OpenALPR Webhook Processor for IP Cameras

[wittaj]

I would guess these pics are in the kb range, so 8GB is probably a lot of pictures stored.

 

[IReallyLikePizza2]

Are you guys storing video? I'm not even sure what that could be used for...

I have 5000ish plates which is taking up 5GB, but 30GB is video, which I'm not even sure what is being used for

 

[biggen]

I have been adjusting it in the UI, but it seems to sync up as my web page says the same

View attachment 80203

I'm running the agent on Linux in Docker so I don't even have that menu. I have to control the agent from the Dashboard.

No I'm not having it record video. I'll log into the container tomorrow and see how much room the pictures are consuming.

 

[mlapaglia]

you can control the agent through the alprd.conf file

 

[biggen]

you can control the agent through the alprd.conf file

That's inside the container though for me and not meant to be accessible to most people. I had to enable the webserver from there to get pictures working.

 

[mlapaglia]

you can map the /etc/openalpr/ folder and modify it outside. Watchman Quick Start — openalpr 2.8.101 documentation

 

[biggen]

you can map the /etc/openalpr/ folder and modify it outside. Watchman Quick Start — openalpr 2.8.101 documentation

Ahh I did not know that!

 

[mlapaglia]

it's in there instructions OpenALPR Agent (alprd) — openalpr 2.2.0 documentation

Edit the configuration file

sudo nano /etc/openalpr/alprd.conf

Configure the following properties:

• company_id (using the value found on the web server)
• site_id (a unique text value that identifies this agent)
• stream (The mjpeg URL for the camera stream. Multiple streams are added as additional stream entries)
• analysis_threads (number of CPU cores dedicated to OpenALPR processing)
• upload_address (set this to http:///[Web_server_IP]/push/)
• store_plates_maxsize_mb (amount of space used for storing license plate images)

Test the daemon from the CLI to make sure your settings are correct. Press CTRL+C once you see that the video stream is connected.

alprd -f

 

[wittaj]

Are you guys storing video? I'm not even sure what that could be used for...

I have 5000ish plates which is taking up 5GB, but 30GB is video, which I'm not even sure what is being used for

What quality/size are your plate photos?

I have 43,000 plate photos taking up 7GB

 

[mlapaglia]

the size of your photos will depend on your camera resolution, i record in 720p so there are less pixels to process and less cpu usage.

 

[biggen]

I have ~3500 plates stored on the local Webhook service. That corresponds to 1.6GB of plate pictures on the client or about 450KB per picture.

My watchman agent is actually showing plates all the way back till the beginning of January when I first signed up. But that is before the webhook service was a thing so I don't have access to them since on the $5/plan you can't look at plates more than 5 days out. So from Jan 1 (my earliest plate pictures) till today I'm only using 4GB of space out of 8GB reserved for them. So I'd say there will be close to two months worth of plate pictures for me with 8GB of space set aside for them judging from current traffic patterns.

 

[mlapaglia]

i bet there's a way to get that plate data since it's still stored in the agent, and the agent isn't aware of what your subscription is.

 

[wittaj]

i bet there's a way to get that plate data since it's still stored in the agent, and the agent isn't aware of what your subscription is.

Is that a challenge to yourself to figure that out LOL

 

[biggen]

i bet there's a way to get that plate data since it's still stored in the agent, and the agent isn't aware of what your subscription is.

But those plate "records" aren't on your local service DB since I had collected those plates before I ever started running your service. So your service doesn't even know about them. The hydrator would have taken care of pulling plates off the OpenALPR cloud service that weren't stored on the local webhook service, but as we found out, they require a commercial license to pull them down.

 

[mlapaglia]

But those plate "records" aren't on your local service DB since I had collected those plates before I ever started running your service. So your service doesn't even know about them. The hydrator would have taken care of pulling plates off the OpenALPR cloud service that weren't stored on the local webhook service, but as we found out, they require a commercial license to pull them down.

i bet the data is in there somewhere, might need to crack the client's db open and take a look.

in the mean time v3.2.0-alpha5 is out with preliminary search abilities. there are probably issues with timezone when searching with date/time filters, and the plate search is case sensitive right now.

 

[wittaj]

I have ~3500 plates stored on the local Webhook service. That corresponds to 1.6GB of plate pictures on the client or about 450KB per picture.

Wow - using the snapshot from BI for Plate Recognizer, my plate pictures are 75KB per picture.

Can you specify a resolution or size for the pics using OpenALPR or is it a direct relation to the resolution of the videostream that you are using?

Does this system take a snapshot of any vehicle or motion or just when it has a plate?

Does this system take the videostream and truly look for a plate and then take a pic only of plates or does it take a pic early like a large semi or bus enters the frame but the plate isn't visible yet?

Sorry for all the questions! I am just trying to figure out as much as I can before I start the trial period of OpenALPR and get my system as close as possible from the start to be usable.

 

[mlapaglia]

probably best to read through the docs OpenALPR Documentation — openalpr 2.8.101 documentation

is 1.6gb really that much? i've got a 50TB server this runs on haha

 

[wittaj]

Thanks I will take a look at that. I know I looked at it before, but a lot was above my skillset. But maybe I have learned some things between then and now LOL.

OMG 50TB of storage - so jealous!

I have 43,000 plate photos taking up 7GB using Plate Recognizer whereas others are getting maybe 7,000 plates with OpenALPR, so just trying to figure out how to get going somewhat efficiently.

 

[mlapaglia]

do you care about the plate images? you can set your retention limit to whatever you want then it will auto prune old images.

 

[biggen]

Wow - using the snapshot from BI for Plate Recognizer, my plate pictures are 75KB per picture.

Can you specify a resolution or size for the pics using OpenALPR or is it a direct relation to the resolution of the videostream that you are using?

Does this system take a snapshot of any vehicle or motion or just when it has a plate?

Does this system take the videostream and truly look for a plate and then take a pic only of plates or does it take a pic early like a large semi or bus enters the frame but the plate isn't visible yet?

Sorry for all the questions! I am just trying to figure out as much as I can before I start the trial period of OpenALPR and get my system as close as possible from the start to be usable.

You can specify the quality of the .jpgs in the .conf file. I never tinkered with it so its set for whatever the default is. It only takes the snapshot when it detects a plate. The agent is always analyzing the video stream. You configure your camera in the agent settings so the agent has a direct connection to the camera. When it detects that a plate comes into the scene that is when it reads it and takes the snapshot.

I have the agent running on a dedicated NUC with a 500GB drive in it. I guess I should modify the 8GB storage setting to something larger since I have plenty of space on that machine.