在Raspberry Pi 4上安装NCNN/MNN深度学习框架(aarch64+armv7l)

小锋学长生活大爆炸

发布于 2021-11-30 20:00:46

2.4K1

发布于 2021-11-30 20:00:46

文章被收录于专栏：小锋学长生活大爆炸

注意：如果尚未安装 OpenCV，请先安装它。

ncnn

Raspberry 64-bit (aarch64)

# check for updates (64-bit OS is still under development!)
$ sudo apt-get update
$ sudo apt-get upgrade
# install dependencies
$ sudo apt-get install cmake wget
$ sudo apt-get install libprotobuf-dev protobuf-compiler
# download ncnn
$ git clone --depth=1 https://github.com/Tencent/ncnn.git
# install ncnn
$ cd ncnn
$ mkdir build
$ cd build
# build 64-bit ncnn
$ cmake -D NCNN_DISABLE_RTTI=OFF \
-D CMAKE_TOOLCHAIN_FILE=../toolchains/aarch64-linux-gnu.toolchain.cmake ..
$ make -j4
$ make install
# copy output to dirs
$ sudo mkdir /usr/local/lib/ncnn
$ sudo cp -r install/include/ncnn /usr/local/include/ncnn
$ sudo cp -r install/lib/libncnn.a /usr/local/lib/ncnn/libncnn.a
# once you've placed the output in your /usr/local directory,
# you may delete the ncnn directory if you have no tools or examples compiled

$ cd ~
$ sudo rm -rf ncnn
sudo /sbin/ldconfig

Raspberry 32-bit (armv7l)

# check for updates
$ sudo apt-get update
$ sudo apt-get upgrade
# install dependencies
$ sudo apt-get install cmake wget
$ sudo apt-get install libprotobuf-dev protobuf-compiler
# download ncnn
$ git clone --depth=1 https://github.com/Tencent/ncnn.git
# install ncnn
$ cd ncnn
$ mkdir build
$ cd build
# build 32-bit ncnn
$ cmake -D PI3=ON \
-D NCNN_DISABLE_RTTI=OFF \
-D CMAKE_EXE_LINKER_FLAGS=-ldl \
-D CMAKE_TOOLCHAIN_FILE=../toolchains/pi3.toolchain.cmake ..
$ make -j4
$ make install
# copy output to dirs
$ sudo mkdir /usr/local/lib/ncnn
$ sudo cp -r install/include/ncnn /usr/local/include/ncnn
$ sudo cp -r install/lib/libncnn.a /usr/local/lib/ncnn/libncnn.a
# once you've placed the output in your /usr/local directory,
# you may delete the ncnn directory if you have no tools or examples compiled
$ cd ~
$ sudo rm -rf ncnn
sudo /sbin/ldconfig

后续

‎如果一切顺利，您将获得两个文件夹。一个包含所有头文件，另一个包含库，如屏幕转储中所示。‎

另请注意包含示例的文件夹。这里涵盖了许多不同类型的深度学习。由于 ncnn 库中的版本更改，对实际深度学习模型的引用有时会导致错误。

MNN

默认cmake、Protobuf、opencv都已安装。

Raspberry 64-bit (aarch64)

如果尝试最新版本报错，则装这个版本：

原地址：https://github.com/alibaba/MNN/tree/6b0c16f24f222239ecce8fb61c51b7f267502670

快速下载地址：https://hub.fastgit.org/alibaba/MNN.git

编辑CMakeLists，打开OpenCL选项。

option(MNN_OPENCL "Enable OpenCL" ON)

sudo apt-get install ocl-icd-opencl-dev -y
./schema/generate.sh
mkdir build
cd build
cmake .. -DMNN_BUILD_CONVERTER=true -DMNN_SEP_BUILD=false 
make -j4
sudo make install 
sudo /sbin/ldconfig

本文参与腾讯云自媒体同步曝光计划，分享自作者个人站点/博客。

原始发表：2021/11/30 ，如有侵权请联系 cloudcommunity@tencent.com 删除

深度学习