MMDeploy安装笔记

阅读原文时间：2023年07月09日阅读：3

MMDeploy的TensorRT教程

conda create -n openmmlab python=3.7 -y
conda activate openmmlab

conda install pytorch==1.8.0 torchvision==0.9.0 cudatoolkit=10.2 -c pytorch -y

# install mmcv
pip install mmcv-full==1.4.0 -f https://download.openmmlab.com/mmcv/dist/cu102/torch1.8/index.html

# install mmdetection
git clone https://github.com/open-mmlab/mmdetection.git
cd mmdetection
pip install -r requirements/build.txt
pip install -v -e .

Download the checkpoint from this link and put it in the {MMDET_ROOT}/checkpoints where {MMDET_ROOT} is the root directory of your MMDetection codebase.

在anaconda中运行下列命令来安装MMDeploy

conda activate openmmlab

git clone https://github.com/open-mmlab/mmdeploy.git
cd mmdeploy
git submodule update --init --recursive
pip install -e . # 安装MMDeploy

install TensorRT through tar file
After installation, you’d better add TensorRT environment variables to bashrc by

cd /the/path/of/tensorrt/tar/gz/file
tar -zxvf TensorRT-8.2.3.0.Linux.x86_64-gnu.cuda-11.4.cudnn8.2.tar.gz

将下面的导入到 ~/.bashrc

export TENSORRT_DIR=$(pwd)/TensorRT-8.2.3.0
export LD_LIBRARY_PATH=$TENSORRT_DIR/lib:$LD_LIBRARY_PATH
install cudnn8.2 through tar file
Extract the compressed file and set the environment variables

cd /the/path/of/cudnn/tgz/file
tar -zxvf cudnn-11.3-linux-x64-v8.2.1.32.tgz

将下面的导入到 ~/.bashrc

export CUDNN_DIR=$(pwd)/cuda
export LD_LIBRARY_PATH=$CUDNN_DIR/lib64:$LD_LIBRARY_PATH

Step6-1: Build Custom Ops

TensorRT Custom Ops

cd ${MMDEPLOY_DIR}
mkdir -p build && cd build

cmake -DCMAKE_CXX_COMPILER=g++-7 -DMMDEPLOY_TARGET_BACKENDS=trt -DTENSORRT_DIR=${TENSORRT_DIR} -DCUDNN_DIR=${CUDNN_DIR} ..

make -j$(nproc)

Step6-2: install Model Converter

cd ${MMDEPLOY_DIR}
pip install -e .

Step6-3: 验证模型是否能够进行转换

python ${MMDEPLOY_DIR}/tools/check_env.py

# 如果正常输出会得到:
# 2022-05-04 10:13:07,140 - mmdeploy - INFO - tensorrt: 8.2.3.0    ops_is_avaliable : True

Step6-4: Convert Model

Once you have installed MMDeploy, you can convert the PyTorch model in the OpenMMLab model zoo to the backend model with one magic spell!

Assume you have installed MMDeploy in ${MMDEPLOY_DIR} and MMDetection in ${MMDET_DIR}

If you do not know where to find the path. Just type pip show mmdeploy and pip show mmdet in your console.

python ${MMDEPLOY_DIR}/tools/deploy.py ${MMDEPLOY_DIR}/configs/mmdet/detection/detection_tensorrt_dynamic-320x320-1344x1344.py ${MMDET_DIR}/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py ${MMDET_DIR}/checkpoints/faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth ${MMDET_DIR}/demo/demo.jpg --work-dir work_dirs \ # 转换好的模型保存目录
--device cuda:0 \ # 将cuda:0 更改成cuda???
--show \ # 展示使用后端推理框架，和原来pytorch推理的两张图片
--dump-info # 输出，可用与SDK

At the same time, an onnx model file end2end.onnx and ene2end.engine deploy.json detail.json pipeline.json (SDK config files) will generate on the work directory work_dirs.

Step6-5: Inference Model

Now you can do model inference with the APIs provided by the backend. But what if you want to test the model instantly? We have some backend wrappers for you.

from mmdeploy.apis import inference_model

deploy_cfg = "/home/zranguai/Deploy/MMDeploy/configs/mmdet/detection/detection_tensorrt_dynamic-320x320-1344x1344.py"
model_cfg = "/home/zranguai/Deploy/mmdetection/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py"
backend_files = ["/home/zranguai/Deploy/MMDeploy/work_dirs/end2end.engine"]
img = "/home/zranguai/Deploy/mmdetection/demo/demo.jpg"
device = 'cuda:0'

result = inference_model(model_cfg, deploy_cfg, backend_files, img=img, device=device)
print(result)

Step6-6: Evaluate Model

You might wonder that does the backend model have the same precision as the original one? How fast can the model run? MMDeploy provides tools to test the model.

python ${MMDEPLOY_DIR}/tools/test.py ${MMDEPLOY_DIR}/configs/mmdet/detection/detection_tensorrt_dynamic-320x320-1344x1344.py ${MMDET_DIR}/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py --model /home/zranguai/Deploy/MMDeploy/work_dirs/end2end.engine --metrics "bbox" --device cuda:0

Step7-1: build MMDeploy SDK for TensorRT

注意: 30系显卡需要将pplcv安装到最新版本。参考issue

cd ${MMDEPLOY_DIR}
mkdir -p build && cd build

cmake -DMMDEPLOY_BUILD_SDK=ON \
      -DCMAKE_CXX_COMPILER=g++-7 \
      -DMMDEPLOY_BUILD_SDK_PYTHON_API=ON \
      -DMMDEPLOY_TARGET_DEVICES="cuda;cpu" \
      -DMMDEPLOY_TARGET_BACKENDS=trt \
      -Dpplcv_DIR=${PPLCV_DIR}/cuda-build/install/lib/cmake/ppl \  # pplcv到最新版本 A high-performance image processing library of openPPL. ref:https://mmdeploy.readthedocs.io/en/latest/build/linux.html#install-dependencies-for-sdk
      -DTENSORRT_DIR=${TENSORRT_DIR} \
      -DCUDNN_DIR=${CUDNN_DIR} \
      -DOpenCV_DIR=/usr/lib/x86_64-linux-gnu/cmake/opencv4 \
      -Dspdlog_DIR=/usr/lib/x86_64-linux-gnu/cmake/spdlog \
      -DMMDEPLOY_CODEBASES=mmdet ..

make -j$(nproc) && make install

Step7-2: build demo

cd ${MMDEPLOY_DIR}/build/install/example
mkdir -p build && cd build

cmake -DOpenCV_DIR=/usr/lib/x86_64-linux-gnu/cmake/opencv4 \
      -DMMDeploy_DIR=${MMDEPLOY_DIR}/build/install/lib/cmake/MMDeploy ..
make object_detection

# suppress verbose logs
export SPDLOG_LEVEL=warn

# running the object detection example
./object_detection cuda ${work_dirs} ${path/to/an/image}
# 例子: ./object_detection cuda ${MMDEPLOY_DIR}/work_dirs ${MMDET_DIR}/demo/demo.jpg

在Clion中调试代码

setting中设置:
CMake options:
-DMMDeploy_DIR=/home/zranguai/Deploy/MMDeploy/build/install/lib/cmake/MMDeploy -DTENSORRT_DIR=/home/zranguai/Deploy/Backend/TensorRT/TensorRT-8.2.3.0 -DCUDNN_DIR=/home/zranguai/Deploy/Backend/TensorRT/cuda

Build directory:
/home/zranguai/Deploy/MMDeploy/build/install

Build options:
object_detection

configuation:
cuda /home/zranguai/Deploy/MMDeploy/work_dirs /home/zranguai/Deploy/MMDeploy/demo/demo.jpg

+++++++++++++++++++我是分割线++++++++++++++

MMDeploy的onnxruntime教程

参考官方教程

Here is an example of how to deploy and inference Faster R-CNN model of MMDetection from scratch.

step1: 创建虚拟环境并且安装MMDetection

Create Virtual Environment and Install MMDetection.

Please run the following command in Anaconda environment to install MMDetection.

conda create -n openmmlab python=3.7 -y
conda activate openmmlab

conda install pytorch==1.8.0 torchvision==0.9.0 cudatoolkit=10.2 -c pytorch -y

# install mmcv
pip install mmcv-full==1.4.0 -f https://download.openmmlab.com/mmcv/dist/cu102/torch1.8/index.html

# install mmdetection
git clone https://github.com/open-mmlab/mmdetection.git
cd mmdetection
pip install -r requirements/build.txt
pip install -v -e .

step2: 下载MMDetectin中训练好的权重

Download the Checkpoint of Faster R-CNN

Download the checkpoint from this link and put it in the {MMDET_ROOT}/checkpoints where {MMDET_ROOT} is the root directory of your MMDetection codebase.

step3: 安装MMDeploy和ONNX Runtime

Install MMDeploy and ONNX Runtime

step3-1: 安装MMDeploy

Please run the following command in Anaconda environment to install MMDeploy.

conda activate openmmlab

git clone https://github.com/open-mmlab/mmdeploy.git
cd mmdeploy
git submodule update --init --recursive
pip install -e .  # 安装MMDeploy

step3-2a: 下载onnxruntime

Once we have installed the MMDeploy, we should select an inference engine for model inference. Here we take ONNX Runtime as an example. Run the following command to install ONNX Runtime:

pip install onnxruntime==1.8.1

Then download the ONNX Runtime library to build the mmdeploy plugin for ONNX Runtime:

step3-2b: 制作onnxruntime的插件(模型转换会需要)

wget https://github.com/microsoft/onnxruntime/releases/download/v1.8.1/onnxruntime-linux-x64-1.8.1.tgz

tar -zxvf onnxruntime-linux-x64-1.8.1.tgz
cd onnxruntime-linux-x64-1.8.1
export ONNXRUNTIME_DIR=$(pwd)
export LD_LIBRARY_PATH=$ONNXRUNTIME_DIR/lib:$LD_LIBRARY_PATH  # 也可将这两句写进~/.bashrc

cd ${MMDEPLOY_DIR} # To MMDeploy root directory
mkdir -p build && cd build

# build ONNXRuntime custom ops
cmake -DMMDEPLOY_TARGET_BACKENDS=ort -DONNXRUNTIME_DIR=${ONNXRUNTIME_DIR} ..
make -j$(nproc)

step3-2c: build MMDeploy SDK(使用C的接口会用到)

# build MMDeploy SDK
cmake -DMMDEPLOY_BUILD_SDK=ON \
      -DCMAKE_CXX_COMPILER=g++-7 \
      -DOpenCV_DIR=/path/to/OpenCV/lib/cmake/OpenCV \  # 这里的opencv安装可参考这里https://mmdeploy.readthedocs.io/en/latest/build/linux.html#install-dependencies-for-sdk
      -Dspdlog_DIR=/path/to/spdlog/lib/cmake/spdlog \  # 这里的spdlog安装可参考这里https://mmdeploy.readthedocs.io/en/latest/build/linux.html#install-dependencies-for-sdk
      -DONNXRUNTIME_DIR=${ONNXRUNTIME_DIR} \
      -DMMDEPLOY_TARGET_BACKENDS=ort \
      -DMMDEPLOY_CODEBASES=mmdet ..
make -j$(nproc) && make install

# build MMDeploy SDK具体案例
cmake -DMMDEPLOY_BUILD_SDK=ON \
-DCMAKE_CXX_COMPILER=g++-7 \
-DOpenCV_DIR=/usr/lib/x86_64-linux-gnu/cmake/opencv4 \  # 通过apt-get安装的
-Dspdlog_DIR=/usr/lib/x86_64-linux-gnu/cmake/spdlog \
-DONNXRUNTIME_DIR=${ONNXRUNTIME_DIR} \
-DMMDEPLOY_TARGET_BACKENDS=ort \
-DMMDEPLOY_CODEBASES=mmdet ..

# 其中${MMDEPLOY_DIR} ${MMDET_DIR} ${ONNXRUNTIME_DIR}都可以写在 ~/.bashrc里面然后source ~/.bashrc生效

补充: 验证后端和插件是否安装成功

python ${MMDEPLOY_DIR}/tools/check_env.py

step4: Model Conversion

Once we have installed MMDetection, MMDeploy, ONNX Runtime and built plugin for ONNX Runtime, we can convert the Faster R-CNN to a .onnx model file which can be received by ONNX Runtime. Run following commands to use our deploy tools:

# Assume you have installed MMDeploy in ${MMDEPLOY_DIR} and MMDetection in ${MMDET_DIR}
# If you do not know where to find the path. Just type `pip show mmdeploy` and `pip show mmdet` in your console.

python ${MMDEPLOY_DIR}/tools/deploy.py \
    ${MMDEPLOY_DIR}/configs/mmdet/detection/detection_onnxruntime_dynamic.py \
    ${MMDET_DIR}/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py \
    ${MMDET_DIR}/checkpoints/faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth \
    ${MMDET_DIR}/demo/demo.jpg \
    --work-dir work_dirs \  # 转换好的模型保存目录
    --device cpu \
    --show \  # 展示使用后端推理框架，和原来pytorch推理的两张图片
    --dump-info  # 输出方便，可用与SDK

# 补充
# ${MMDEPLOY_DIR}和${MMDET_DIR}已经写进了~/.bashrc
# 转换好了模型可以通过python接口进行推理
例如: Inference Model
Now you can do model inference with the APIs provided by the backend. But what if you want to test the model instantly? We have some backend wrappers for you.

from mmdeploy.apis import inference_model

result = inference_model(model_cfg, deploy_cfg, backend_files, img=img, device=device)

If the script runs successfully, two images will display on the screen one by one. The first image is the infernce result of ONNX Runtime and the second image is the result of PyTorch. At the same time, an onnx model file end2end.onnx and three json files (SDK config files) will generate on the work directory work_dirs.

step5: Run MMDeploy SDK demo

After model conversion, SDK Model is saved in directory ${work_dir}.

Here is a recipe for building & running object detection demo.

cd build/install/example

# path to onnxruntime ** libraries **
export LD_LIBRARY_PATH=/path/to/onnxruntime/lib
# 例子: export LD_LIBRARY_PATH=/home/zranguai/Deploy/Backend/ONNXRuntime/onnxruntime-linux-x64-1.8.1/lib

mkdir -p build && cd build
cmake -DOpenCV_DIR=path/to/OpenCV/lib/cmake/OpenCV \
      -DMMDeploy_DIR=${MMDEPLOY_DIR}/build/install/lib/cmake/MMDeploy ..
make object_detection

# 例子:
# cmake -DOpenCV_DIR=/usr/lib/x86_64-linux-gnu/cmake/opencv4 \
#       -DMMDeploy_DIR=${MMDEPLOY_DIR}/build/install/lib/cmake/MMDeploy ..

# suppress verbose logs
export SPDLOG_LEVEL=warn

# running the object detection example
./object_detection cpu ${work_dirs} ${path/to/an/image}
# 例子: ./object_detection cpu ${MMDEPLOY_DIR}/work_dirs ${MMDET_DIR}/demo/demo.jpg

If the demo runs successfully, an image named "output_detection.png" is supposed to be found showing detection objects.

++++++++++++++++++++++++++++++++分割线++++++

MMDeploy的OpenVINO教程

conda create -n openmmlab python=3.7 -y
conda activate openmmlab

conda install pytorch==1.8.0 torchvision==0.9.0 cudatoolkit=10.2 -c pytorch -y

# install mmcv
pip install mmcv-full==1.4.0 -f https://download.openmmlab.com/mmcv/dist/cu102/torch1.8/index.html

# install mmdetection
git clone https://github.com/open-mmlab/mmdetection.git
cd mmdetection
pip install -r requirements/build.txt
pip install -v -e .

Download the checkpoint from this link and put it in the {MMDET_ROOT}/checkpoints where {MMDET_ROOT} is the root directory of your MMDetection codebase.

在anaconda中运行下列命令来安装MMDeploy

conda activate openmmlab

git clone https://github.com/open-mmlab/mmdeploy.git
cd mmdeploy
git submodule update --init --recursive
pip install -e . # 安装MMDeploy

pip install openvino-dev

step4-1: 根据官网提示: 这里的openvino不需要custom ops

step4-2: 可选项: 下载用于使用OpenVINO的SDK

Optional. If you want to use OpenVINO in MMDeploy SDK, please install and configure it by following the guild
参考安装教程

OpenVINO安装

tar -xvzf l_openvino_toolkit_p_2020.4.287.tgz
cd l_openvino_toolkit_p_2020.4.287
sudo ./install_GUI.sh 一路next安装
cd /opt/intel/openvino/install_dependencies
sudo ./install_openvino_dependencies.sh
vi ~/.bashrc
把如下几行放置到 bashrc 文件尾

set env for openvino

source /opt/intel/openvino_2021/bin/setupvars.sh # 注意找到是自己的路径
export INTEL_OPENVINO_DIR=/opt/intel/openvino_2021
export LD_LIBRARY_PATH=/opt/intel/openvino_2021/deployment_tools/inference_engine/lib/intel64
source ~/.bashrc 激活环境
模型优化配置步骤

cd /opt/intel/openvino_2021/deployment_tools/model_optimizer/install_prerequisites

sudo ./install_prerequisites.sh # 可以只安装onnx的

step4-3: build MMDeploy SDK(openvino)

cd ${MMDEPLOY_DIR} # To MMDeploy root directory
mkdir -p build && cd build

cmake -DMMDEPLOY_BUILD_SDK=ON \
      -DCMAKE_CXX_COMPILER=g++-7 \
      -DOpenCV_DIR=/path/to/OpenCV/lib/cmake/OpenCV \
      -Dspdlog_DIR=/path/to/spdlog/lib/cmake/spdlog \
      -DInferenceEngine_DIR=${INTEL_OPENVINO_DIR}/deployment_tools/inference_engine/share \
      -DMMDEPLOY_TARGET_BACKENDS=openvino \
      -DMMDEPLOY_CODEBASES=mmdet ..
make -j$(nproc) && make install

# build MMDeploy SDK具体案例
cmake -DMMDEPLOY_BUILD_SDK=ON \
      -DCMAKE_CXX_COMPILER=g++-7 \
      -DOpenCV_DIR=/usr/lib/x86_64-linux-gnu/cmake/opencv4 \  # 这里设置apt-get下载的opencv
      -Dspdlog_DIR=/usr/lib/x86_64-linux-gnu/cmake/spdlog \
      -DInferenceEngine_DIR=${INTEL_OPENVINO_DIR}/deployment_tools/inference_engine/share \
      -DMMDEPLOY_TARGET_BACKENDS=openvino \
      -DMMDEPLOY_CODEBASES=mmdet ..

# ${INTEL_OPENVINO_DIR}写进了~/.bashrc

补充: 验证后端和插件是否安装成功(注意openvino不需要安装插件)

python ${MMDEPLOY_DIR}/tools/check_env.py

# 当把这个export LD_LIBRARY_PATH=/opt/intel/openvino_2021/deployment_tools/inference_engine/lib/intel64写进~/.bashrc里面时候，会导致出现libopencv_ml.so.4.5: cannot open shared object file: No such file or directory  ？？

Step5: Model Conversion(这一步也可以放在step4-1前面)

# Assume you have installed MMDeploy in ${MMDEPLOY_DIR} and MMDetection in ${MMDET_DIR}
# If you do not know where to find the path. Just type `pip show mmdeploy` and `pip show mmdet` in your console.

python ${MMDEPLOY_DIR}/tools/deploy.py \
    ${MMDEPLOY_DIR}/configs/mmdet/detection/detection_openvino_dynamic-300x300.py \
    ${MMDET_DIR}/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py \
    ${MMDET_DIR}/checkpoints/faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth \
    ${MMDET_DIR}/demo/demo.jpg \
    --work-dir work_dirs \  # 转换好的模型保存目录
    --device cpu \
    --show \  # 展示使用后端推理框架，和原来pytorch推理的两张图片
    --dump-info  # 输出，可用与SDK

# 补充
# ${MMDEPLOY_DIR}和${MMDET_DIR}已经写进了~/.bashrc
# 转换好了模型可以通过python接口进行推理
例如: Inference Model
Now you can do model inference with the APIs provided by the backend. But what if you want to test the model instantly? We have some backend wrappers for you.

from mmdeploy.apis import inference_model

result = inference_model(model_cfg, deploy_cfg, backend_files, img=img, device=device)

例子:
deploy_cfg = "/home/zranguai/Deploy/MMDeploy/configs/mmdet/detection/detection_openvino_dynamic-300x300.py"
model_cfg = "/home/zranguai/Deploy/mmdetection/configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py"
backend_files = [ "/home/zranguai/Deploy/MMDeploy/work_dirs/end2end.xml",]
img = "/home/zranguai/Deploy/mmdetection/demo/demo.jpg"
device = 'cpu'

from mmdeploy.apis import visualize_model
visualize_model(model_cfg, deploy_cfg, backend_files[0], img, device, show_result=True)

python ${MMDEPLOY_DIR}/tools/check_env.pyIf the script runs successfully, two images will display on the screen one by one. The first image is the infernce result of OpenVINO and the second image is the result of PyTorch. At the same time, an onnx model file end2end.onnx and ene2end.bin(contains the weights and biases binary data) end2end.xml(describes the networks topology) end2end.mapping deploy.json detail.json pipeline.json (SDK config files) will generate on the work directory work_dirs.

Step6: Run MMDeploy SDK demo(for openvino)

After model conversion, SDK Model is saved in directory ${work_dir}.

Here is a recipe for building & running object detection demo.

cd build/install/example

# path to openvino ** libraries **
export LD_LIBRARY_PATH=/path/to/onnxruntime/lib/intel64
# 例子: export LD_LIBRARY_PATH=/opt/intel/openvino_2021/deployment_tools/inference_engine/lib/intel64

mkdir -p build && cd build
cmake -DOpenCV_DIR=path/to/OpenCV/lib/cmake/OpenCV \   # openvino中opencv路径
      -DMMDeploy_DIR=${MMDEPLOY_DIR}/build/install/lib/cmake/MMDeploy ..
make object_detection

# 例子2:
# cmake -DOpenCV_DIR=/opt/intel/openvino_2021/opencv/cmake \
#       -DMMDeploy_DIR=${MMDEPLOY_DIR}/build/install/lib/cmake/MMDeploy ..

# suppress verbose logs
export SPDLOG_LEVEL=warn

# running the object detection example
./object_detection cpu ${work_dirs} ${path/to/an/image}
# 例子: ./object_detection cpu ${MMDEPLOY_DIR}/work_dirs ${MMDET_DIR}/demo/demo.jpg
# 可能出现的错误: 上面导出的xml的name: torch-jit-export version="11",解决: 重新安装就好了

If the demo runs successfully, an image named "output_detection.png" is supposed to be found showing detection objects.

手机扫一扫

移动阅读更方便

你可能感兴趣的文章

MMDeploy安装笔记

MMDeploy的TensorRT教程

将下面的导入到 ~/.bashrc

将下面的导入到 ~/.bashrc

Step6-1: Build Custom Ops

Step6-2: install Model Converter

Step6-3: 验证模型是否能够进行转换

Step6-4: Convert Model

Assume you have installed MMDeploy in ${MMDEPLOY_DIR} and MMDetection in ${MMDET_DIR}

If you do not know where to find the path. Just type pip show mmdeploy and pip show mmdet in your console.

Step6-5: Inference Model

Step6-6: Evaluate Model

Step7-1: build MMDeploy SDK for TensorRT

Step7-2: build demo

在Clion中调试代码

MMDeploy的onnxruntime教程

step1: 创建虚拟环境并且安装MMDetection

Create Virtual Environment and Install MMDetection.

step2: 下载MMDetectin中训练好的权重

Download the Checkpoint of Faster R-CNN

step3: 安装MMDeploy和ONNX Runtime

Install MMDeploy and ONNX Runtime

step3-1: 安装MMDeploy

step3-2a: 下载onnxruntime

step3-2b: 制作onnxruntime的插件(模型转换会需要)

step3-2c: build MMDeploy SDK(使用C的接口会用到)

补充: 验证后端和插件是否安装成功

step4: Model Conversion

step5: Run MMDeploy SDK demo

MMDeploy的OpenVINO教程

step4-1: 根据官网提示: 这里的openvino不需要custom ops

step4-2: 可选项: 下载用于使用OpenVINO的SDK

set env for openvino

step4-3: build MMDeploy SDK(openvino)

补充: 验证后端和插件是否安装成功(注意openvino不需要安装插件)

Step5: Model Conversion(这一步也可以放在step4-1前面)

Step6: Run MMDeploy SDK demo(for openvino)

If you do not know where to find the path. Just type `pip show mmdeploy` and `pip show mmdet` in your console.