CodeProject.AI Version 2.0

harleyl7

Pulling my weight
Joined
Jun 4, 2021
Messages
260
Reaction score
223
@MikeLud1 do you use the Rockchip for all detection or just LPR? My cpu can't really handle the load of CPAI and I don't want to waste the watts by running my GTX 1080. So will the Rockchip work for LPR and main detection? ~9 cameras, but not all are running AI
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,219
Reaction score
4,261
Location
Brooklyn, NY
@MikeLud1 do you use the Rockchip for all detection or just LPR? My cpu can't really handle the load of CPAI and I don't want to waste the watts by running my GTX 1080. So will the Rockchip work for LPR and main detection? ~9 cameras, but not all are running AI
My main Blue Iris system I have RTX 3060 for AI. The Orange Pi5 Rockchip will do both LPR and object detection. There is a bug that causes the detection times to slowdown after a while that I have not found the fixed yet.
 

harleyl7

Pulling my weight
Joined
Jun 4, 2021
Messages
260
Reaction score
223
Ah gotcha. Guess I need to throw a 1060 in my server. But just hold off on LPR for a little. The GTX 1080 just eats too much power
 
Joined
Jun 2, 2021
Messages
28
Reaction score
11
Location
usa
Welp all I had to was be patient, nothing was changed since i made my last post and I see the delivery model is being used now. Here's a break down of my settings for anyone else who runs into similar issues.

Docker config
YAML:
version: '3.3'
services:
  ai-server:
    network_mode: host
#    ports:
#      - '32168:32168'
    container_name: CodeProject.AI-Server
    volumes:
      - '/opt/senseai/ai:/etc/codeproject/ai'
      - '/opt/senseai/modules:/app/preinstalled-modules/ObjectDetectionYOLOv5-6.2/custom-models'
    image: 'codeproject/ai-server:cuda12_2-2.5.0'
    restart: always
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              capabilities: [gpu]
Note the 2nd item in the volumes section, that's where the custom models from CodeProject.AI-Custom-IPcam-Models/PT Models at main · MikeLud/CodeProject.AI-Custom-IPcam-Models are in my case

BI Settings:

Go to settings > AI
The use custom models option should be checked, but the path can be blank

Screenshot 2024-01-16 181248.png

For the camera setup the AI section in the alert section as below

The labels for delivery services as far as I know are: fedex,usps,ups,amazon,garbage,dhl
I have "to confirm" set as: person,cat,dog,fedex,usps,ups,amazon,garbage,dhl
"custom models" is set as: ipcam-general,ipcam-animal,delivery

I have to use Nothing found:0 to prevent cancels from showing up in the alerts.
Screenshot 2024-01-16 161529.png


I still can't get the old clip to show delivery object analysis but the live feed is definitely checking the delivery object as I see it being used in the AI logs in Code Project Ai's web interface.
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,219
Reaction score
4,261
Location
Brooklyn, NY
Welp all I had to was be patient, nothing was changed since i made my last post and I see the delivery model is being used now. Here's a break down of my settings for anyone else who runs into similar issues.

Docker config
YAML:
version: '3.3'
services:
  ai-server:
    network_mode: host
#    ports:
#      - '32168:32168'
    container_name: CodeProject.AI-Server
    volumes:
      - '/opt/senseai/ai:/etc/codeproject/ai'
      - '/opt/senseai/modules:/app/preinstalled-modules/ObjectDetectionYOLOv5-6.2/custom-models'
    image: 'codeproject/ai-server:cuda12_2-2.5.0'
    restart: always
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              capabilities: [gpu]
Note the 2nd item in the volumes section, that's where the custom models from CodeProject.AI-Custom-IPcam-Models/PT Models at main · MikeLud/CodeProject.AI-Custom-IPcam-Models are in my case

BI Settings:

Go to settings > AI
The use custom models option should be checked, but the path can be blank

View attachment 183049

For the camera setup the AI section in the alert section as below

The labels for delivery services as far as I know are: fedex,usps,ups,amazon,garbage,dhl
I have "to confirm" set as: person,cat,dog,fedex,usps,ups,amazon,garbage,dhl
"custom models" is set as: ipcam-general,ipcam-animal,delivery

I have to use Nothing found:0 to prevent cancels from showing up in the alerts.
View attachment 183050


I still can't get the old clip to show delivery object analysis but the live feed is definitely checking the delivery object as I see it being used in the AI logs in Code Project Ai's web interface.
It looks like Blue Iris has not pulled the custom model list. To have it pull the list restart Blue Iris service.

1705462164017.png
 
Joined
Jun 2, 2021
Messages
28
Reaction score
11
Location
usa
It looks like Blue Iris has not pulled the custom model list. To have it pull the list restart Blue Iris service.

View attachment 183054
I've restarted the BI service multiple times using the Net command but I restarted the whole system this time and still no change on this page.
Screenshot 2024-01-16 223454.png


Something is wrong with the host system's network connectivity to AI I believe. Clicking on the open ai dashboard link leads to an infinite loading screen in IE. I also removed a bunch of what I felt were unnecessary stuff from this windows install and that could've broken something too.
Weird that the AI detection calls go through.
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,219
Reaction score
4,261
Location
Brooklyn, NY
I've restarted the BI service multiple times using the Net command but I restarted the whole system this time and still no change on this page.
View attachment 183055


Something is wrong with the host system's network connectivity to AI I believe. Clicking on the open ai dashboard link leads to an infinite loading screen in IE. I also removed a bunch of what I felt were unnecessary stuff from this windows install and that could've broken something too.
Weird that the AI detection calls go through.
Maybe you need to reinstall the windows version, below a link to the current version.

 
Last edited:

nils_92

n3wb
Joined
Dec 13, 2019
Messages
12
Reaction score
2
Location
Germany
with 2.5.0 rc5 i seem to have lost do "options" on the modules I cannot change the size anymore
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
25,060
Reaction score
48,873
Location
USA
with 2.5.0 rc5 i seem to have lost do "options" on the modules I cannot change the size anymore
Same - I have lost the gear icon on the far right for each module that lets you set those things:

1705499530094.png
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
25,060
Reaction score
48,873
Location
USA
I am on the latest version:

1705499979326.png

CTRL+F5 and starting/restopping didn't bring the gear settings icon back.


Further it shows an update is available for the modules yet when you try to update it says they are already installed:

1705499694701.png
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,219
Reaction score
4,261
Location
Brooklyn, NY
I am on the latest version:

View attachment 183091

CTRL+F5 and starting/restopping didn't bring the gear settings icon back.


Further it shows an update is available for the modules yet when you try to update it says they are already installed:

View attachment 183087
CodeProject.AI will be releasing a new version most likely later today that the update will work , the updates do not work on 2.5-RC5
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
25,060
Reaction score
48,873
Location
USA
CodeProject.AI will be releasing a new version most likely later today that the update will work , the updates do not work on 2.5-RC5
OK - hopefully that update will fix the missing gear icon as well!
 
Top