I have an ESXi hypervisor running Windows 10 for BI for many years now and started using CPAI for detection a couple of years ago using the custom ipcam models.
Detection works fairly well but the detection times are really bad (500+ ms) so I am experimenting with using a Coral USB I had access to hooked up to a Raspberry Pi with CPAI running in a Docker container.
It works but I'm having a lot of issues. Using YOLO v8 I observed that the detection process seems to hang and become unresponsive after a few hours. Using YOLO v5 (medium) models it has been working, kind of, but lots of inaccurate detections (detecting a car driving by with 65% accuracy as "person).
Does anyone have this working well including night time IR identification of persons?
If so please let me know what your settings are, this is feeling kind of like a lost cause.
Detection works fairly well but the detection times are really bad (500+ ms) so I am experimenting with using a Coral USB I had access to hooked up to a Raspberry Pi with CPAI running in a Docker container.
It works but I'm having a lot of issues. Using YOLO v8 I observed that the detection process seems to hang and become unresponsive after a few hours. Using YOLO v5 (medium) models it has been working, kind of, but lots of inaccurate detections (detecting a car driving by with 65% accuracy as "person).
Does anyone have this working well including night time IR identification of persons?
If so please let me know what your settings are, this is feeling kind of like a lost cause.