pyncnn

python wrapper of ncnn with pybind11, only support python3.x now.

Prerequisites

On Unix (Linux, OS X)

A compiler with C++11 support
CMake >= 2.8.12

On Windows

Visual Studio 2015
CMake >= 3.1

Build

clone ncnn and pybind11, build and install with default setting, if you change the install directory, change the cmake commond with your setting.

cd /path/to/ncnn
mkdir build && cd build
cmake ..
make -j4
make install

update ENVS_EXECUTABLE,ENVS_LIB and ENVS_INCLUDE to your path in CMakeLists.txt
remove find_package(Python3) in CMakeLists.text, this step is optional, it's up to your environment
change /path/to to your path and running the following cmd

cd /path/to/pyncnn
mkdir build
cd build
cmake -DCMAKE_PREFIX_PATH=/path/to/ncnn/build/install/lib/cmake/ncnn/ ..
make
如果编译失败
则进入src 目录手动编译
cp /path/to/build/install/lib/libncnn.a ./
g++ -O3 -shared -std=c++11 -fopenmp -o ncnn`python3-config --extension-suffix` -fPIC `python3 -m pybind11 --includes` -I/path/to/build/install/include/ncnn/ main.cpp ./libncnn.a
cd ../python
python3 setup.py install
如果Python导入ncnn出错，可以尝试 python3 setup.py bdist_wheel   
然后cd dist   pip3 install ncnn-0.0.1-py3-none-any.whl

Tests

test

cd /path/to/pyncnn/tests
python3 test.py

benchmark

cd /path/to/pyncnn/tests
python3 benchmark.py

benchmark gpu(build ncnn with vulkan)

cd /path/to/pyncnn/tests
python3 benchmark_gpu.py

numpy

ncnn.Mat->numpy.array, with no memory copy

mat = ncnn.Mat(...)
mat_np = np.array(mat)

numpy.array->ncnn.Mat, with no memory copy

mat_np = np.array(...)
mat = ncnn.Mat(mat_np)

model zoo

install requirements

pip install tqdm requests portalocker opencv-python

then you can import ncnn.model_zoo and get model list as follow:

import ncnn
import ncnn.model_zoo as model_zoo

print(model_zoo.get_model_list())

all model in model zoo has example in examples folder

rainfly123 / pyncnn