Unblink is a free and open-source smart camera monitoring app that uses AI to watch your video feeds in real time, right on your device. It can spot people, objects, and even understand what’s happening (like “someone entering the door” or “a package arriving”).
You can search past footage using keywords (“dog,” “car,” “movement”) and stream videos with lightning speed, all while keeping your data private.
It upgrades your existing setup with real-time object detection and smart search, letting you find specific moments instantly, like a delivery or a pet’s mischief, without scrubbing through hours of footage.
It uses D-FINE; a State-of-the-art object detection for identifying and tracking objects in real-timeSmolVLM2 and Moondream 3 which is a Vision-language models for understanding context and answering questions about camera feeds
Best of all, it is self-hosted, meaning you get sub-second streaming and advanced monitoring while keeping your data 100% private and free from subscription fees. Make your cameras smart today!
What can it do?
- Contextual understanding
- Object detection
- Intelligent search across your video feeds.
- Sub-second video streaming
Features
- Real-Time AI Vision
Run advanced AI models on your camera streams instantly — no cloud dependency. - Contextual Understanding
AI interprets what’s happening in live feeds using vision-language models (VLMs). - Object Detection & Tracking
Powered by D-FINE model — detects and tracks people, vehicles, packages, and more in real time. - Smart Search Across Footage
Search past videos using natural language: “Show me when someone arrived at the front door” or “Find all clips with dogs.” - Sub-Second Video Streaming
Ultra-fast, low-latency playback for smooth monitoring. - Multi-Camera Dashboard
View and manage multiple cameras simultaneously with real-time status indicators. - Support for Multiple Protocols
Connect RTSP, MJPEG, and other common camera streams — works with NVRs and IP cams. - Natural Language Interaction
Ask questions like “Is the garage door open?” or “Who was near the back gate?” — get answers instantly. - Webhook Alerts
Send real-time notifications via webhooks when events are detected (e.g., motion, object presence). - Role-Based Access Control
Securely manage user permissions — ideal for homes, offices, or shared spaces. - Self-Hosted & Privacy-First
All processing runs locally or on your server. Your data never leaves your network. - Optional Remote Engine (GPU-Accelerated)
AI inference runs in a separate engine (Python-based) that can be hosted remotely or on your own GPU server. - Lightweight Web App
Built with Bun + SolidJS — fast, modern, and responsive. - Future Features (Coming Soon):
- Video recording & playback
- Motion detection
- ONVIF support
- Automation rules
FAQs
Why is my CPU usage so high?
D-FINE object detection is resource-intensive. If you experience performance issues, you could consider disabling object detection from the Settings page. I would add some optimization to this soon.
Where is the code to run the models?
The model inference code is in a separate repository at https://github.com/tri2820/unblink-engine. This separation allows the AI models to run with GPU acceleration in Python, while keeping the app lightweight.
Currently I have the engine hosted on my GPU server that you can use (the client app automatically connects to it), so hosting the engine yourself is optional. If you need to, you can mofidy ENGINE_URL env var and the client app will connect there instead.
License
- AGPL-3.0 license
