CodeProject.AI Version 2.5

Dudleydogg said:
it was working prior to this last code project update, like this Morning was all good. My GPU is only 2 years old but yes its a 3090

its 3/4 of inch longer than the 970 I can't win. going to see if its work unstacking the fans on the radiator.
Click to expand...
Well everything started as CPU, took few moments for the PC to get the new card recognized. then this happened
1708234069759.png
 
  • Like
Reactions: MikeLud1
MikeLud1 said:
Your GPU is on the old side and having the latest CUDA 12.2 installed is one of the issues that the module is not using CUDA.
Did you ever have CUDA working with the ALPR module? We might be able to get it to work if we use an older version on CUDA, cuDNN, and Paddle
Click to expand...
it was working prior to this last code project update, like this Morning was all good. My GPU is only 2 years old but yes its a 3090
Dudleydogg said:
Well everything started as CPU, took few moments for the PC to get the new card recognized. then this happened
View attachment 186792
Click to expand...
yep its doing its thing now just need some daylight to test out the LPR I get them good during the day. I notice the hardware must grab the cuDNN it can use because now my stats have changed again
1708234961315.png
 
MikeLud1 said:
Your GPU is on the old side and having the latest CUDA 12.2 installed is one of the issues that the module is not using CUDA.
Did you ever have CUDA working with the ALPR module? We might be able to get it to work if we use an older version on CUDA, cuDNN, and Paddle
Click to expand...
it was working prior to this last code project update, like this Morning was all good. My GPU is only 2 years old but yes its a 3090
MikeLud1 said:
Check Task Manager to see what is spiking you CPU
Click to expand...
I believe was just a glitch lot going on with new GPU, after a reboot everything is back tonormal now. Made good progress, did not even being to think it was my hardware Thank you for the Assist. the 1080 totally did not fit in this mini nzxt case I had to pull the radiator forward like 4 centimeters and mount the front cage on stand off's just enough I could still get the front cover on. its a packed tight case .
1708235150941.png
 
Dudleydogg said:
it was working prior to this last code project update, like this Morning was all good. My GPU is only 2 years old but yes its a 3090

I believe was just a glitch lot going on with new GPU, after a reboot everything is back tonormal now. Made good progress, did not even being to think it was my hardware Thank you for the Assist. the 1080 totally did not fit in this mini nzxt case I had to pull the radiator forward like 4 centimeters and mount the front cage on stand off's just enough I could still get the front cover on. its a packed tight case .
View attachment 186794
Click to expand...
Im walking around in my garage trying to get one of the cameras to see a License Tag LOL want to see if this works, in Models I put 'alpr'
 
Dudleydogg said:
yes says something about using GPU paddle but your cuda device is not set properly
Click to expand...

00:23:12:ALPR_adapter.py: C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddle\base\framework.py:688: UserWarning: You are using GPU version Paddle, but your CUDA device is not set properly. CPU device will be used by default.
00:23:12:ALPR_adapter.py: warnings.warn(
00:23:25:Face Processing: No face found in image

Logs with same colour are for the same request
it runs till I try to process a test image then shuts down.
 
Last edited:
Dudleydogg said:
00:23:12:ALPR_adapter.py: C:\Program Files\CodeProject\AI\modules\ALPR\bin\windows\python39\venv\lib\site-packages\paddle\base\framework.py:688: UserWarning: You are using GPU version Paddle, but your CUDA device is not set properly. CPU device will be used by default.
00:23:12:ALPR_adapter.py: warnings.warn(
00:23:25:Face Processing: No face found in image

Logs with same colour are for the same request
it runs till I try to process a test image then shuts down.
Click to expand...
Post your install.log file

1708237023532.png
 
MikeLud1 said:
From the quick look everything looks normal. I need to get some sleep because I have to go into work this morning. We will continue troubleshooting later when I get back home for work.
Click to expand...
yea I was ripping out software prior to this new GPU so I put back the 11.8 and the module does not crash now, but only reads tags from test images. If I hold a tag in front of camera its not catching it. but will see what it does tomorrow during the day I have 2 LPR cameras at the gate
all good thank you very much. made good progress.
 
Skinny1 said:
I am having the same issue after upgrading Codeprojet Ai 2.5.4. The Custom Model field is blank in Blue Iris Ai window. I have restarted and rebooted to no avail.

I have checked the Codeproject directory, and the Custom Models are there.

Any other ideas?

Thanks,

Rick
Click to expand...
All: Blue Iris sends a call to CodeProject.AI.Server to list the models when Blue Iris starts (or starts the service if you are running it as a service.)
CodeProject.AI.Server service needs to be running BEFORE Blue Iris starts.
That's why sometimes you will restart the machine and that field will be blank.
To make things easier, in services.msc, select Blue Iris server service, right click, select properties and change the startup from "Automatic" to "Automatic (Delayed Start)".
Then whenever the machine reboots, CodeProject.AI.Server will start, then Blue Iris will start, and you will have the custom models listed.
(If you have "use Custom Models" checked.)
HTH

Sent from my iPlay_50 using Tapatalk
 
  • Like
Reactions: stxrc and Dudleydogg
For fun I ran the deer image through a bunch of different models and sizes. Here is a link to the output:


Note:
  • All models were run on a Coral TPU with an 8-bit model.
  • There is no label for 'deer' in the COCO label set, so a correct answer is to not see anything in the image.
  • The non-YOLO models were downloaded from the Coral example models page.
  • The input image has been stretched to fit the input tensor of the model. So 300x300 or 640x640, etc. One YOLOv8 model has an input that was stretched during model creation to already fit a landscape aspect ratio.
  • If the file is missing, nothing was found.
 
mailseth said:
For fun I ran the deer image through a bunch of different models and sizes. Here is a link to the output:


Note:
  • All models were run on a Coral TPU with an 8-bit model.
  • There is no label for 'deer' in the COCO label set, so a correct answer is to not see anything in the image.
  • The non-YOLO models were downloaded from the Coral example models page.
  • The input image has been stretched to fit the input tensor of the model. So 300x300 or 640x640, etc. One YOLOv8 model has an input that was stretched during model creation to already fit a landscape aspect ratio.
  • If the file is missing, nothing was found.
Click to expand...

Where do you get these Test files images?
 
MikeLud1 said:
From the quick look everything looks normal. I need to get some sleep because I have to go into work this morning. We will continue troubleshooting later when I get back home for work.
Click to expand...
things are rolling this Morning, the 1080 is rocking, Its even running cooler than the 970 was. Tags are hit or Miss but they are there, I thought there was a way to see in Bi logs all the tags detected.
1708291630683.png
 
You must log in or register to reply here.
Top Bottom