Multi camera AI security system

I am trying to run yolov4 in 6 cameras at the same time. However, I am only able to open 4, the 5th camera feed is empty.
How can I run 6 cameras in python? Thanks.

It looks like you are using Jetson Nano so moving topic to help it get better attention.

Did you check with gst-launch-1.0 command to confirm 6 cameras?

Just curiously I was wondering what you expectation is with 6 cameras and yolov4. Are you running a tiny model or something? Because the full sized model of yolov4 running on a nano will be able to process just over one frame per second, so even 4x cameras seems very ambitious.

However, if you want to try and catch intruders then you may be able to make this work somewhat. If that’s the case you may want to try install my yolo powered security project “StalkedByTheState”.

Normally on a nano though I recommend only processing 4x cameras realistically, but that’s running the full model. If you reduce the image size from 416 (Already a compromise for the nano) in steps of 32 from memory, you may very well be able to process 6x cameras in a meaningful manner for security, at the cost of some accuracy, meaning you will have more false positives but these can be mitigated.

The project is easy to install with just two commands installing the whole system.

Kim Hendrikse