SiteWatch Edge

The need

Analyze workplace imagery and queue incident metadata where connectivity is unreliable.

Why it was developed

Run detection, local warnings and delayed synchronization on the local machine.

How to use it

1. Create a Python environment; install requirements-lock.txt and the local project as instructed by the repository README.
2. Fetch the licensed detector models once. Start sitewatch-web with the configured ONNX model on loopback port 7860.
3. Test images/video and rules locally; configure AWS synchronization separately if required.

How it can save time or money

Processes video locally and queues events, avoiding a mandatory cloud-inference request for every frame and reducing manual evidence assembly.

These are potential benefits, not measured savings. Include setup, maintenance, review and local operating costs when comparing your own workflow.

Current scope & limits

Advisory application; field performance needs separate verification. The selected model imposes AGPL requirements. Optional AWS can introduce charges. Existing source link was not accessible in the connected account inventory.