FastDeploy/examples/application/java/yolov8

English | 简体中文

YOLOv8 Java Deployment Example

This directory provides examples that java/InferDemo.java uses Java to call FastDeploy C++ API and finish the deployment of YOLOv8 model。

Before deployment, two steps require confirmation

• 1. Software and hardware should meet the requirements. Please refer to FastDeploy Environment Requirements
• 2. Download the precompiled deployment library and samples code according to your development environment. Refer to FastDeploy Precompiled Library

Using Java to call C++ API can be divided into two steps:

• Generate dynamic link library in C++ side.
• Call the dynamic link library in Java side.

Generates dynamic link library

First, switch the path to the cpp directory and copy jni.h and jni_md.h which in jdk directory to current directory cpp.

cp /PathJdk/jdk-17.0.6/include/jni.h ./
cp /Pathjdk/jdk-17.0.6/include/linux/jni_md.h ./

Then, execute the following command in the cpp directory to compile and generate the dynamic link library.

Note: you will need to specify the location of the FASTDEPLOY_INSTALL_DIR pre-compile library at compile time, but also the location of your own compiled FastDeploy library.

mkdir build && cd build
cmake .. -FASTDEPLOY_INSTALL_DIR /fast-deploy-path
make -j

After successful compilation, the dynamic link library will be stored in the cpp/build directory, ending in .so under Linux and .dll under Windows.

Invoke dynamic link libraries using JAVA

Switch the path to the java directory and use the following command to add Fastdeploy library path to the environment variable. Note the path of the FastDeploy library replaced with your own.

source /Path/to/fastdeploy-linux-x64-1.0.4/fastdeploy_init.sh

Download the YOLOv8 model file and test images.

wget https://bj.bcebos.com/paddlehub/fastdeploy/yolov8s.onnx
wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/000000014439.jpg

Compiling Java files.

javac InferDemo.java

After compiling, run the following command to get the predicted result, where the first parameter indicates the path of the downloaded model and the second parameter indicates the path of the test image.

java InferDemo ./yolov8s.onnx ./000000014439.jpg

Then visualized inspection result is saved in the local image vis_result.jpg.