CodeProject AI on Github

[amdscooter]

Below is the list of supported GPU's by generation for the ALPR module

Screenshot 2024-12-16 174608.jpg

thanks mike!

[TimG]

Perfect. Thank you.

[amdscooter]

Perfect. Thank you.

I guess no CUDA 12 unless you have a $40K + GPU in your system. Not going to be a problem for me

[tripp]

Perfect. Thank you.

I guess no CUDA 12 unless you have a $40K + GPU in your system. Not going to be a problem for me

I'm running Cuda 12.6.85 with an RTX 4090 with zero issues. Latency is in the 60ms-100ms range running YOLOv8 only.

[GI0UTE]

Below is the list of supported GPU's by generation for the ALPR module

Screenshot 2024-12-16 174608.jpg

Great info thanks. Would trying to use a too high a version of cuda cause issues. I am using a Tesla p4 but have installed cuda 12 version of codeProjectAI GPU, in a docker under unraid. I am new to codeProjectAI, so just learning, but have it working ok, but alpr is very hit or miss, despite the plate being very clear in the images/ video.

Maybe I should try the cuda 11 version? Thanks des.

[TimG]

For me, ALPR didn't work at all with the wrong CUDA. Do you have any issues ?

Hit or miss is probably due to the camera, the placement or the camera settings. My present camera doesn't work in the dark but I won't update that camera yet.

[GI0UTE]

Thanks for the reply. I do have the AI working ok on other cameras including this one. It's picking up cars and people, even the cat, but the number plate reading is very hit and miss. I have great images of the plates from the camera, day and night, it's just it seems inconsistent with reading them. They are very clear and easily read by a human from the image.

Anyhow, I will start a separate thread tomorrow about my issues as they are not directly related to this thread. I am sure it's just my settings not being correct in blue iris. Its a very capable piece of software, but I don't have the base knowledge needed, and due to updates, older threads don't seem to follow what I see on screen.

Thanks