IPC-HFW5241E-Z12E LPR Blue Iris Triggering Help Needed

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,141
Reaction score
4,118
Location
Brooklyn, NY
I just installed my IPC-HFW5241E-Z12E LPR today and need some help with triggering. I have a IPC-T5442T-ZE (LPR FOV is the red box) as an overview that I could use for triggering. I have the camera dialed in with good day and night images just need help with the triggering.
Is it better to use an overview camera, ONVIF from the LPR camera, or Blue Iris to do the triggering.

Thanks,
Mike

LPR.jpg
@395777237.jpg
Camera 6 2021-11-04 10.07.06.775 PM.jpg
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,435
Reaction score
47,557
Location
USA
IVS ONVIF triggering is problematic for LPR duty, especially at night with the black image.

Since you have a great camera overview already, I would suggest setting up a clone of that camera to trigger the LPR camera.

You can also set it up as motion in BI.

I found this to be the setup that worked best for me. The one zone approach at this location allowed me to catch the front plate of the cars going right to left and the back plate of the cars going left to right. Occasionally I would get a large truck or a very fast vehicle that was missed, but I also run this continuous motion in the event that happens or it missed the motion. And then at night I simply make the object size box a lot smaller like the size of the plate.

1636079501175.png

The other way is since this thread originated, BI now has DeepStack integrated with it. It has added the ability to only send when AI detects a vehicle. You can assign a leading edge trigger now as well with DeepStack, But some fields of view are still problematic and you could do as another member did here and create a custom model based on your field of view that will only take a snapshot when a plate is in view. Member said it took about an hour or so to train the custom model.
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,141
Reaction score
4,118
Location
Brooklyn, NY
IVS ONVIF triggering is problematic for LPR duty, especially at night with the black image.

Since you have a great camera overview already, I would suggest setting up a clone of that camera to trigger the LPR camera.

You can also set it up as motion in BI.

I found this to be the setup that worked best for me. The one zone approach at this location allowed me to catch the front plate of the cars going right to left and the back plate of the cars going left to right. Occasionally I would get a large truck or a very fast vehicle that was missed, but I also run this continuous motion in the event that happens or it missed the motion. And then at night I simply make the object size box a lot smaller like the size of the plate.

View attachment 107380

The other way is since this thread originated, BI now has DeepStack integrated with it. It has added the ability to only send when AI detects a vehicle. You can assign a leading edge trigger now as well with DeepStack, But some fields of view are still problematic and you could do as another member did here and create a custom model based on your field of view that will only take a snapshot when a plate is in view. Member said it took about an hour or so to train the custom model.
Thanks Wittaj,

I just set up a LPR Group with the clone overview and LPR Camera. In the clone I set the motion to trigger with a zone crossing, I will see how this work overnight.

Thanks
Mike
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,141
Reaction score
4,118
Location
Brooklyn, NY
I had to change the overview camera to another one for triggering the one I setup last night was not triggering reliably enough to get a good LPR. Below are some LPR images, some of the images are nice and clear some are not. Is there any tweaks to get a better image or is it just a bad license plate.

@453015078.jpg
@453030049.jpg
@452968892.jpg
@452974316.jpg
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,435
Reaction score
47,557
Location
USA
Those look pretty good.

There will be some plates that are problematic - either an out of state plate that the numbers/letters are printed instead of raised, temp paper tags, dirty/rusty or a cover on them etc.

Post a screenshot of your settings.

Some things that come to mind would be:
  • slowing down the shutter to let more light in. I probably wouldn't drop below 1/1,000
  • is your shutter a fixed number or a range? I have found the range to be problematic.
  • adjust a bigger spread with contrast and brightness to help the image pop.
  • adjust NR - try both ways - a lower number and a higher number. I run mine lower and it seems to help it pop a little brighter.
  • try to keep the gain to a small spread or the same number in both blocks
 

Flintstone61

Known around here
Joined
Feb 4, 2020
Messages
6,587
Reaction score
10,894
Location
Minnesota USA
very tough angle. Your duration of time to catch a plate is going to be very short. But you are getting them. SO thats a Win.
I have a longer angle, and so if the lighting and plates are funkified somehow, I get a few more nanoseconds to reacquire the plate at longer or shorter focal distance ( if they're coming or going)
and usually 99% of them are readable. My worst time of day is afternoon with the sun moving in and out of the tree shadows. thats when I miss a few. But my criminal activity is largely at night out back.
You are dialed in pretty good. I hope you can get a higher %age of reads.
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,141
Reaction score
4,118
Location
Brooklyn, NY
Those look pretty good.

There will be some plates that are problematic - either an out of state plate that the numbers/letters are printed instead of raised, temp paper tags, dirty/rusty or a cover on them etc.

Post a screenshot of your settings.

Some things that come to mind would be:
  • slowing down the shutter to let more light in. I probably wouldn't drop below 1/1,000
  • is your shutter a fixed number or a range? I have found the range to be problematic.
  • adjust a bigger spread with contrast and brightness to help the image pop.
  • adjust NR - try both ways - a lower number and a higher number. I run mine lower and it seems to help it pop a little brighter.
  • try to keep the gain to a small spread or the same number in both blocks
I had my shutter set to fixed 1/2000. I will try fixed 1/1000. I will also try adjusting the contrast and brightness. I have my NR and sharpness set low, for now I will leave them alone. The gain set to 60 same number in both blocks.

Thanks,
Mike
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,141
Reaction score
4,118
Location
Brooklyn, NY
The plate on the left is a parked car and the right is moving with the new settings. I will see how the new setting work overnight.

@453455600.jpg
 

biggen

Known around here
Joined
May 6, 2018
Messages
2,539
Reaction score
2,765
I think you are dialed in fine. As @wittaj noted, some plates are just mangled, dirty, not reflective enough, etc... This lead to me actually installing external IR to really blast those dirty plates that the camera built-in IR just wasn't powerful enough to illuminate. I now get better than 99% of plate reads of cars going by at night whereas I missed a handful each night before without. Basically, if the plate is viewable now and not hidden by a trailer, I capture it.

I also use my overview 5442 setup in BI to trigger my LPR cam to record. That way my LPR doesn't record blank road for hours and hours a day. My overview cam is always recording but sending a trigger upon motion to the LPR cam for it to begin recording for 20 seconds. Which is more than enough time for the car to clear the FOV.
 
Last edited:

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,435
Reaction score
47,557
Location
USA
Also remember at these tighter zoom and field of view that the focus is more defined to the area you focused it, so a parked car on the fringe of the field of view will be not as clear as a plate in the target zone.
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,141
Reaction score
4,118
Location
Brooklyn, NY
Those look pretty good.

There will be some plates that are problematic - either an out of state plate that the numbers/letters are printed instead of raised, temp paper tags, dirty/rusty or a cover on them etc.

Post a screenshot of your settings.

Some things that come to mind would be:
  • slowing down the shutter to let more light in. I probably wouldn't drop below 1/1,000
  • is your shutter a fixed number or a range? I have found the range to be problematic.
  • adjust a bigger spread with contrast and brightness to help the image pop.
  • adjust NR - try both ways - a lower number and a higher number. I run mine lower and it seems to help it pop a little brighter.
  • try to keep the gain to a small spread or the same number in both blocks
To All,

Thanks for all the help setting up my LPR. Below are my final night camera settings also some images. The hardest part was setting up the zone crossing triggering in the overview camera, this was the first time setting BI triggering, all my other cameras I am triggering with IVS from the cameras.
I am running DeepStack, is there any improvement that I can use DeepStack for LPR. Also what are other AI options that I can run local on my Blue Iris server.

Thanks
Mike
1636255834775.png1636255863100.png1636255899930.png
1636255929353.png1636255958916.png1636255986676.png
1636256022053.png1636256049643.png
1636257935072.png1636258020115.png
 

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
24,435
Reaction score
47,557
Location
USA
Those images look even better than before!

You could train a Deepstack model to look for the plates in your field of view and then trigger based on that.

A member here did that and said it was about an hour or so taking his photos and drawing the rectangle around the plates to train the model.
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
2,141
Reaction score
4,118
Location
Brooklyn, NY
Those images look even better than before!

You could train a Deepstack model to look for the plates in your field of view and then trigger based on that.

A member here did that and said it was about an hour or so taking his photos and drawing the rectangle around the plates to train the model.
I might give it a try when I get more images to train a Deepstack model.
 
Top