Viseron

Viseron is a self-hosted, local network video recorder and computer vision application. It provides tools for surveillance and object detection, allowing users to monitor locations such as homes or offices using local hardware. The system operates as a local-only solution, ensuring that video data and analysis remain on the user's own infrastructure rather than being sent to external cloud services.

The software is deployed as a Docker container, which simplifies the installation process across different operating systems. Users manage the system and edit configuration files through a built-in web interface to set up video capture, define monitoring zones, and configure AI processing parameters. This web-based management allows for the coordination of multiple camera feeds from a single centralized dashboard.

Key features

• Object detection for identifying specific items in video feeds
• Motion detection to trigger recording or alerts
• Face recognition for identifying individuals
• Support for RTSP network video capture
• Integration with Google Coral Edge TPU for hardware acceleration
• CUDA support for GPU-accelerated processing
• Compatibility with YOLO and Darknet models
• TensorFlow integration for computer vision tasks
• License plate recognition for vehicle monitoring

Viseron utilizes a component-based architecture, where specific functionalities are enabled by adding individual components. These components allow the system to handle various tasks ranging from basic video recording to complex AI analysis. Users can discover and implement these modules through a dedicated Component Explorer. The software supports a wide range of hardware accelerators, including NVIDIA GPUs and Google Coral devices, to process high-resolution video streams efficiently. This architecture allows the system to scale based on the available compute resources of the host machine.

This application is designed for users who require a local surveillance system with integrated AI capabilities and hardware acceleration for real-time video analysis.