An Automatic License Plate Recognition tool based on the ultimateALPR-SDK [0].
Please find the info for building/using ultimateALPR-SDK at [0]. It applies to this tool as well.
-
Clone the ultimateALPR:
$ git clone https://github.com/DoubangoTelecom/ultimateALPR-SDK.git
-
Download the tensorflow and copy it into the
binaries/linux/x86_64
folder (https://github.com/DoubangoTelecom/ultimateALPR-SDK/blob/master/samples/c%2B%2B/README.md#linux).$ tar -C ${path-to-the-ultalpr}/ultimateALPR-SDK/binaries/linux/x86_64/ -xf ~/Downloads/libtensorflow_r1.14_cpu+gpu_linux_x86-64.tar.gz
-
Set up build of the lp-recognizer:
$ mkdir build && cd build $ cmake ../lp-recognizer/ -DCMAKE_C_FLAGS="-L${path-to-the-ultalpr}/ultimateALPR-SDK/binaries/linux/x86_64 -I${path-to-the-ultalpr}/ultimateALPR-SDK/c++ -I/${path-to-the-ultalpr}/ultimateALPR-SDK/samples/c++" -DCMAKE_CXX_FLAGS="-L/${path-to-the-ultalpr}/ultimateALPR-SDK/binaries/linux/x86_64 -I${path-to-the-ultalpr}/ultimateALPR-SDK/c++ -I${path-to-the-ultalpr}/ultimateALPR-SDK/samples/c++" $ make
-
Running the lp-recognizer:
Before we run the tool install openCL library:
$ sudo apt install ocl-icd-opencl-dev
$ export LD_LIBRARY_PATH=${path-to-the-ultalpr}/ultimateALPR-SDK/binaries/linux/x86_64
$ cd ../examples/
$ ../build/lp-recognizer --assets ${path-to-the-ultalpr}/ultimateALPR-SDK/assets --config ../config.json --port 8989
The server expects just paths to the images to be processed. The client will recieve a JSON object that containts the outcome of using the [0] library.