DeepStack LPR Custom Model

How is the License-Plate model performing fo you?


  • Total voters
    12

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
1,157
Reaction score
2,578
Location
Brooklyn, NY
Attached is a DeepStack custom model that can be used to confirm if a license plate is in the FOV of your LPR camera. The DeepStack labels for the model are "DayPlate" and "NightPlate".

Let me know how the custom model works for you. If you are having poor results and want to add your LPR images to the model let me know and I will post instructions on how to label the images so I can added them to the custom model.

This model was created with the help of @aesterling. Below are the states where the plates were captured and the amount of images used to train the model.
State Where Plates Were CapturedDay Plate Count Used For Training the Custom ModelNight Plate Count Used For Training the Custom Model
New York530530
Minnesota291103

1636905203071.png

1636905418204.png
 

Attachments

Last edited:

wittaj

IPCT Contributor
Joined
Apr 28, 2019
Messages
15,106
Reaction score
28,160
Location
USA
Awesome. Thanks for creating this! This is a great model for those using Plate Recognizer with BI.

So the items we put in DS are

Day Plate
Night Plate
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
1,157
Reaction score
2,578
Location
Brooklyn, NY
Awesome. Thanks for creating this! This is a great model for those using Plate Recognizer with BI.

So the items we put in DS are

Day Plate
Night Plate
Below is how I have the DS AI Trigger set
AI Setting.jpg

Also I have an Alert Action to trigger an overview camera.
Alert Action.jpg
 

aesterling

Getting comfortable
Joined
Oct 9, 2017
Messages
331
Reaction score
306
Nice job, @MikeLud1, glad I could contribute my LPR images to help train the model. This is what I've been using to only submit LPR images with a confirmed license plate to Plate Recognizer to stay below the monthly quota. The addition of night images helps, as well!

Thanks!
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
1,157
Reaction score
2,578
Location
Brooklyn, NY
There is a new version of the DeepStack LPR Custom Model in the first post. There was a BI issue with having spaces in the labels, so the new model labels are "DayPlate" and "NightPlate".

Attached is a DeepStack custom model that can be used to confirm if a license plate is in the FOV of your LPR camera. The DeepStack labels for the model are "DayPlate" and "NightPlate".

Let me know how the custom model works for you. If you are having poor results and want to add your LPR images to the model let me know and I will post instructions on how to label the images so I can added them to the custom model.

This model was created with the help of @aesterling. Below are the states where the plates were captured and the amount of images used to train the model.
State Where Plates Were CapturedDay Plate Count Used For Training the Custom ModelNight Plate Count Used For Training the Custom Model
New York530530
Minnesota291103
View attachment 108342








View attachment 108343
 

dimammx

Getting the hang of it
Joined
Apr 9, 2015
Messages
127
Reaction score
68
I was talking about Mark As Vehicle: option which sounds like an option too use custom tags to tell BI what you want to consider a vehicle
 

dimammx

Getting the hang of it
Joined
Apr 9, 2015
Messages
127
Reaction score
68
Not to hijack the tread but anyone running more then 1 custom Deeps tack models? Whenever i have more than 1 custom model processing time increases to 2 sec and after few requests just freezes.
 

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
1,157
Reaction score
2,578
Location
Brooklyn, NY
Not to hijack the tread but anyone running more then 1 custom Deeps tack models? Whenever i have more than 1 custom model processing time increases to 2 sec and after few requests just freezes.
You are probably using the CPU version of DeepStack and it can not handle more than one model
 

dimammx

Getting the hang of it
Joined
Apr 9, 2015
Messages
127
Reaction score
68
You are probably using the CPU version of DeepStack and it can not handle more than one model
No it is GPU with under 100ms average processing time it is just as soon as 2nd custom model added times increase to 2-3 seconds and shell freezes, so just trying to see if people are able to run 2 models or more.
1637645756868.png
 
Last edited:

MikeLud1

IPCT Contributor
Joined
Apr 5, 2017
Messages
1,157
Reaction score
2,578
Location
Brooklyn, NY
Attached is a DeepStack custom model that can be used to confirm if a license plate is in the FOV of your LPR camera. The DeepStack labels for the model are "DayPlate" and "NightPlate".

Let me know how the custom model works for you. If you are having poor results and want to add your LPR images to the model let me know and I will post instructions on how to label the images so I can added them to the custom model.

This model was created with the help of @aesterling. Below are the states where the plates were captured and the amount of images used to train the model.
State Where Plates Were CapturedDay Plate Count Used For Training the Custom ModelNight Plate Count Used For Training the Custom Model
New York530530
Minnesota291103
View attachment 108342








View attachment 108343
There is a new version of the DeepStack LPR Custom Model in the first post. This version should see a big improvement in the analyze times for CPU version of DS and minor improvement in the analyze times for GPU version of DS
 

aesterling

Getting comfortable
Joined
Oct 9, 2017
Messages
331
Reaction score
306
There is a new version of the DeepStack LPR Custom Model in the first post. This version should see a big improvement in the analyze times for CPU version of DS and minor improvement in the analyze times for GPU version of DS
That's great! What allows for these improvements?
 
Top