

Linux x86_64: unblink-node_linux_amd64.tar.gz
Linux ARM64: unblink-node_linux_arm64.tar.gz
Windows x86_64: unblink-node_windows_amd64.zip
Windows ARM64: unblink-node_windows_arm64.zip

Extract the archive and run:

unblink-node

On Windows, run:

.\unblink-node.exe

With Go Toolchain

Install from source:

go install github.com/zapdos-labs/unblink/cmd/unblink-node@main

# Make sure you have Go’s bin dir on PATH
unblink-node

# Or you can run
# $(go env GOPATH)/bin/unblink-node

Uninstall:

rm -f "$(go env GOBIN)/unblink-node"
# if GOBIN is empty, use:
rm -f "$(go env GOPATH)/bin/unblink-node"

On first run, authorize the node by opening the displayed URL in your browser.

Development

# Install dependencies
make install

# Run development environment (tmux with server + node + app)
make dev

# Typecheck code
make typecheck

# Generate protobuf code
make proto

Environment Variables

Copy .env.example to .env and configure:

cp .env.example .env

Architecture

Server: Publicly reachable server that manages nodes, handles authentication, and stores configuration
Node: Private proxy that runs in your network and forwards traffic to local cameras (RTSP, MJPEG)
Frontend: SolidJS with TypeScript, Vite, and Ark UI components
Database: PostgreSQL with pgx
Video: go2rtc for RTSP/WebRTC handling