Explore the L-detection preview

This guide covers the development preview, not production deployment. You need a development package with its scripts; a public download source will be added later.

1. Understand the components

• Client: the interface for camera viewing, regions and event management.
• Site: processes cameras and AI at a deployment location.
• Central: manages configuration and connects sites.

Closing the client does not automatically stop Site and Central services.

2. Prepare the environment

You need Windows, an NVIDIA GPU compatible with the development package’s CUDA runtime, an H.264 RTSP camera and a suitable ONNX model.

If you have been provided source code with the scripts, build from the project directory:

./scripts/build.ps1

Run runtime and media-tool installation scripts only according to the documentation supplied with that version. A complete end-user installer has not been confirmed.

3. Start and check a camera

For the development package with scripts, start with:

./scripts/start-dev.ps1

Connect the client to the correct Central/Site services. Configure your RTSP camera, verify its live image and then set up regions and rules. Check actual video rather than relying only on a service status.

4. Stop and understand the limits

To stop the development services:

./scripts/stop-dev.ps1

Long-running production operation and 32–64 camera loads have not been accepted. Hide passwords and camera credentials in URLs, screenshots and logs when reporting a problem.