Hi all, I've successfully integrated deepfaune into my Video alerting full-features security system StalkedByTheState. The yellow box around the image represents the zone of interest that I defined and the blue crosses represents matching the category of interest in the zone, in this case wolves. This will generate a new event into the system that can result in multiple actions, such as generating a video alert or flashing flood lights or other actions to try and deter the wolf from coming further if this was outside a farm for example.

StalkedByTheState is an advanced security system that an monitor multiple cameras and trigger video alerts or other actions when it gets a match. The AI computer vision is abstracted behind a network interface which made it a simple task to integrate deepfaune.



Deepfaune first uses megadetector to detect the presence of either a person, an animal or a car. I've joined this two steps and present the API in the form of a traditional yolo API so it appears to detect and classify in one step.



This announcement is a bit of a pre-release really. The code is working but I still have to release this in a usable for on github.



StalkedByTheState is designed to be able to run locally on NVidia's AI on the edge Jetson servers.



It's now possible to define a json file defining what animals you wish to detect and generate video alerts for them. Or even invoke local control automatically from the Jetson server directly. There is support for local relay control, inputs and outputs.



I still have to complete the installation script to support the new Jetson Orin series computers, it's working on them, but I've manually installed it at the moment. When this is complete I'll make a new announcement.