Self-hosted, local-only NVR and AI computer vision software for object, motion, and face detection
- Stars3.2k
- Forks395
- Open Issues62
MIT
- Python
- TypeScript
- JavaScript

About Viseron
Viseron is self-hosted, local-only NVR and AI computer vision software for keeping an eye on a home, office, or any other space. All video analysis runs on hardware you control, with nothing sent to a cloud service.
It connects to RTSP and IP cameras and layers several detectors on top: object detection, motion detection, face recognition, and license plate recognition. Detections can drive recording, so storage goes toward footage that matters rather than continuous streams.
Viseron can use hardware acceleration through Coral EdgeTPU, CUDA, and OpenCL to keep detection fast on modest machines. You run it as a Docker container and configure everything through a built-in web interface, with capabilities added through a component system.
Key features
- Object detection, motion detection, and face recognition
- License plate recognition for vehicles
- RTSP and IP camera support with detection-based recording
- Hardware acceleration via Coral EdgeTPU, CUDA, and OpenCL
- Built-in web interface with a component-based config
Details
- First released
- 2020
- Cameras
- RTSP · IP
- Detection
- Objects · Motion · Faces · Plates
- Hardware
- Coral · CUDA · OpenCL
- Deployment
- Self-hosted · Docker
- Language
- Python
