wpiman
Getting comfortable
I am on 6.0.3.8There were a couple of early versions of BI6 where Ken changed the JSON format, but he has changed it back so if you are up to date it should be working.
I am on 6.0.3.8There were a couple of early versions of BI6 where Ken changed the JSON format, but he has changed it back so if you are up to date it should be working.
Check the API key didn't accidentally get reset. Can you give us the exact error message from the ALPR DB logs?I was just looking at my and realized the database hasn't update in a month. I'll take a look at it... I think the format from Blue Iris might have changed? There messages are still going in but it is complaining about the key I think
Check the API key didn't accidentally get reset. Can you give us the exact error message from the ALPR DB logs?
I'm using:So I ran the database command listed above...
Still didn't work.
I ended up having to change my Blue Iris HTTP alert POST body from:
{"ai_dump":&JSON, "Image":"&ALERT_JPEG", "camera":"&CAM", "timestamp":"&ALERT_TIME"}
To:
{"plate_number":"&PLATE", "Image":"&ALERT_JPEG", "camera":"&CAM", "timestamp":"&ALERT_TIME"}
Yeah, I am not sure why the ai_dump stopped working for me. I cloned a camera and one is working with Codeproject and the second ALPR one is working with working with the Blue Iris AI. Both are showing up with that change.I'm using:
{ "ai_dump":&JSON, "Image":"&ALERT_JPEG", "camera":"&CAM", "ALERT_PATH": "&ALERT_PATH", "ALERT_CLIP": "&ALERT_CLIP", "timestamp":"&ALERT_TIME" }
and it's been working fine through the 5.x to 6.x transition, once I checked all the other required boxes.
Yeah, I am not sure why the ai_dump stopped working for me. I cloned a camera and one is working with Codeproject and the second ALPR one is working with working with the Blue Iris AI. Both are showing up with that change.
Yeah, I am not sure why the ai_dump stopped working for me. I cloned a camera and one is working with Codeproject and the second ALPR one is working with working with the Blue Iris AI. Both are showing up with that change.
. I've now got that turned off, I hope, and for all you Win10 folks out there, it's a different registry setting than the one that worksd in Win10.On a related note, what's are my options for offloading more AI load onto another device, either a PCIe GPU or ???So I just clicked the one radio button to switch from the "local service" (e.g. CPAI) to the Built-In AI:
View attachment 240023
then turned off the Object Detection and LPR in the CPAI dashboard:
View attachment 240021
and I'm still getting plates, so I guess it really is that simple! After a few days I will uninstall CPAI completely.
Is the built-in service any more efficient re: CPU and RAM than using CPAI?
I am using codeproject AI on a linux docker that does one camera.On a related note, what's are my options for offloading more AI load onto another device, either a PCIe GPU or ???
When the driveways are busy, my CPU nearly pegs, and stays that way through the high traffic period.
Which GPU? Or does it not matter because they're all fast enough?I had CPAI running on an 11th gen i3 and used the iGPU, and with Mike's YOLO v11 DirectML code, it was very good. Inference times were around 80-100ms. On another system, I'm running on an 11th gen laptop with an Nvidia GPU in Proxmox and it's 25ms with 4 cameras. The laptop is basically idle and uses very little power.
@iwanttosee that’s a good find. Thanks for sharing. If you add a known plate from your detections first, you should then see the button to add one without having seen it before. This is the original empty state from before it was possible to add them manually.
I have time to work on it this week.
Which GPU? Or does it not matter because they're all fast enough?
Thanks! I'm going to move this discussion to this new thread, as my question really has nothing to do with the ALPR database!For this application, I dont think it matters. I have a RTX3000 (Turing architecture) with 6GB of vram.