LPR: Database & Alerts

[kjinxx2]

I put together a pretty neat system with my LPR cams and a number of other software installations that does the following:

1) IP Cam for license plate photo & ONVIF trigger to Blue Iris
2) Blue Iris to pass photos to CodeProject.AI for license plate detection & first attempt at reading
3) Blue Iris to post alert photo to SFTP server & send MQTT topic which is subscribed by Node Red server
4) Node Red to determine if plate number was read & if it's a standard 6 digit plate, else pass image location to local OpenALPR server and process using Python script
5) Python script sends MQTT topic back out and subscribed Node Red server receives updated plate info
6) Final results are posted into LAMP stack (stored in MySQL database)
7) Node Red runs a query and determines if the detected vehicle is worthy of alerting (based on an algorithm I came up with) or if is specifically identified as an 'alert' plate - at that point I can whitelist the plate, or do nothing and I'll be notified again when that plate passes again
8) Node Red send MQTT topic with Alert details to my local subscribed Home Assistant server which handles the notification & subsequent actions

I'm looking to use a service to check the registered address of a potentially alerted license plate - if the address is local then do not alert, etc etc. I have seen some with access to look up registration info on here before, but not sure if it's simply because they work for a government agency.

Is there a service that gives registration info based on license plate number?

Thanks

 

[wittaj]

We wish we had that capability, but we have been told that information is illegal.

What people have done is paid for a per lookup (like $9.99) for publicly available information, but it is maybe 50% accurate.

 

[kjinxx2]

Ahh that makes sense, I stumbled across a few sites that appeared to offer a service you just described for $9.99/check or $29.99/mo but wasn't sure if it would give me what I needed anyway. I'm scanning ~6k plates per month so it would have to have some sort of REST API or similar and obviously a 50% accuracy rate isn't ideal either

 

[SIR VEYOR]

Probably the best you can do to narrow your outliers is have a “template” or similar of your states assorted plates.

Even easier is if you have county numbers on the plates. Then you can screen out your county and maybe adjacent ones.

 

[kjinxx2]

Probably the best you can do to narrow your outliers is have a “template” or similar of your states assorted plates.

Even easier is if you have county numbers on the plates. Then you can screen out your county and maybe adjacent ones.

No county numbers but my algorithm flags new plates that have visited multiple times within the last month - so I'm just trying to eliminate false positives on new neighbors/vehicles. Even delivery trucks etc. are eliminated as they have a pattern of appearing multiple times more than 30 days in the past. I'll keep digging and report back if I come up with any other methods to help identify potential vehicles that are casing the neighborhood