OpenALPR (Rekor Scout) compared to Platerecognizer

CamCrazy

Pulling my weight
Aug 23, 2017
522
202
UK
Interested to get others feedback, I dabbled with Platerecognizer due to the integration with Blue Iris and it seemed OK, a few errors, missed some and I didn't like the costs associated with going above 2500 lookups, I stopped using it after a day and didn't bother for a while. Then I decided to give OpenALPR a try after it cropped up on my radar whilst researching Platerecognizer. Only been a day or two but for me it is leagues ahead in both detection accuracy and integration, not to mention the cost being only $5 a month for home use. I would like to store locally but that is $49 from memory, so cloud storage is default for now.

Kind of wishing Blue Iris went with this option now, maybe it was harder to integrate or not possible at the time. Wonder if it might get added as an option at some point. It works well enough standalone that it probably isn't a big issue but I would sure pay some reasonable money ($10-20 a month) to have an all in one solution with Ai and Plate recognition :thumb: If it was solid then I could maybe stomach a bit more money in fact, especially if it was all on site.
 
  • Like
Reactions: Hina Chow
  • Like
Reactions: CamCrazy
You can store locally with OpenALPR on the $5/month subscription. There are two well known programs that allow you do that: OpenALPR Webhook Processor for IP Cameras and OpenALPR Tool - Save and Query CSV Exports Steps I took To Get Mine Working. ZERO Programming Experience.

I've been using OpenALPR for over a year. I love it. I use the program in the first link. I have 103K plates stored since January of this year.

View attachment 100160View attachment 100161
I was being lazy but will look at those two options and implement one :) Pleased to hear you have been happy with it, I can see why even with very limited time using it, incredible results in fact and the analysis tools look useful. Appreciate the response and links, thank you.
 
You can store locally with OpenALPR on the $5/month subscription. There are two well known programs that allow you do that: OpenALPR Webhook Processor for IP Cameras and OpenALPR Tool - Save and Query CSV Exports Steps I took To Get Mine Working. ZERO Programming Experience.

I've been using OpenALPR for over a year. I love it. I use the program in the first link. I have 103K plates stored since January of this year.

View attachment 100160View attachment 100161
If there is a crash course on setting this up using Docker then feel free to pass on tips, I am pretty good with this stuff but my brain is struggling today!! Docker is installed and running fine and for some reason I am now stuck, OpenALPR is of course running fine as you would expect, I now need to fill in the gaps. I wish I was better at this kind of thing I really do :facepalm::rolleyes:
 
I briefly tried PlateRecognizer, but didn't like it because it relied on the camera alerts to send a photo to PR. While testing I quickly realized even my little street would be costing me money. I went with OpenALPR and find it works great. I am on the $5/month plan.Since it uses a live video feed and looks for plates 24/7, it works much better. I have two cameras on it now, running 700 plates per week. One camera does not read plates at night as it is not actually set as an LPR, but an overview that happens to work well during daylight. With even a 6th gen CPU, and all of that and DeepStack on 4-5 cameras, out of the dozen, CPU runs around 20%.

I have not set up anything other than the OpenALPR yet, non of the programs @biggen mentioned, but hope to do so before too long.
 
Where is your Docker running? I run mine from an Ubuntu VM so I just created a start.sh file in that VM with the following settings:

Code:
#!/bin/bash

docker run -d \
--name=openalprwebhookprocessor \
--restart unless-stopped \
--net=bridge \
-v /home/user/webhook_alpr/app/config/:/app/config/ \
-p 3859:80 \
mlapaglia/openalprwebhookprocessor:v3.16.0

When I run that file, it will pull the latest image and immediately run the container. Then you can use a browser to go to the IP of the machine running the container like this You have to create the directory structure first before you run that .sh file. So I created a /home/user/webhook_alpr/app/config directory structure. Then when you run the .sh file, it will created the .db and user file inside the /config directory.

I don't know anything about running Docker in Windows or using the dotnet installation he has available. I'm a Linux geek by nature.
 
Last edited:
I tried running with dotnet command and got this error:

1630328806686.png

Docker is running in Windows on this computer rather than Linux.
 
I played around with the Windows install a few months ago for shits and giggles, and I could never get it to work either. The documentation for the Windows install needs to be more detailed. I couldn't even determine what version of dotnet to install as there are about half a dozen to choose from.
 
  • Like
Reactions: CamCrazy
I played around with the Windows install a few months ago for shits and giggles, and I could never get it to work either. The documentation for the Windows install needs to be more detailed. I couldn't even determine what version of dotnet to install as there are about half a dozen to choose from.
Indeed, I was being a bit blind though and have now pulled down into Docker and slowly working through the thread! I've setup plenty of IP kit and general networking over the years but this is going to test me I can tell! I also don't have the option for a static IP address which probably won't help with the webhook side.
 
You can run something like a RPi from inside your network that will connect to a DDNS server and keep your WAN IP address updated and correct. I do this with one of my VMs. Simple cron job that connects to afraid.org every 5 minutes. Afraid.org even provides free domain names.
 
  • Like
Reactions: CamCrazy
You can run something like a RPi from inside your network that will connect to a DDNS server and keep your WAN IP address updated and correct. I do this with one of my VMs. Simple cron job that connects to afraid.org every 5 minutes. Afraid.org even provides free domain names.
Thanks, another little job for me, what takes the smart people like you 20 minutes probably takes me 2 hours :rofl: but I am a gluten for punishment and determined to get this working now!! Next job is to get the webhook working.
 
Ha! I've been in networking since 1999 so its been a long road of learning that you are really never done with. Doing little projects like this will pay off in the future when you need to do something similar again and you can draw on these experiences to make it a hell of a lot faster and easier.
 
  • Like
Reactions: CamCrazy
Ha! I've been in networking since 1999 so its been a long road of learning that you are really never done with. Doing little projects like this will pay off in the future when you need to do something similar again and you can draw on these experiences to make it a hell of a lot faster and easier.
I know, also been in the IT and networking 'game' since around 2002 but coding/programming not my area of expertise at all! never give up is usually my attitude. You'd think I would know better by know :facepalm:
 
With Rekor Agent is everyone using a detection mask? I seem to get varying performance, one day it is fine and another it just starts lagging and missing plates. Seems like switching to mainstream 1080 feed sometimes works fine and another time it takes too much CPU. For now I think running sub stream is more reliable until maybe GPU can take some load. I tried setting processor cores from 1 to 2 in Rekor Agent but it makes little difference.

Will try feeding it the second substream set to H265 instead of 264 and see if that helps.
 
I'm not using a detection mask, and my two LPRs work fine. One is a dedicated 2MP, and the other a 4 MP that just happens to work for LPR even though it was not setup to do so, it covers my remtoe garage and happens to get plates off the road during the day. I am thinking that with my Dahuas, I had to use 264 to get Rekor to find the stream, but I could be wrong. I'm running an i7-6700 without a GPU and normal CPU is 20% but will spike on motion for DS and ALPR.
 
  • Like
Reactions: CamCrazy
I'm not using a detection mask, and my two LPRs work fine. One is a dedicated 2MP, and the other a 4 MP that just happens to work for LPR even though it was not setup to do so, it covers my remtoe garage and happens to get plates off the road during the day. I am thinking that with my Dahuas, I had to use 264 to get Rekor to find the stream, but I could be wrong. I'm running an i7-6700 without a GPU and normal CPU is 20% but will spike on motion for DS and ALPR.
Thanks for the information, appreciated. I just have a single 2MP PTZ running LPR at the moment but will be switching it to a non PTZ zoom in due course. I am running an older i5-4690 but with 16GB RAM and with my 6 blue iris cams running Ai it sits around 20-35%. Rekor is currently being fed a H265 720p sub stream which balances between quality and CPU usage. I plan to drop in a GTX 1660 GPU soon and am hopeful this will free up the CPU. I do have another machine which was extravagantly built for Blue Iris although currently being used for other tasks, it has a newer i5 with 32GB RAM but no GPU as yet, can't face switching so might leave as is for now!
 
  • Like
Reactions: TVille
I should have added I'm only running the i7-6700 with 8 GB of ram. It is a "new" (eBay special) machine that only came with 8 GB, and it runs well enough that I haven't paid to get an additional 8 GB yet. I'm running a dozen cameras. I have a much older GPU sitting around that I will eventually try out, but haven't gotten around to it yet.
 
I should have added I'm only running the i7-6700 with 8 GB of ram. It is a "new" (eBay special) machine that only came with 8 GB, and it runs well enough that I haven't paid to get an additional 8 GB yet. I'm running a dozen cameras. I have a much older GPU sitting around that I will eventually try out, but haven't gotten around to it yet.
Wow, OK. that is impressive, assume all the cameras are 2MP or above. Do all the cameras have Ai/Deepstack running? I am very interested to see how the GPU helps, I've read good things on the forum here. What is the older GPU you have if you don't mind me asking.