Hi all, I've just discovered this site. I have an open source project that would be ideally suited for helping in human/wildlife conflict situations. The project is actually an outgrowth from a home security system that I added state of the art computer vision object detection to in 2019. (But the project was used with other sensors from 2011).



The website is the following:



 

and uses the latest object detection models. Currently yolov7, yolov6 and others will full size model support on SBCs.

In July I will making a new release to support the new Seeed Studio recomputers that uses the new Jetson Orin boards.

This project is also unique in that it supports scanning via multiple models and has a simple web socket wrapper around the model detection so it's easy to add new models to the system.

By multi-model. I mean very flexibly. You could use one model to detect poachers and another model to detect specialized animal detection models all with in the same image and even different models in different parts of the images. The models can even be cascaded. i.e. to double check a human detection with an additional model. Or even the detection of a certain human after first detecting a person. The combinations are endless.

And it projects video alerting and very flexible state control via a webpages state machine that connects to user defined buttons. And also some local home automation/device control as well.

Let me know if this is of interest to any of your programs.

Kim Hendrikse