Full ALPR Database System for Blue Iris!

My ALPR database randomly stopped receiving plate data.

I do not see any errors in the ALPR app log files. Blue iris is still recognizing plates. Nothing has changed in any of my Alert settings on the Blue Iris end, my API key is the same.

What's the best way to test or diagnose why?
 
Check the Blue Iris log files to confirm what happens when a plate is detected. You should see a successful call to the ALPR database, or an error.

If you see an error there, then the problem is with the call. I'd start by verifying the IP of the ALPR DB server hasn't accidently been changed.

Note: if the ALPR DB is on the same server as BI, you can use "localhost" as the IP rather than the explicit IP.

Can you post the relevant section of the BI log?
 
I think I’m going to publish some core n8n automations for other blue Iris users to use that they can easily download and apply.

What I’ve really been thinking about, though, is what a seriously powerful futuristic home surveillance system would look like. And I’m not exactly sure.

I feel like it would be super cool to have an iron man Jarvis style AI voice agent to ask anything to and immediately get the footage pulled up exactly how you want it or get any question answered.

I’m particularly really interested in voice agents because of the speed with which you can do things, but I can’t seem to really brainstorm much to really take this to the next level beyond retrospective research or analysis of what is going on in the current moment.

We all know that, for the most part, cameras just record crimes, not stop them. I’m interested in the potential for more active deterrence for that, but also just for general day-to-day, non crime purposes. Most of us aren’t getting robbed on a daily basis and there are so many other things that can be built around this that I’m not sure how to think about.

I’m curious if anyone has any visions or ideas for what a futuristic Jarvis style home surveillance/monitoring system could look like.

I think facial recognition management is probably a good next step, but I want to think a bit bigger than that. There really isn’t much technological limitation to building things that would have previously been sci-fi.

Having a good license plate recognition system is great, but I’m sure we would all love to have an all intelligent Jarvis watching over our properties and taking action on things all day long like something out of a movie.

Please do share any ideas. I’m curious about anything anyone might be able to think of.
 
  • Like
Reactions: Skinny1
wittaj said:
Any more work on the TPMS system?
Click to expand...
Yes, I’ve already built out the page to show the TPMS live and receive it, but the decoder that is popular, built into RTL 433, has been extremely unreliable for me. It worked great at one point, so I’m not entirely sure why.

I certainly do think this is a key advantage in vehicle detection intelligence though. Especially considering the fact that it’s so easy to use existing foundation models to associate the TPMS IDs to the plates now. I have a fully built-out box mounted to a tree in my yard with all the hardware in it. It has been analyzing the TPMS for a while and it just isn’t reliable enough.

It’s been on my mind and I’m still thinking about how to make it work properly. It looks cool right now and kind of works, but I want it getting every single vehicle that passes and not failing to decode all the time.
 
Last edited:
  • Like
Reactions: AlwaysSomething and Skinny1
Personally, I'd love to see more features related to LPR than have agentic features to slurp out video snips or use it for super intelligent automation. Here is why:

  • Limited by video quality. Our video sources are probably great in day time, but at night, it's usually a big compromise.
  • Privacy with AI. These cameras are in our homes and on our property, and our Overlords already know how many times I flush my toilet.
  • Security of our property vs quality of life. Security is more important. I rather help catch the guy that doesn't pick up after their dog than have Javis tell me how often the girl in the yoga pants walks by the camera.

My data shows that I am correctly 40%+ of the OCR, and frequently it's the same plates. Would love to see auto-correction features on the front end to compensate for OCR issues. Some AI to post process images of cars to auto-build descriptions of make, model, color, etc would be super useful so I can search for "green 4 door honda" and have it show me a bunch of plates of cars that match this description.

Anyway, for me, I rather see us go deeper into ALPR rather than expand wider into general camera intelligence.
 
  • Like
  • Haha
Reactions: algertc, AlwaysSomething and Skinny1
I am stuck on the opening screen, i have the option to verify and skip migration, or backfill,. when i choose verify and skip migration i get an erorr "relation "plate_reads" does not exist" how do i solve this issue (windows 11-docker)
 
PeteJ said:
Personally, I'd love to see more features related to LPR than have agentic features to slurp out video snips or use it for super intelligent automation. Here is why:

  • Limited by video quality. Our video sources are probably great in day time, but at night, it's usually a big compromise.
  • Privacy with AI. These cameras are in our homes and on our property, and our Overlords already know how many times I flush my toilet.
  • Security of our property vs quality of life. Security is more important. I rather help catch the guy that doesn't pick up after their dog than have Javis tell me how often the girl in the yoga pants walks by the camera.

My data shows that I am correctly 40%+ of the OCR, and frequently it's the same plates. Would love to see auto-correction features on the front end to compensate for OCR issues. Some AI to post process images of cars to auto-build descriptions of make, model, color, etc would be super useful so I can search for "green 4 door honda" and have it show me a bunch of plates of cars that match this description.

Anyway, for me, I rather see us go deeper into ALPR rather than expand wider into general camera intelligence.
Click to expand...

Re general automations, I wasn’t going to build that into this app. Posted here just because it’s kind of adjacent.

I’d agree with your points as far as ALPR goes. I’ve just been playing around with a bunch of new AI tools recently and find it exciting to see what u can do.

Right now the build I have has the capability to do all the backfilling of vehicle descriptions and tags and much more. I might add a simple button to just fill vehicle descriptions but the nice thing about having it as an agent is the flexibility it offers to allow everyone, each with their different tags and needs, to ask and get a pretty reliable behavior.

It’s a nice shortcut instead of having to program functionality for every individual use case.

Re privacy:
There will be a model provider configuration screen in settings that allows you to use your own local models. Nothing ever gets sent to any AI provider unless you set it up and turn it on.

I’ve been pretty privacy conscious with all of the settings.
 
  • Like
Reactions: PeteJ
1 7/3/2025 10:16:01.866 PM Cam2 Web: No stream: 500

7/3/2025, 10:18:40 PM [ERROR] Error processing request: SyntaxError: Unexpected token ',', ...""ai_dump":, "Image":"... is not valid JSON
7/3/2025, 10:18:41 PM [INFO] POST /api/plate-reads
7/3/2025, 10:18:41 PM [ERROR] Error processing request: SyntaxError: Unexpected token ',', ...""ai_dump":, "Image":"... is not valid JSON
7/3/2025, 10:19:12 PM [INFO] GET /logs
7/3/2025, 10:19:12 PM [INFO] Middleware checking path: /logs, Session ID from cookie: 0cfea52080639e6b4faa6651795d8d85b79ff5e0cdb454aff1e91c9e71351670
7/3/2025, 10:19:12 PM [INFO] POST /api/verify-session
7/3/2025, 10:19:12 PM [INFO] GET /api/check-update
7/3/2025, 10:19:12 PM [ERROR] Error checking update status: Error: connect ECONNREFUSED 172.18.0.2:5433
 
{ "ai_dump":&JSON, "Image":"&ALERT_JPEG", "camera":"&CAM", "ALERT_PATH": "&ALERT_PATH", "ALERT_CLIP": "&ALERT_CLIP", "timestamp":"&ALERT_TIME" }

generates the errors above.

{ "ai_dump":"&PLATE", "Image":"&ALERT_JPEG", "camera":"&CAM", "ALERT_PATH": "&ALERT_PATH", "ALERT_CLIP": "&ALERT_CLIP", "timestamp":"&ALERT_TIME" }

also generates the same error
 
You must log in or register to reply here.
Top Bottom