mmdeploy/demo/java

Usage

step 1. Install Apache Ant.

Please check the Apache Ant existence using ant -h in the command line.

If there is no Apache Ant installed, please follow the command below.

sudo apt-get update
sudo apt-get install ant

Set environment variable

export ANT_HOME=/usr/share/ant
export PATH=${ANT_HOME}/bin:${PATH}

step 2. Build OpenCV jar package (PoseTracker only).

PoseTracker demo needs OpenCV Java, if you are interested in PoseTracker demo, you need to build OpenCV jar package first.

Using OpenCV-4.7.0 as example:

export OPENCV_VERSION=4.7.0
export JAVA_AWT_INCLUDE_PATH=${JAVA_HOME}
export JAVA_AWT_LIBRARY=${JAVA_HOME}
export JAVA_INCLUDE_PATH=${JAVA_HOME}/include
export JAVA_INCLUDE_PATH2=${JAVA_HOME}/include/darwin
export JAVA_JVM_LIBRARY=${JAVA_HOME}

wget https://github.com/opencv/opencv/archive/${OPENCV_VERSION}.zip
unzip ${OPENCV_VERSION}.zip && rm -rf ${OPENCV_VERSION}.zip && cd opencv-${OPENCV_VERSION}
mkdir -p build && cd build
cmake -DCMAKE_BUILD_TYPE=RELEASE -DBUILD_JAVA=ON ..
make -j8 && make install

step 3. Build MMDeploy Java API.

Please check the existence of *.class in ${MMDEPLOY_DIR}/csrc/mmdeploy/apis/java/mmdeploy.

If there is no existence of *.class, please follow this tutorial to build Java class.

step 4. Build demo.

Use ImageClassification as example.

First, you should set your mmdeploy path, opencv path, model path and image path to ${MMDEPLOY_DIR}, ${OPENCV_DIR}, ${MODEL_PATH} and ${IMAGE_PATH}. And then follow the bash codes.

cd demo/java
export TASK=ImageClassification
export ARGS=cpu\ ${MODEL_PATH}\ ${IMAGE_PATH}
ant -DtaskName=${TASK} -DjarDir=${OPENCV_DIR}/build/bin -DlibDir=${OPENCV_DIR}/build/lib:${MMDEPLOY_DIR}/build/lib -Dcommand=${ARGS}

As for PoseTracker, you should execute:

cd demo/java
export TASK=PoseTracker
export ARGS=cpu\ ${DET_MODEL_PATH}\  ${POSE_MODEL_PATH}\ ${VIDEO_PATH}
ant -DtaskName=${TASK} -DjarDir=${OPENCV_DIR}/build/bin -DlibDir=${OPENCV_DIR}/build/lib:${MMDEPLOY_DIR}/build/lib -Dcommand="${ARGS}"