Manual Installation Steps
If for some reason, you prefer not to use Plate Recognizer Installer, you can go through the steps below to manually install Snapshot SDK with Docker.
For the commands below, make sure to change XXXXX with your License Key and YYYYY with your Plate Recognizer Token. Get your token and license key.
List of Docker Images​
There are Docker images for various architectures and engine customizations for some countries. The image structure is
platerecognizer/<architecture>:<country>-<version>
<architecture>
- x86
alpr
- x86 with Nvidia GPU
alpr-gpu
- Raspberry Pi
alpr-raspberry-pi
- Apple Silicon M1/M1 Pro use
alpr-raspberry-pi
- Nvidia Jetson (with GPU)
alpr-jetson
oralpr-jetson:r35
. We support 2 Jetpack versions:r35.3.1
andr32.7.1
(default). - ZCU
alpr-zcu104
- x86
<country>
: leave blank for global engine,thailand
,germany
,korea
,japan
,egypt
.<version>
: leave blank for latest.- To see the list of tags, select an image from Docker Hub and then click "Tags".
Here are a couple of examples:
- Image for x86 with a specific version:
platerecognizer/alpr:21-09-13
- Image for x86 customized for
thailand
:platerecognizer/alpr:thailand
- Image for Jetson:
platerecognizer/alpr-jetson
Installation​
Installation on an Intel CPU​
Abbreviated installation steps below. For a detailed, step-by-step guide, see the Install Docker and SDK on Windows and the associated FAQ for help.
- Sign up and log in.
- Subscribe to a SDK plan.
- Install Docker on your local machine. See requirements.
- Get our SDK image. In a terminal run
docker pull platerecognizer/alpr
- Install and run the container:
docker run --restart="unless-stopped" -t -p 8080:8080 -v license:/license \
-e TOKEN=4805bee*********** -e LICENSE_KEY=AX******** \
platerecognizer/alpr
If you use the engine for Thai vehicles, replace platerecognizer/alpr
by platerecognizer/alpr:thailand
.
Installation on a Raspberry Pi​
Follow the steps 1) to 3) of the standard installation instructions.
- Get our SDK image. In a terminal run
docker pull platerecognizer/alpr-raspberry-pi
- Run the container. Initialization can take up to 10-20 seconds.
Use the command 5) but with image
platerecognizer/alpr-raspberry-pi
Installation for Desktop GPU (Nvidia Only)​
If you have an Nvidia Jetson, follow these instructions. Make sure you have installed the right Nvidia drivers and have Nvidia-docker installed. Tested with drivers version 450. Follow the steps 1) to 3) of the standard installation instructions.
- Get our SDK image. In a terminal run
docker pull platerecognizer/alpr-gpu
- Run the container.
Option 1 (new version of nvidia-docker), notice the --gpus all
:
docker run --gpus all --restart="unless-stopped" -t -p 8080:8080 -v license:/license -e TOKEN=YYYYY -e LICENSE_KEY=XXXXX platerecognizer/alpr-gpu
Option 2 (Deprecated nvidia-docker2 version):
docker run --runtime nvidia --restart="unless-stopped" -t -p 8080:8080 -v license:/license -e TOKEN=YYYYY -e LICENSE_KEY=XXXXX platerecognizer/alpr-gpu
Installation for Nvidia Jetson​
Configure the Nvidia Jetson according to the official Jetson Nano Developer Kit Guide. Our image has been tested on Jetpack 4.6. For more Jetson information, check our FAQ.
Follow the steps 1) to 3) of the standard installation instructions.
- Get our SDK image. In a terminal run
docker pull platerecognizer/alpr-jetson
- Run the container. The initial prediction can take up to 2-3 minutes. After the first prediction, inference speed is around 300ms.
docker run --runtime nvidia --restart="unless-stopped" -t -p 8080:8080 \
-v license:/license -e TOKEN=YYYYY -e LICENSE_KEY=XXXXX \
platerecognizer/alpr-jetson
- Having problems running the container? Check our Jetson FAQ.
Previously available PROCESS_GPU_FRACTION
environment variable is no longer supported.
Installation for Kubernetes​
Refer to the documentation with Helm charts here