Yes, one thing to look for: there was a specific MS update (sorry , I don't have the exact KB # to reference at the moment) to .NET in January of this year that caused this mess on my system for a couple of days. I held off on the update for almost two months, then decided to install it, but had to remove this update quickly as it was causing CPAI crashes and instability. After removing this update, my system reverted to its usual complete stability (I guess count me among those rare cases that generally find CPAI quite reliable and stable).
AI: not responding
- Thread starter BassTeQ
- Start date
I've been troubleshooting 'AI: not responding' and confirmed with a packet capture on localhost that Blue Iris sends an HTTP POST to CodeProject.AI (CPAI), the image goes into the queue (objectdetection_queue), is then dequeued, then nothing happens. Blue Iris waits 15 seconds (Blue Iris's AI timeout setting) then "AI: not responding" is seen. Below I have two working exchanges where CPAI responds properly to the request and a problem example. I've pulled the image for analysis out of the packet capture and manually run it through CPAI to see if the problem is the image to analyze but that works perfectly fine. I also see no difference in the HTTP POST between a working and non-working exchange going to CPAI.
Using CodeProject.ai version 2.9.5 on Windows 10 with Object Detection (YOLOv5 6.2) 1.10.0.
Notice in the problem section compared to the others, there is no log of Object Detection retrieved command... I looked through CPAI's github page to where the request is printed but I am stuck trying to debug how/where the handoff to the YOLOv5 module happens to try and see why intermittently YOLOv5 doesn't process it even though it is taken out of the CPAI queue.
As a test, I've moved CPAI off the Windows machine running Blue Iris to a docker container (codeproject/ai-server:2.9.7) using the same Object Detection (YOLOv5 6.2) 1.10.0 and am still seeing an occasional 'AI: no responding' with the same exact signature as above. To mean this means the issue is within CPAI itself or the detection module.
Using CodeProject.ai version 2.9.5 on Windows 10 with Object Detection (YOLOv5 6.2) 1.10.0.
Code:
### Working
2025-03-19 21:57:07: Client request 'custom' in queue 'objectdetection_queue' (#reqid 6cce23d2-327e-4439-a9fb-8263e02e1956)
2025-03-19 21:57:07: Request 'custom' dequeued from 'objectdetection_queue' (#reqid 6cce23d2-327e-4439-a9fb-8263e02e1956)
2025-03-19 21:57:07: Object Detection (YOLOv5 6.2): Retrieved objectdetection_queue command 'custom' in Object Detection (YOLOv5 6.2)
2025-03-19 21:57:07: Object Detection (YOLOv5 6.2): Detecting using ipcam-general in Object Detection (YOLOv5 6.2)
2025-03-19 21:57:07: Response rec'd from Object Detection (YOLOv5 6.2) command 'custom' (#reqid 6cce23d2-327e-4439-a9fb-8263e02e1956) ['No objects found'] took 199ms
##### Problem - AI: not responding #####
2025-03-19 21:57:53: Client request 'custom' in queue 'objectdetection_queue' (#reqid 5315f07e-6404-46ba-9001-3c970561e4dd)
2025-03-19 21:57:53: Request 'custom' dequeued from 'objectdetection_queue' (#reqid 5315f07e-6404-46ba-9001-3c970561e4dd)
################################
### Working
2025-03-19 21:58:01: Client request 'custom' in queue 'objectdetection_queue' (#reqid bb5296be-c4da-41f4-9ed3-1e933c02c8c9)
2025-03-19 21:58:01: Request 'custom' dequeued from 'objectdetection_queue' (#reqid bb5296be-c4da-41f4-9ed3-1e933c02c8c9)
2025-03-19 21:58:01: Object Detection (YOLOv5 6.2): Retrieved objectdetection_queue command 'custom' in Object Detection (YOLOv5 6.2)
2025-03-19 21:58:01: Object Detection (YOLOv5 6.2): Detecting using ipcam-general in Object Detection (YOLOv5 6.2)
2025-03-19 21:58:01: Response rec'd from Object Detection (YOLOv5 6.2) command 'custom' (#reqid bb5296be-c4da-41f4-9ed3-1e933c02c8c9) ['No objects found'] took 153msNotice in the problem section compared to the others, there is no log of Object Detection retrieved command... I looked through CPAI's github page to where the request is printed but I am stuck trying to debug how/where the handoff to the YOLOv5 module happens to try and see why intermittently YOLOv5 doesn't process it even though it is taken out of the CPAI queue.
As a test, I've moved CPAI off the Windows machine running Blue Iris to a docker container (codeproject/ai-server:2.9.7) using the same Object Detection (YOLOv5 6.2) 1.10.0 and am still seeing an occasional 'AI: no responding' with the same exact signature as above. To mean this means the issue is within CPAI itself or the detection module.
Last edited:
I just upgraded to .33 and I am getting this now, but I am still using Deepstack 
So I really have no clue what is going on. It was working fine before. When I restart blueiris it says AI is responding and I can get to the Deepstack webpage at 127.0.0.1:82.
I rolled back to 5.9.8.5 but it didn't seem to change anything. Now I am on 5.9.9.19 and I just get AI Alert Cancelled nothing found when its clearly me on the front port and it used to catch that no problem.
Well I overrode one camera and pointed its AI to a blue-onyx instance I was testing on my NAS. It seems to have worked right and BI got the AI confirmation no problem. Noooot sure what is going on with my native Deepstack install. The only thing that changed on the system was Blueiris updates =(
So I really have no clue what is going on. It was working fine before. When I restart blueiris it says AI is responding and I can get to the Deepstack webpage at 127.0.0.1:82.
I rolled back to 5.9.8.5 but it didn't seem to change anything. Now I am on 5.9.9.19 and I just get AI Alert Cancelled nothing found when its clearly me on the front port and it used to catch that no problem.
Well I overrode one camera and pointed its AI to a blue-onyx instance I was testing on my NAS. It seems to have worked right and BI got the AI confirmation no problem. Noooot sure what is going on with my native Deepstack install. The only thing that changed on the system was Blueiris updates =(
