CP.AI not making any object identifications

[subterminal]

I just setup a new BI server w/ CP.AI to replace my old one. It's my first run at AI. I've got it all setup, but CP.AI isn't successfully identifying any objects.

I've played around with the custom models, initially only having ipcam-combined as an option, but gradually adding everything. I've also lowered confidence and increased the amount of images sent to CP. Nothing seems to work.

Additionally, it's not saving images of faces, however, I assume that this is a product of the object recognition not working. If it can't identify that something is a face, it isn't gonna save it.

Attached are my setting configs.

 

[Vettester]

I just setup a new BI server w/ CP.AI to replace my old one. It's my first run at AI. I've got it all setup, but CP.AI isn't successfully identifying any objects.

I've played around with the custom models, initially only having ipcam-combined as an option, but gradually adding everything. I've also lowered confidence and increased the amount of images sent to CP. Nothing seems to work.

Additionally, it's not saving images of faces, however, I assume that this is a product of the object recognition not working. If it can't identify that something is a face, it isn't gonna save it.

Attached are my setting configs.

Have you tried CPAI explorer to make sure AI is working with your custom models?

 

[subterminal]

Have you tried CPAI explorer to make sure AI is working with your custom models?

No, how does that work? I just upload random photos? I don't even see that option in my dashboard. My server doesn't have WAN connectivity, fwiw, but I can certain add that if needed to install a module or something.

 

[MikeLud1]

No, how does that work? I just upload random photos? I don't even see that option in my dashboard. My server doesn't have WAN connectivity, fwiw, but I can certain add that if needed to install a module or something.

Click on CodeProject.AI Explorer to access Explorer

 

[subterminal]

Hrm... It doesn't seem to detect anything. I've tried various photos of people and dogs. Not a single one has worked.

 

[subterminal]

Finally got it working.

I restored WAN connectivity, then updated the modules. I also downgraded CUDA toolkit to 11.7. I'm not sure which fixed it, but I'm assuming it was the latter. Didn't see any documentation that specified a CUDA version, but I did notice that YOLOv5 module description mentioned CUDA 11.7