Computer vision tutorial for installing and using PyTorch, OpenCV and NCNN in C++.
OpenCV (Open Source Computer Vision Library) is a library of programming functions mainly aimed at real-time computer vision. Originally developed by Intel, it was later supported by Willow Garage then Itseez (which was later acquired by Intel)
First step is the install OpenCV to your local environment which is based from link and works well in Ubuntu 20.04 (Do not guarantee for other ubuntu versions) (Do not need to do this installation in your project folder).
sudo apt-get update
sudo apt-get upgrade
sudo apt install build-essential cmake git pkg-config libgtk-3-dev \
libavcodec-dev libavformat-dev libswscale-dev libv4l-dev \
libxvidcore-dev libx264-dev libjpeg-dev libpng-dev libtiff-dev \
gfortran openexr libatlas-base-dev python3-dev python3-numpy \
libtbb2 libtbb-dev libdc1394-22-dev libopenexr-dev \
libgstreamer-plugins-base1.0-dev libgstreamer1.0-dev
mkdir ~/opencv_build && cd ~/opencv_build
git clone https://github.com/opencv/opencv.git
git clone https://github.com/opencv/opencv_contrib.git
cd ~/opencv_build/opencv
mkdir -p build && cd build
cmake -D CMAKE_BUILD_TYPE=RELEASE \
-D CMAKE_INSTALL_PREFIX=/usr/local \
-D INSTALL_C_EXAMPLES=ON \
-D INSTALL_PYTHON_EXAMPLES=ON \
-D OPENCV_GENERATE_PKGCONFIG=ON \
-D OPENCV_EXTRA_MODULES_PATH=~/opencv_build/opencv_contrib/modules \
-D BUILD_EXAMPLES=ON ..
make -j8
sudo make install
pkg-config --modversion opencv4
# Expected Output
4.4.0
Then add commands that given below will be include OpenCV library to CMakeLists.txt.
find_package(OpenCV REQUIRED)
include_directories(${OpenCV_INCLUDE_DIRS})
target_link_libraries(istation ${OpenCV_LIBS})
PyTorch is an open source machine learning (ML) framework based on the Python programming language and the Torch library. It is one of the preferred platforms for deep learning research. The framework is built to speed up the process between research prototyping and deployment. Can reach official installing C++ distributions of PyTorch from link. Our installation based on this documentation.
wget https://download.pytorch.org/libtorch/nightly/cpu/libtorch-shared-with-deps-latest.zip
unzip libtorch-shared-with-deps-latest.zip
mv libtorch include/
rm -r libtorch-shared-with-deps-latest.zip
Then properties of library will be added to CMakeLists.txt manually. istation will be your project name. Example of only using libtorch in C++, CMakeLists.txt as shown in below.
cmake_minimum_required(VERSION 3.0 FATAL_ERROR)
project(istation)
find_package(Torch REQUIRED)
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${TORCH_CXX_FLAGS}")
add_executable(istation main.cpp)
target_link_libraries(istation "${TORCH_LIBRARIES}")
set_property(TARGET istation PROPERTY CXX_STANDARD 14)
if (MSVC)
file(GLOB TORCH_DLLS "${TORCH_INSTALL_PREFIX}/lib/*.dll")
add_custom_command(TARGET istation
POST_BUILD
COMMAND ${CMAKE_COMMAND} -E copy_if_different
${TORCH_DLLS}
$<TARGET_FILE_DIR:istation>)
endif (MSVC)
Last step as shown in below to prepare PyTorch in terminal that on your codebase path.
mkdir build
cd build
cmake -DCMAKE_PREFIX_PATH=include/libtorch ..
cmake --build . --config Release
ncnn is a high-performance neural network inference computing framework optimized for mobile platforms. ncnn is deeply considerate about deployment and uses on mobile phones from the beginning of design.
ncnn installation based on its own installation guide. You should find best fit for your environment. In our repository, we are going to work with Ubuntu 20.04 so will download dependencies with respect to this direction. Imagine downloaded ncnn-20220216-ubuntu-2004.zip from releases. (Replace ncnn folder name if necessary)
unzip ncnn-20220216-ubuntu-2004.zip
mv ncnn-20220216-ubuntu-2004 include
rm -r ncnn-20220216-ubuntu-2004.zip
Then, add required to CMakeLists.txt.
set(ncnn_DIR ${CMAKE_SOURCE_DIR}/include/ncnn-20220216-ubuntu-2004/lib/cmake/ncnn)
find_package(ncnn REQUIRED)
target_link_libraries(istation ${TORCH_LIBRARIES} ${OpenCV_LIBS} ncnn)
Prepared CMakeLists.txt includes all the requirements!
ONNX Simplifier is presented to simplify the ONNX model. It infers the whole computation graph and then replaces the redundant operators with their constant outputs.
Clone repository and place it to your base folder.
https://github.com/daquexian/onnx-simplifier
We are going to give an example from torchvision.models.resnet50(pretrained=True) but you can apply every other network that you customized. (Check ncnn and Onnx supported layers)
# Check `pytorch2onnx2ncnn.py` setup for modifications
cd utils/
python pytorch2onnx2ncnn.py
Expected output R50.onnx (if not changed save path) will be written to utils/. It is time to simplify model.
(This is optional you do not need to do this if not necessary)
cd ../onnx-simplifier/
python -m onnxsim ../utils/R50.onnx ../utils/R50-Simplified.onnx
## Expected Output
Simplifying...
Ok!
Now .onnx model can be converted to the ncnn.
cd ../include/ncnn-20220216-ubuntu-2004/bin/
./onnx2ncnn ../../../utils/R50-Simplified.onnx ../../../utils/R50.param ../../../utils/R50.bin
Now we have .param and .bin files. We can write code with these informations but there are one more step to optimize networks.
(This is optional you do not need to do this if not necessary)
- Network size before optimize --> 102.1 MB
- Network size after optimize --> 51.1 MB
./ncnnoptimize ../../../utils/R50.param ../../../utils/R50.bin ../../../utils/R50-Optimized.param ../../../utils/R50-Optimized.bin 65536
We are going to move our models to build folder.
mv utils/R50-Optimized.bin build/Models/
mv utils/R50-Optimized.param build/Models/
Layers can be inspected from netron.app if necessary. Now everything ready to implement our codes. There is and example in src/main.cpp
We are going to give an example from torchvision.models.resnet50(pretrained=True) but you can apply every other network that you customized. (Check ncnn and Onnx supported layers)
# Check `pytorch2script.py` setup for modifications
cd utils/
python pytorch2script.py
mv R50.pt ../build/Models/
Expected output R50.pt (if not changed save path) will be written to utils/. It is time to simplify model. This .pt file is ready to use in your implementation. For details check src/main.cpp --> pytorch2script function.
CMake
CMake Tools
C/C++
C/C++ Extension Pack