I used to use the ipcamera binding + dahua IVS to generate an event when a person crossed a border. But I have since switched to just relying on Frigate’s detection, which it sends via mqtt, so I no longer use the ipcamera binding now.
I also used to trigger an LPR process manually but stopped doing it and zoomed the camera out and use it as a general camera instead of special LPR. It was more for fun and not for specific use.
When a person or car triggers the camera (via Frigate), a rule would instruct my openhab page to show the corresponding camera feed on a wall mounted monitor in the kitchen.
ONVIF is basically just a discovery API, as it lets openHAB to discover the RTSP stream that the camera has. It also allows the motion and other events to be seen by openhab. So an ONVIF camera has the easiest setup as the events are then seen inside of openHAB so you can then trigger things to happen like turn lights on/off and send notifications. The other ways of integrating cameras are much more fiddly to setup and can be time consuming. I am not a fan of MQTT purely because it is one extra service that has to be installed and setup creating complexity for new users. A lot of the camera brands have rest API;s that openHAB uses without the end user needing to open up or learn any of the API.
Are you an engineer at imou wanting to improve the way the cameras can be added to openHAB? If so, start with ONVIF first as that will get the cameras working in multiple home automation platforms and software like Frigate and Blue Iris etc…
Frigate has the benefit of being able to use hardware acceleration for image detection. I use PCI Coral accelerators with frigate. This does mean you likely will have a separate X86 computer with a PCI slot for your camera system (probably a good idea anyways).
