export TARGET_ARCH=aarch64
export TARGET_TRIPLE=aarch64-linux-gnu
export CROSS_COMPILE=/usr/bin/$TARGET_TRIPLE-

colcon build --packages-select dnn_benchmark_example \
   --merge-install \
   --cmake-force-configure \
   --cmake-args \
   --no-warn-unused-cli \
   -DCMAKE_TOOLCHAIN_FILE=`pwd`/robot_dev_config/aarch64_toolchainfile.cmake

  colcon build --packages-select dnn_benchmark_example \
     --merge-install \
     --cmake-force-configure \
     --cmake-args \
     --no-warn-unused-cli \
     -DPLATFORM_X86=ON \
     -DTHIRD_PARTY=`pwd`/../sysroot_docker
  

export COLCON_CURRENT_PREFIX=./install
source ./install/local_setup.bash

# For the example used in config, input configuration file, and local image for re-insertion
# Copy according to the actual installation path (the installation path in the docker is install/lib/dnn_benchmark_example/config/, the copy command is cp -r install/lib/dnn_benchmark_example/config/ .).
cp -r install/lib/dnn_benchmark_example/config/ .

# Run: Re-insert prediction using local jpg format images in synchronous mode, output performance metrics fps and latency through logging, and set log level to warn
ros2 run dnn_benchmark_example dnn_benchmark_example --ros-args --log-level warn

# Run default model evaluation, turn off fps information output, and change the number of processed cyclic images to 1000
ros2 run dnn_benchmark_example dnn_benchmark_example --ros-args --log-level warn -p show_fps_log:=0 -p statistic_cycle:=1000

# Change evaluation model
# The test model of dnn_benchmark_example is in the /opt/hobot/model/x3/basic/ path of the RDK X3 development board, installed via `apt install hobot_models_basic_1.0.0_arm64.deb` command, and when running switch the evaluation model using the -p option. If users need to evaluate their own model, place the model in the config folder, similarly reconfigure the model file path and model name through -p. For example, to evaluate the fcos model, change the model file by configuring -p model_file_name:=/opt/hobot/model/x3/basic/fcos_512x512_nv12.bin, and -p statistic_cycle:=50 to change the number of processed cyclic images to 50 (i.e., evaluate the fcos_512x512_nv12.bin model, output performance indicators for every 50 images processed)
ros2 run dnn_benchmark_example dnn_benchmark_example --ros-args --log-level warn -p model_file_name:=/opt/hobot/model/x3/basic/fcos_512x512_nv12.bin -p statistic_cycle:=50

export ROS_LOG_DIR=/userdata/
export LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:./install/lib/

# For the example used in config, copy according to the actual installation path
cp -r install/lib/dnn_benchmark_example/config/ .

# Run: Re-insert prediction using local jpg format images in synchronous mode, output performance metrics fps and latency through logging, and set log level to warn
./install/lib/dnn_benchmark_example/dnn_benchmark_example --ros-args --log-level warn

# Run default model evaluation, turn off fps information output, and change the number of processed cyclic images to 1000
./install/lib/dnn_benchmark_example/dnn_benchmark_example --ros-args --log-level warn -p show_fps_log:=0 -p statistic_cycle:=1000

# Change evaluation model
# The test model of dnn_benchmark_example is in the /opt/hobot/model/x3/basic path of the RDK X3 development board, installed via `apt install hobot_models_basic_1.0.0_arm64.deb` command, and when running switch the evaluation model using the -p option. If users need to evaluate their own model, place the model in the config folder, similarly reconfigure the model file path and model name through -p. For example, to evaluate the fcos model, change the model file by configuring -p model_file_name:=/opt/hobot/model/x3/basic/fcos_512x512_nv12.bin, and -p statistic_cycle:=50 to change the number of processed cyclic images to 50 (i.e., evaluate the fcos_512x512_nv12.bin model, output performance indicators for every 50 images processed)
./install/lib/dnn_benchmark_example/dnn_benchmark_example --ros-args --log-level warn -p model_file_name:=/opt/hobot/model/x3/basic/fcos_512x512_nv12.bin -p statistic_cycle:=50

export COLCON_CURRENT_PREFIX=./install
source ./install/local_setup.bash

# Copy the model, input configuration file, and local images used for the example based on the actual installation path (the installation path in docker is install/lib/dnn_benchmark_example/config/, copy command: cp -r install/lib/dnn_benchmark_example/config/ .).
cp -r install/lib/dnn_benchmark_example/config/ .

# Run: perform inference using local jpg format images in synchronous mode, output performance metrics fps and latency in logs, and set log level to warn
ros2 run dnn_benchmark_example dnn_benchmark_example --ros-args --log-level warn

# Run evaluation with specified model, disable fps information output, and change the number of processed images per cycle to 50
ros2 run dnn_benchmark_example dnn_benchmark_example --ros-args --log-level warn -p model_file_name:=config/mobilenetv1_224x224_nv12_pyramid.bin -p show_fps_log:=0 -p statistic_cycle:=50

echo performance > /sys/class/devfreq/devfreq0/governor
echo performance > /sys/devices/system/cpu/cpufreq/policy0/scaling_governor