Openalpr: Is there a script that does the steps for GPU compile and install?

[moeiscool]

I see a lot of steps here to install with GPU https://github.com/openalpr/openalpr/wiki/Compilation-instructions-(Ubuntu-Linux) i was wondering if there is a script to do it all at once.

perhaps an apt install ..? just like non GPU version?

[moeiscool]

So it turns out figuring out OpenALPR with NVIDIA GPU is a lot harder than it's supposed to be. So for anyone who is left hanging like I was you can check out these scripts. I tested with a Zotac NVIDIA GT 1030.

if running nvidia-smi in terminal displays data about your graphics card you can skip steps 1 and 2.

1. First you need to install the NVIDIA CUDA drivers.

sudo apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64/7fa2af80.pub
sudo echo "deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /" | sudo tee -a /etc/apt/sources.list
sudo apt update
sudo apt -y install cuda-drivers
sudo reboot

1. Install CUDA Toolkit

sudo apt-get -y install cuda-toolkit-9-1

1. so of course you need OpenCV as well. Here is the install script for that https://github.com/ShinobiCCTV/Shinobi/blob/dev/INSTALL/opencv-cuda.sh

2. Here is the install script for OpenALPR. It will compile with GPU capability if your drivers and OpenCV are installed correctly.
   https://github.com/ShinobiCCTV/Shinobi/blob/dev/INSTALL/openalpr-gpu-easy.sh

3. Enjoy and forget asking why this wasn't provided in the first place!

Update

You can review a formalized instruction for this here https://shinobi.video/docs/object

[Raven888888]

I am working on Openalpr on jetson nano.
I installed Openalpr via

echo "deb https://deb.openalpr.com/snapshot/jetson40-2.7.101/ jetson40 main" > /etc/apt/sources.list.d/openalpr.list apt-get update && apt-get install -y openalpr libopenalpr-dev libalprstream-dev libvehicleclassifier-dev

However, when I try alpr <imagefile> --gpu 1 I get this error
ERROR loading detection library -- libalpredgegpu.so.2

Is the step provided by @moeiscool safe to use in jetson nano?

[moeiscool]

@Raven888888 if you want it running on Jetson Nano you should use these scripts

1. Install OpenCV 3.4, OpenALPR needs it

https://gitlab.com/Shinobi-Systems/Shinobi/blob/master/INSTALL/jeston-opencv3-4.sh

1. Build OpenALPR from source

https://gitlab.com/Shinobi-Systems/Shinobi/blob/master/INSTALL/openalpr-gpu-easy.sh

thats it. Jetson already has CUDA drivers.

[Raven888888]

@moeiscool Thank you very much. I tried your method and it is working amazing now!

I needed to do some further debugging due to unmet dependencies and missing libraries when I integrate it into Deepstream. This is only specific to my case and you should not bother if you don't plan to run it with Deepstream.

From 1 fps to 20+fps on Jetson nano! Cheers!

[moeiscool]

@Raven888888 i will gladly update my scripts for deepstream support :) please let me know what I need to do

[saltzmanjoelh]

@moeiscool Thank you very much. I tried your method and it is working amazing now!

I needed to do some further debugging due to unmet dependencies and missing libraries when I integrate it into Deepstream. This is only specific to my case and you should not bother if you don't plan to run it with Deepstream.

From 1 fps to 20+fps on Jetson nano! Cheers!

Thanks @moeiscool for the install scripts!

@Raven888888 How are you getting 20 fps on a Jetson Nano? I used the same build scripts and ran the test.py script included in the source and I'm getting Processing Time: 843.126160.

[moeiscool]

it’s probably because the python script is pointed at cpu by default. you
need to specify lbgpu in openalpr conf.

On Sat, Mar 14, 2020 at 9:38 PM saltzmanjoelh [e-mail]
wrote:

@moeiscool https://github.com/moeiscool Thank you very much. I tried
your method and it is working amazing now!

I needed to do some further debugging due to unmet dependencies and
missing libraries when I integrate it into Deepstream. This is only
specific to my case and you should not bother if you don't plan to run it
with Deepstream.

From 1 fps to 20+fps on Jetson nano! Cheers!

Thanks @moeiscool https://github.com/moeiscool for the install scripts!

@Raven888888 https://github.com/Raven888888 How are you getting 20 fps
on a Jetson Nano? I used the same build scripts and ran the test.py script
included in the source and I'm getting Processing Time: 843.126160.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/openalpr/openalpr/issues/635#issuecomment-599167025,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AB6USQFBHPSEYMGTZLCG7FDRHRLSPANCNFSM4ETJIU6A
.

>

Thanks!
Moe Alam
Shinobi Systems
CEO & Founder
shinobi.video

[saltzmanjoelh]

@moeiscool I made sure that it is using the gpu setting. It even logs “loading cuda classifier”

Are you using the cli binary, writing your own source code or something else?

[moeiscool]

i’ve used both CLI and the nodejs module. i don’t use python

On Sun, Mar 15, 2020 at 8:20 AM saltzmanjoelh [e-mail]
wrote:

@moeiscool https://github.com/moeiscool I made sure that it is using
the gpu setting. It even logs “loading cuda classifier”

Are you using the cli binary, writing your own source code or something
else?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/openalpr/openalpr/issues/635#issuecomment-599223994,
or unsubscribe
https://github.com/notifications/unsubscribe-auth/AB6USQECTJ4EKLPMJWVDRHLRHTW25ANCNFSM4ETJIU6A
.

>

Thanks!
Moe Alam
Shinobi Systems
CEO & Founder
shinobi.video

[saltzmanjoelh]

Thanks @moeiscool. With cli, would you mind sending the output with the —debug flag so that I can compare please? 🙏

[TrueWodzu]

1. Install OpenCV 3.4, OpenALPR needs it

https://gitlab.com/Shinobi-Systems/Shinobi/blob/master/INSTALL/jeston-opencv3-4.sh

@moeiscool Hi, may I ask why 3.4? The official documentation states that it needs 2.4.