Full ALPR Database System for Blue Iris!

Update Released:
  • Support for 24 hour time
  • Fixed max records pruning
  • Time based recognition filtering
  • Pagination for database page
  • Notification time zone fix
  • Live feed plate image modal
  • Search matching improvements
  • UI Improvements
 
Thanks for this update! With over 11,000 records in my database this version loads almost instantaneously.
I top you with almost 25k of records.

1733785863674.png
 
  • Like
Reactions: Vettester
I only have 1300 so far. Those are some crazy numbers. Really makes me think about the crowdsourcing potential - I'm not sure exactly what that would look like.

One thought that came to mind was keeping a daytime image for each plate. Especially from the front, lots of times we see decent facial ID through the windshield. That could definitely come in handy.

--

While thinking about this, an idea I was exploring many months ago came back to mind. Listening to and logging 300-400MHz radio signals near your road. Why? Tire pressure monitors. TPMS sensors all have a unique ID and send their reading to your car wirelessly with no encryption. This signal can be received from up to something like 50 feet away.

So what does this mean and why might someone want this? Well, if your vehicle of interest is smart enough to take its plate off, you're kind of SOL, at least at night when you don't have much else to work off of, like a vehicle description. The device IDs of a vehicle's TPMS sensors could serve as a secondary piece of identifiable information in situations where we aren't able to capture a license plate. By capturing both, the TPMS ID's can be associated to the license plates of the cars that they came from. So, if a vehicle first passes your property at some other time with its plate mounted (which I would guess happens in the majority of cases), be it earlier that day, 3 months earlier, or even after an incident occurs, the system will still have sufficient data to identify the vehicle even with its license plate removed...

This is obviously getting way more into the weeds, but I've had so many incidents and am so paranoid now that I just might do it. I also just think it's fascinating to think about the fact that you could. Could easily be set up with some basic equipment for no more than $100-300.
 
Using generative AI to capture the make/model/color during daylight hours would also be an awesome addition. Something like this:
IMG_3505.jpeg
 
  • Like
Reactions: Skinny1
  • Like
Reactions: algertc
Interesting ok that would be really nice to have if it worked well.
It would be nice to incorporate something like this into your app as an option. You could add an analyze with AI option button on your image page.
 
It would be nice to incorporate something like this into your app as an option. You could add an analyze with AI option button on your image page.
What I’d really like to do is have it come from codeproject with everything else or be automated in some other fashion for people who may want to use a paid service. The paid option I’d be open to adding an application side integration since CP isn’t really an option.

When I think about this, the most valuable prospect is the crowdsourcing side of things. Especially if you guys are getting thousands of plates per month, this could really generate a pretty reliable open database of license plates similar to the information you would get from Kelly blue books or Oreilly or any data broker.

I don’t know anyone else in my area that could be convinced to roll a full ALPR setup, but if you’re located somewhere with more closely located users or could convince any friends to collect data also, the opportunity is intriguing.

This is pretty topical right now too amid the UNH situation. Aggregating data from different recognition locations and allowing for some semi-big data analytics would enable an unparalleled level of insight and monitoring of your traffic.
 
@algertc, I just noticed that your last update seems to have broken the time distribution chart. All columns are listed as 12am.

Screen Shot 2024-12-19 at 7.08.04 AM.png