OpenALPR Webhook Processor for IP Cameras

Yeah, I don't trust Rekor to decide that you need a higher plan to retain local images or something
 
  • Like
Reactions: biggen
Ok v4.1.0-alpha3 now stores plates locally. i'm still testing it, if you run the agent scrape it should pull in all the images since the last time it was run, right now old images won't show up. I also added a db index that should make scraping run a little faster.
 
  • Like
  • Love
Reactions: CamCrazy, biggen and IReallyLikePizza2
Awesome!

So you're saying at some point I should be able to get it to pull in the 180K images I have already linked in there from OpenALPR?
 
IReallyLikePizza2 said:
Awesome!

So you're saying at some point I should be able to get it to pull in the 180K images I have already linked in there from OpenALPR?
Click to expand...
Ya before this branch gets released. if you want to try it now delete your processor.db (back it up first) restart your service and run an agent scrape, it will grab all the plates currently on the openalpr agent and populate the processor.
 
v4.1.0-alpha4 will scrape the images of all the missing plate groups after the regular scrape finishes.
 
biggen said:
I haven't had a chance to try this yet. Anyone?
Click to expand...

Yes I have tried it. I think that things are working correctly as far as the plates being store locally. I changed my Endpoint URL and I still got the plates to preview. My question is though does it still need to connect to the agent for anything or is everything sent thru the Webhook? I would like to get my Agent back into the Docker but not ready to pull that trigger if it still needed for the Processor to work with.

I do have a problem with the Scrapper not doing anything. I get the following message in the log

Code: 
11/03/2021 16:01:58 +00:00 Finished OpenALPR Agent scrape.
11/03/2021 16:01:58 +00:00 no metadata found for given date range: ""
11/03/2021 16:01:58 +00:00 Scraping between 1635953041829 and 1636039441829
11/03/2021 16:01:58 +00:00 no metadata found for given date range: ""
11/03/2021 16:01:58 +00:00 Scraping between 1635866641829 and 1635953041829
11/03/2021 16:01:58 +00:00 no metadata found for given date range: ""
11/03/2021 16:01:58 +00:00 Scraping between 1635780241829 and 1635866641829
 
  • Like
Reactions: biggen
the webhook does not carry the full resolution image. when a webhook comes in the processor calls the agent and gets the image 1 time, then stores it locally.

no metadata found for given date range: "" means your agent didn't see any plates during those dates. the numbers are unix epoch milliseconds. you can plug them in here to see what date it converts to Epoch Converter
 
I've not tried it out yet, this weekend I might try move Rekor to my Blue Iris machine to use GPU, and then at the same time give this a shot
 
  • Like
Reactions: CamCrazy and biggen
Just so I get it right, if I update the latest and then run the scrape, it will go and downloaded all the images from the agent and put them into the container data?

Once thats done, I assume I can completely power down the OpenALPR box if I wanted to, and everything would still show up going back? I wonder if I should try this out before migrating, then I don't even need to mess with the OpenALPR database moving over
 
ya plates will show up once you run the scrape. you still need openalpr agent running though if you want to get new plates obviously.

v4.1.0-alpha15 is the latest i am testing right now

make sure you back up your db
 
  • Like
Reactions: biggen
That is so cool! Thanks!

As soon as I finish work I'm going to give it a shot
 
Right now I'm on mlapaglia/openalprwebhookprocessor:latest

Do I just replace latest with the build? Like mlapaglia/openalprwebhookprocessor:v4.1.0-alpha15
 
CamCrazy said:
I'm all ears on how this works out, I've had mixed results with the GPU so far :wtf:
Click to expand...

It wasn't good, still had 20% CPU load from OpenALPR, and it missed a few plates
 
I believe the CPU is still used for motion detection, then it uses the GPU to process the plates
 
I guess that's understandable, but 20% CPU seems excessive considering that's about what BI is using, but I have 12 cameras in there
 
Basically in CPU mode mine uses around 6-8% CPU idle, in GPU mode it use around 30% CPU idle! events smooth out CPU spikes when running GPU though, on balance I can't say which is best! tough call.
 
You must log in or register to reply here.
Top Bottom