HyperLPR3 Documentation

To cross-compile Android shared libraries, ensure the $ANDROID_NDK environment variable is set. Run the Android build script. The resulting libraries for arm64-v8a and armeabi-v7a will be placed in build/release_android/.

sh command/build_release_android_share.sh

To quickly integrate HyperLPR3 into an Android project, add the JitPack repository and the HyperLPR3 dependency to your build.gradle files, then initialize and use the SDK.

# Step 1: Add JitPack to your root build.gradle
allprojects {
	repositories {
		...
		maven { url 'https://jitpack.io' }
	}
}

# Step 2: Add dependency to your app build.gradle
dependencies {
	implementation 'com.github.HyperInspire:hyperlpr3-android-sdk:1.0.3'
}

# Step 3: Use in your code
HyperLPR3.getInstance().init(this, new HyperLPRParameter());

// Recognize a plate from a bitmap
Plate[] plates = HyperLPR3.getInstance().plateRecognition(bitmap, HyperLPR3.CAMERA_ROTATION_0, HyperLPR3.STREAM_BGRA);

To compile the C++ shared library for Linux or Mac, ensure you have placed or linked third-party dependencies in the project root (same level as CMakeLists.txt).

Note: When compiling C++ projects, OpenCV 4.0 and MNN 2.0 must be used.

Run the build script:

sh command/build_release_linux_share.sh

Output is located in build/linux/install/hyperlpr3, containing include, lib, and resource directories.

To compile the C++ shared libraries for Linux or Mac, ensure third-party dependencies (OpenCV 4.0+ and MNN 2.0+) are placed in the project root. Run the provided build script.

sh command/build_release_linux_share.sh

To install the HyperLPR3 Python package, use the following command. It supports Python 3 and is compatible with Windows, Mac, Linux, and Raspberry Pi.

python -m pip install hyperlpr3

To quickly integrate the HyperLPR3 SDK into an Android project, add the JitPack repository and the dependency in your build.gradle files.

Step 1: Add JitPack to root build.gradle

allprojects {
    repositories {
        ...
        maven { url 'https://jitpack.io' }
    }
}

Step 2: Add dependency

dependencies {
    implementation 'com.github.HyperInspire:hyperlpr3-android-sdk:1.0.3'
}

Step 3: Usage Initialize the instance once (e.g., at app startup) and then call plateRecognition with a bitmap.

// Initialization
HyperLPR3.getInstance().init(this, new HyperLPRParameter());

// exec recognition
Plate[] plates = HyperLPR3.getInstance().plateRecognition(bitmap, HyperLPR3.CAMERA_ROTATION_0, HyperLPR3.STREAM_BGRA);

You can use Docker to compile the Linux-x86 shared library. First, build the image, then use docker-compose to run the compilation.

1. Build the image:

docker build -t hyperlpr_build .

2. Start compilation:

docker-compose up build_linux_x86_shared_lib

Output is located in build/linux.

To compile the Android shared library, you must have the NDK installed and the $ANDROID_NDK environment variable configured. It is recommended to use NDK version 21.

Run the build script:

sh command/build_release_android_share.sh

Output is located in build/release_android/ containing arm64-v8a (64-bit) and armeabi-v7a (32-bit) directories.

Install the HyperLPR3 package using pip. This version supports Python 3 on Windows, Mac, Linux, and Raspberry Pi.

Python Dependencies:

• opencv-python (>3.3)
• onnxruntime (>1.8.1)
• fastapi (0.92.0)
• uvicorn (0.20.0)
• loguru (0.6.0)
• python-multipart
• tqdm
• requests

python -m pip install hyperlpr3

Use the lpr3 command-line interface to test license plate recognition on images via URL or local file path. Use the -det flag to specify detection level (e.g., high).

# image url
lpr3 sample -src https://koss.iyong.com/swift/v1/iyong_public/iyong_2596631159095872/image/20190221/1550713902741045679.jpg

# image path
lpr3 sample -src images/test_img.jpg -det high