Blue Iris and DeepStack ALPR

[Futaba]

Both the LPR real camera and the cropped ALPR camera.

Thanks much! That is the reason why I wasn't getting the logfile.txt and the json files.

Would be great to make this clearer for others who will install the tool in the future. Something like:

For all cameras that you will create during this installation process, go to each of the cameras' "Camera Settings...", under the "Record" tab, uncheck "Use default" next to "Folder:" under "Video", and enter "&CAM\%Y%m%d\&CAM.%Y%m%d_%H%M%S" in the text edit field. This will save jpg files into the "Alerts" folder and in separate folders for each day's recording.

And insert this instruction during the camera creation steps. Thanks.

Additional suggestion for installation instruction:

• Might want to mention that SD_ALPR_config.ini is in "C:\Program Files (x86)\DeepStackALPR" if default folder was used during DeepStackALPR setup
• For the per camera AI dialog box, would be good to say:
  - Right click your LPR camera window, select "Camera Settings...", then click the "Trigger" tab, click the "Artificial Intelligence..." button. In the AI dialog box, change to the following settings.
• Note that Blue Iris's web server needs to be enabled. And the web server port number in DS_ALPR_config.ini need to match the BI's web server port set in the "Web server" tab in the "Blue Iris Settings" dialog box.
• Same for Deepstack's port number

 

[Psylent126]

Question...Does anyone know what the following indicates?



I know it was suggested that it was due to running two cameras with the API, but I am no longer doing so and I am still getting this error. from looking at the log, it appears that this is where things are breaking for me and I am not quite sure what the error signifies and can't find any references to it through searching.

Thanks.

 

[Psylent126]

Question...Does anyone know what the following indicates?

View attachment 130125

I know it was suggested that it was due to running two cameras with the API, but I am no longer doing so and I am still getting this error. from looking at the log, it appears that this is where things are breaking for me and I am not quite sure what the error signifies and can't find any references to it through searching.

Thanks.

So I am still at a loss with this. I do believe I have narrowed down when the process fails, but I am not sure what is different about my setup versus ones that work. This is what I have determined so far:

• A vehicle activates the motion trigger and a series of images are saved to D:\ALPR-Image\<CAM>\<Date/Time>\ (where <Cam>\<Date/Time> is &CAM\%Y%m%d\&CAM.%Y%m%d_%H%M%S as specified in the instructions).
• DeepStack activates and analyzes the images using the custom DayPlate/NightPlate model included with the DeepStackALPR download.
• If DeepStack finds a valid image, then it saves the alert image(s) and accompanying DAT files to my alerts folder at C:\Alerts\ALPRS\.
• The images saved by DeepStack are annotated with DayPlate or NightPlate as expected.
• This all shows in the Blue Iris log as expected.
• It is at this point that I seem to get the Plate: error 80002efd generated by the ALPRS camera as shown in my previous post (The ALPRS camera is my physical LPR camera).
• I never see the ALPR user account that I created for DeepStackALPR log in or do anything in the system log.
• I never see the ALPR camera (The ALPR camera is the virtual camera that is supposed to do the crop and such) mentioned in the Blue Iris Log file.
• No files are saved to the C:\Program Files\Blue Iris 5\www\ALPR\ folder as specified in the DS_ALPR_config.ini.
• The 'file path.txt' file is not created or updated.
• No files are saved to the C:\Alerts\ALPR\* folder as specified in the DS_ALPR_config.ini file.

I have checked the permissions on the install folder for DeepStackALPR to ensure that it is writeable and that all the accounts on the PC have full access to read/write/modify/execute files and folder in that folder and subfolders. I have also verified that the firewall setting for DeepStackALPR allow all incoming UDP and TCP connections for the program. I have also checked and verified that the DeepStackALPR user specified in the DS_ALPR_Config.ini file has admin rights in Blue Iris. From what I can tell the DeepStackALPR service is running and has spawned to processes that appear to run simultaneously but don't appear to actually do anything when called. I have tried shutting down the service and manually running DeepStackALPR.exe on port 5000 from the command line as referenced in another post in this thread, but after it starts, nothing further happens (even when vehicles trigger the DeepStack AI in Blue Iris and alerts are generated.

As far as I can tell everything has been set up according to the instructions written by MikeLud1, but I just can't get this thing to work. Does anyone have any further suggestions or notice something that I may have missed?

Any help would be very appreciated as I have been trying to get this to work for the better part of a week now without much luck.

Thank you.

 

[Psylent126]

Question...Does anyone know what the following indicates?

View attachment 130125

I know it was suggested that it was due to running two cameras with the API, but I am no longer doing so and I am still getting this error. from looking at the log, it appears that this is where things are breaking for me and I am not quite sure what the error signifies and can't find any references to it through searching.

Thanks.

So I am still at a loss with this. I do believe I have narrowed down when the process fails, but I am not sure what is different about my setup versus ones that work. This is what I have determined so far:

• A vehicle activates the motion trigger and a series of images are saved to D:\ALPR-Image\<CAM>\<Date/Time>\ (where <Cam>\<Date/Time> is &CAM\%Y%m%d\&CAM.%Y%m%d_%H%M%S as specified in the instructions).
• DeepStack activates and analyzes the images using the custom DayPlate/NightPlate model included with the DeepStackALPR download.
• If DeepStack finds a valid image, then it saves the alert image(s) and accompanying DAT files to my alerts folder at C:\Alerts\ALPRS\.
• The images saved by DeepStack are annotated with DayPlate or NightPlate as expected.
• This all shows in the Blue Iris log as expected.
• It is at this point that I seem to get the Plate: error 80002efd generated by the ALPRS camera as shown in my previous post (The ALPRS camera is my physical LPR camera).
• I never see the ALPR user account that I created for DeepStackALPR log in or do anything in the system log.
• I never see the ALPR camera (The ALPR camera is the virtual camera that is supposed to do the crop and such) mentioned in the Blue Iris Log file.
• No files are saved to the C:\Program Files\Blue Iris 5\www\ALPR\ folder as specified in the DS_ALPR_config.ini.
• The 'file path.txt' file is not created or updated.
• No files are saved to the C:\Alerts\ALPR\* folder as specified in the DS_ALPR_config.ini file.

I have checked the permissions on the install folder for DeepStackALPR to ensure that it is writeable and that all the accounts on the PC have full access to read/write/modify/execute files and folder in that folder and subfolders. I have also verified that the firewall setting for DeepStackALPR allow all incoming UDP and TCP connections for the program. I have also checked and verified that the DeepStackALPR user specified in the DS_ALPR_Config.ini file has admin rights in Blue Iris. From what I can tell the DeepStackALPR service is running and has spawned to processes that appear to run simultaneously but don't appear to actually do anything when called. I have tried shutting down the service and manually running DeepStackALPR.exe on port 5000 from the command line as referenced in another post in this thread, but after it starts, nothing further happens (even when vehicles trigger the DeepStack AI in Blue Iris and alerts are generated.

As far as I can tell everything has been set up according to the instructions written by MikeLud1, but I just can't get this thing to work. Does anyone have any further suggestions or notice something that I may have missed?

Any help would be very appreciated as I have been trying to get this to work for the better part of a week now without much luck.

Thank you.

Just wanted to post a follow-up on the issues that I was having. So after much troubleshooting and experimentation, I determined that Windows appeared to be blocking the DeepStackALPR.exe from executing. I was able to go in an create an exclusion for the DeepStackALPR process and folder within Windows Defender. After creating the exclusions, the application performed as expected. Now I am working on fine tuning the triggers for the plate capture and trying to improve the OCR accuracy.

If any one has any suggestions on how to better improve the OCR accuracy or improved OCR models they can point me to, I would appreciate the assistance.

Thank you.

 

[Futaba]

ALPR has been working very well. There are a couple things I wonder if they can be improved on:

• When one car is followed closely by another car, the second car's plate is not detected. Might this be improved with BI settings?
• Sometimes OCR misreads license plate characters
• automatically rotate license plate capture. Cars that I capture around my neighborhood circle are always turning left. Depends on which picture was used to get the plate, sometimes it is not rotated enough, sometimes it is rotated too much. Not really a big deal, just a nit pick

 

[Psylent126]

ALPR has been working very well. There are a couple things I wonder if they can be improved on:

• When one car is followed closely by another car, the second car's plate is not detected. Might this be improved with BI settings?
• Sometimes OCR misreads license plate characters
• automatically rotate license plate capture. Cars that I capture around my neighborhood circle are always turning left. Depends on which picture was used to get the plate, sometimes it is not rotated enough, sometimes it is rotated too much. Not really a big deal, just a nit pick

I would also add that it would be nice to have the ability to use more than one LPR camera with the application.

 

[Futaba]

The vast majority was detected correctly. Some of my misidentified "plates":

 

[Psylent126]

MikeLud1,

Two questions...

1. Does your program currently work with the new BI integrated SenseAI or strictly DeepStack?
2. Does your program work with more than one camera at a time? I believe from previous posts you indicated that it might not, but I wanted to verify.

Thank you.

 

[MikeLud1]

MikeLud1,

Two questions...

1. Does your program currently work with the new BI integrated SenseAI or strictly DeepStack?
2. Does your program work with more than one camera at a time? I believe from previous posts you indicated that it might not, but I wanted to verify.

Thank you.

1. Does your program currently work with the new BI integrated SenseAI or strictly DeepStack?

At this time only DeepStack, I am going to try to make a SenseAI ALPR module once SenseAI has GPU support..

2. Does your program work with more than one camera at a time? I believe from previous posts you indicated that it might not, but I wanted to verify.

Only works with one camera.

 

[Psylent126]

At this time only DeepStack, I am going to try to make a SenseAI ALPR module once SenseAI has GPU support..

Only works with one camera.

Hello MikeLud1,

Just checking to see if now that there is GPU support available for SenseAI, if you have made any progress in converting your ALPR module so that it works for it?

It was a wonderful tool to have with Deepstack and I really hope that you are able to convert it to work with SenseAI as well.

 

[MikeLud1]

Hello MikeLud1,

Just checking to see if now that there is GPU support available for SenseAI, if you have made any progress in converting your ALPR module so that it works for it?

It was a wonderful tool to have with Deepstack and I really hope that you are able to convert it to work with SenseAI as well.

I am using the current version with CodeProject.AI. All you need to do is to add the two highlight models to the folder shown below. Then change the ports as show below.
I am going to try and make a CodeProject.AI ALPR module, I should hopefully have done sometime next month.

 

[pbc]

Okay, decided to give this great project a go, but seem to be having some issues in terms of setup. I've installed the executable, and I think did everything within the pdf instructions. But the new "camera" doesn't seem to be connecting properly and/or saving images. Some shots below:


What I'm seeing with the newly added LPRPlate Camera:






LPR Camera Setup:




LPRPlate "Camera" Setup:


Config file:



Note: I didn't do anything with the NSSM in the ALPR directory, not sure if I am supposed to run ALPR or adjust the NSSM to run it automatically (didn't see anything in the instructions)?

Maybe I have one of my IP Addresses or ports off?

 

[MikeLud1]

Okay, decided to give this great project a go, but seem to be having some issues in terms of setup. I've installed the executable, and I think did everything within the pdf instructions. But the new "camera" doesn't seem to be connecting properly and/or saving images. Some shots below:


What I'm seeing with the newly added LPRPlate Camera:
View attachment 140090
View attachment 140092View attachment 140093




LPR Camera Setup:
View attachment 140094View attachment 140095
View attachment 140099


LPRPlate "Camera" Setup:
View attachment 140096View attachment 140097

Config file:
View attachment 140102


Note: I didn't do anything with the NSSM in the ALPR directory, not sure if I am supposed to run ALPR or adjust the NSSM to run it automatically (didn't see anything in the instructions)?

Maybe I have one of my IP Addresses or ports off?

Post a screenshot of the LPRPlate configuration settings

 

[pbc]

Post a screenshot of the LPRPlate configuration settings

Different screen shots vs the ones posted above?

 

[MikeLud1]

Different screen shots vs the ones posted above?

View attachment 140122View attachment 140123

Click on the below Configure... and post this screenshot.

 

[pbc]

Click on the below Configure... and post this screenshot.

View attachment 140124

Here you go (and thank you!)

 

[pbc]

Just going back through the instructions pdf. For this step....



Is this done in the "Alerts" tab, or in the Trigger tab where it says "immediate actions", and the "Configure Run action"? Mine looks like this, but had it in the Alert tab vs the Immediate Actions section under Trigger.

Also, under Trigger, for the dummy camera that is set to do the cropped images, should I be activating AI? I had assumed no as it didn't show in the instructions, but now not sure.

 

[pbc]

Coule other things: Should this be empty? I.e., is it only populated when I drag a .dat file into it?



In the setup pdf for ALPR, the "default object detection" is not checked, shoudl I uncheck mine? I am also using AI for person, etc detections on other cameras.

 

[MikeLud1]

Just going back through the instructions pdf. For this step....

View attachment 140136

Is this done in the "Alerts" tab, or in the Trigger tab where it says "immediate actions", and the "Configure Run action"? Mine looks like this, but had it in the Alert tab vs the Immediate Actions section under Trigger.

Also, under Trigger, for the dummy camera that is set to do the cropped images, should I be activating AI? I had assumed no as it didn't show in the instructions, but now not sure.

View attachment 140137

Did you create the below folder and placed the attached APLR.JPG in it? I also attached my config file if that helps

 

[JNDATHP]

View attachment 140137

For Parameters, you have &ALERT+PATH

Note the + sign.

I have the underscore sign. &ALERT_PATH