【ビルド版】NVIDIA Jetson Nanoで NVIDIA版 Caffe NVCaffeを GPUパワーで動かしてキモイ絵をモリモリ量産

【2020年版】NVIDIA Jetson Nano、Jetson Xavier NXの便利スクリプト

NVIDIA Jetson Nanoで OpenCV 3をビルドしてインストールする方法、NVCaffe等の OpenCV 4未対応を動かす

NVCC src/caffe/util/math_functions.cu
In file included from src/caffe/util/bbox_util.cu:10:0:
./include/caffe/util/bbox_util.hpp:1:10: fatal error: opencv2/core/core.hpp: No such file or directory
 #include <opencv2/core/core.hpp>
          ^~~~~~~~~~~~~~~~~~~~~~~

jetson@jetson-desktop:~/nvcaffe$ apt-cache search libopencv-core3.2
libopencv-core-dev - development files for libopencv-core3.2
libopencv-core3.2 - computer vision core library

sudo apt-get install -y libopencv-core-dev libopencv-core3.2

In file included from src/caffe/util/bbox_util.cu:10:0:
./include/caffe/util/bbox_util.hpp:2:10: fatal error: opencv2/highgui/highgui.hpp: No such file or directory
 #include <opencv2/highgui/highgui.hpp>
          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
Makefile:623: recipe for target '.build_release/cuda/src/caffe/util/bbox_util.o' failed

jetson@jetson-desktop:~$ sudo ln -s /usr/include/opencv4/opencv2/ /usr/include/opencv2

jetson@jetson-desktop:~$ ls -l /usr/include/opencv2/core/
-rw-r--r-- 1 root root  21525 12月 14 00:46 affine.hpp
-rw-r--r-- 1 root root   2819 12月 14 00:46 async.hpp
-rw-r--r-- 1 root root  26499 12月 14 00:46 base.hpp
-rw-r--r-- 1 root root   1197 12月 14 00:46 bindings_utils.hpp
-rw-r--r-- 1 root root    845 12月 14 00:46 bufferpool.hpp
-rw-r--r-- 1 root root   6975 12月 14 00:46 check.hpp
-rw-r--r-- 1 root root 128652 12月 14 00:46 core_c.h
-rw-r--r-- 1 root root   2366 12月 14 00:46 core.hpp

src/caffe/util/bbox_util.cpp: In function ‘void caffe::VisualizeBBox(const std::vector<cv::Mat>&, const caffe::Blob*, float, const std::vector<cv::Scalar_<double> >&, const std::map<int, std::__cxx11::basic_string<char> >&, const string&)’:
src/caffe/util/bbox_util.cpp:2239:42: error: ‘CV_FILLED’ was not declared in this scope
                   CV_RGB(255, 255, 255), CV_FILLED);
                                          ^~~~~~~~~
src/caffe/util/bbox_util.cpp:2239:42: note: suggested alternative: ‘CLD_KILLED’
                   CV_RGB(255, 255, 255), CV_FILLED);
                                          ^~~~~~~~~
                                          CLD_KILLED
src/caffe/util/bbox_util.cpp:2274:48: error: there are no arguments to ‘CV_FOURCC’ that depend on a template parameter, so a declaration of ‘CV_FOURCC’ must be available [-fpermissive]
         cv::VideoWriter outputVideo(save_file, CV_FOURCC('D', 'I', 'V', 'X'),
                                                ^~~~~~~~~
src/caffe/util/bbox_util.cpp:2274:48: note: (if you use ‘-fpermissive’, G++ will accept your code, but allowing the use of an undeclared name is deprecated)
src/caffe/util/bbox_util.cpp: In instantiation of ‘void caffe::VisualizeBBox(const std::vector<cv::Mat>&, const caffe::Blob*, float, const std::vector<cv::Scalar_<double> >&, const std::map<int, std::__cxx11::basic_string<char> >&, const string&) [with Dtype = float; std::__cxx11::string = std::__cxx11::basic_string<char>]’:
src/caffe/util/bbox_util.cpp:2293:43:   required from here

...

src/caffe/util/bbox_util.cpp:2274:57: error: ‘CV_FOURCC’ was not declared in this scope
src/caffe/util/bbox_util.cpp:2274:57: note: suggested alternative: ‘CV_16UC’
         cv::VideoWriter outputVideo(save_file, CV_FOURCC('D', 'I', 'V', 'X'),
                                                ~~~~~~~~~^~~~~~~~~~~~~~~~~~~~
                                                CV_16UC
src/caffe/util/bbox_util.cpp: In instantiation of ‘void caffe::VisualizeBBox(const std::vector<cv::Mat>&, const caffe::Blob*, float, const std::vector<cv::Scalar_<double> >&, const std::map<int, std::__cxx11::basic_string<char> >&, const string&) [with Dtype = half_float::half; std::__cxx11::string = std::__cxx11::basic_string<char>]’:
src/caffe/util/bbox_util.cpp:2305:43:   required from here

...

src/caffe/util/im_transforms.cpp:472:42: error: ‘CV_YCrCb2BGR’ was not declared in this scope
       cv::cvtColor(ycrcb_image, out_img, CV_YCrCb2BGR);
                                          ^~~~~~~~~~~~
src/caffe/util/im_transforms.cpp:485:22: error: ‘CV_IMWRITE_JPEG_QUALITY’ was not declared in this scope
     params.push_back(CV_IMWRITE_JPEG_QUALITY);
                      ^~~~~~~~~~~~~~~~~~~~~~~
src/caffe/util/im_transforms.cpp:488:33: error: ‘CV_LOAD_IMAGE_COLOR’ was not declared in this scope
     out_img = cv::imdecode(buf, CV_LOAD_IMAGE_COLOR);
                                 ^~~~~~~~~~~~~~~~~~~
src/caffe/util/im_transforms.cpp:488:33: note: suggested alternative: ‘CV_HAL_DFT_STAGE_COLS’
     out_img = cv::imdecode(buf, CV_LOAD_IMAGE_COLOR);
                                 ^~~~~~~~~~~~~~~~~~~
                                 CV_HAL_DFT_STAGE_COLS

...

src/caffe/util/im_transforms.cpp:659:34: note: suggested alternative: ‘CV_SSE2’
     cvtColor(*out_img, *out_img, CV_HSV2BGR);
                                  ^~~~~~~~~~
                                  CV_SSE2
Makefile:610: recipe for target '.build_release/src/caffe/util/im_transforms.o' failed
make: *** [.build_release/src/caffe/util/im_transforms.o] Error 1

nano ../src/caffe/layers/video_data_layer.cpp
#include <opencv2/videoio.hpp>
using namespace cv;

#define CV_CAP_PROP_FRAME_COUNT CAP_PROP_FRAME_COUNT
#define CV_CAP_PROP_POS_FRAMES  CAP_PROP_POS_FRAMES
#define CV_LOAD_IMAGE_COLOR     LOAD_IMAGE_COLOR

nano ../src/caffe/layers/window_data_layer.cpp
         cv_img = cv::imread(image.first, cv::IMREAD_COLOR);

/src/caffe/util/bbox_util.cpp:2186:42: error: ‘CV_FILLED’ was not declared in this scope
                   CV_RGB(255, 255, 255), CV_FILLED);
nano ../src/caffe/util/bbox_util.cpp
　CV_RGBを cv::Scalarに書き換え。更に、RGBの順番を BGRに変更。
　CV_FILLED cv::FILLED

　等々、更に他にも必要なので省略。。。

【ビルド版】Raspberry Piで DeepDreamを動かしてキモイ絵をモリモリ量産 Caffe Deep Learning Framework

【インストール版】Raspberry Piで DeepDreamを動かしてキモイ絵をモリモリ量産 Caffe Deep Learning

【ビルド版】NVIDIA Jetson Nanoで Caffe DeepDreamを GPUパワーで動かしてキモイ絵をモリモリ量産

cd
# nvcaffeディレクトリに git cloneする
# NVCaffe
# NVCaffe May 1, 2019 v0.17.3 fd6cf7a
git clone https://github.com/NVIDIA/caffe nvcaffe --depth 1 -b v0.17.3
cd nvcaffe

# Makefile.configファイルを作成（参考例のファイルからコピー）する
cp Makefile.config.example Makefile.config

# Makefile.configファイルを Jetson Nano環境に合わせて編集する
# nano Makefile.config

# コマンドラインでサクッと Makefile.configファイルを変更する
sed -i 's/^# USE_CUDNN/USE_CUDNN/' Makefile.config
# sed -i 's/BLAS :=.*/BLAS := atlas/' Makefile.config
sed -i 's/x86_64-linux-gnu/aarch64-linux-gnu/' Makefile.config
sed -i 's/# OPENCV_VERSION/OPENCV_VERSION/' Makefile.config
sed -i 's/arch=compute_60,code=sm_60/arch=compute_53,code=sm_53/' Makefile.config
sed -i 's/# WITH_PYTHON_LAYER/WITH_PYTHON_LAYER/' Makefile.config

grep "BLAS " Makefile.config
grep "CUDA_" Makefile.config
grep "_DIRS :" Makefile.config
grep "OPENCV_" Makefile.config
grep "_CUDNN" Makefile.config
grep "CPU_" Makefile.config
grep "WITH_" Makefile.config

user@user-desktop:~/N/caffe$ grep "BLAS " Makefile.config
# BLAS choice:
BLAS := open

user@user-desktop:~/N/caffe$ grep "CUDA_" Makefile.config
CUDA_DIR := /usr/local/cuda
# CUDA_DIR := /usr
CUDA_ARCH :=    -gencode arch=compute_50,code=sm_50 \

user@user-desktop:~/N/caffe$ grep "_DIRS :" Makefile.config
INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/aarch64-linux-gnu/hdf5/serial

user@user-desktop:~/N/caffe$ grep "OPENCV_" Makefile.config
OPENCV_VERSION := 3

user@user-desktop:~/N/caffe$ grep "_CUDNN" Makefile.config
USE_CUDNN := 1

user@user-desktop:~/N/caffe$ grep "CPU_" Makefile.config

user@user-desktop:~/N/caffe$ grep "WITH_" Makefile.config
WITH_PYTHON_LAYER := 1

user@user-desktop:~/N/caffe$ diff Makefile.config.example Makefile.config
6c6
< # USE_CUDNN := 1
---
> USE_CUDNN := 1
21c21
< # OPENCV_VERSION := 3
---
> OPENCV_VERSION := 3
36c36
<               -gencode arch=compute_60,code=sm_60 \
---
>               -gencode arch=compute_53,code=sm_53 \
85c85
< # WITH_PYTHON_LAYER := 1
---
> WITH_PYTHON_LAYER := 1
89c89
< LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu/hdf5/serial
---
> LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/aarch64-linux-gnu/hdf5/serial

# Caffe は、システムのパッケージ マネージャーから利用できるいくつかのライブラリに依存します。
# Ubuntu 14.04 の場合、次のコマンドで必要なライブラリがインストールされます。

sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler libgflags-dev libgoogle-glog-dev liblmdb-dev libatlas-base-dev git

sudo apt-get install --no-install-recommends libboost-all-dev

# Makefile.config, BLAS := atlas
sudo apt-get -y install libatlas-base-dev protobuf-compiler libprotoc-dev libboost-dev libgflags-dev libgoogle-glog-dev libhdf5-dev libleveldb-dev liblmdb-dev libopencv-dev libsnappy-dev libboost-system-dev libboost-filesystem-dev libboost-thread-dev libboost-python-dev python-skimage python-protobuf python-numpy python-pil

# Makefile.config, BLAS := open
sudo apt-get -y install libopenblas-dev

# libturbojpegをインストールする
# sudo apt-get -y install libturbojpeg libturbojpeg-dev
# for Raspbian OS Raspberry Pi 3B+
# Note, selecting 'libturbojpeg0-dev' instead of 'libturbojpeg-dev'
sudo apt-get -y install libturbojpeg libturbojpeg0-dev
ls -l /usr/lib/aarch64-linux-gnu/libturbojpeg.so
# sudo ln -s /usr/lib/aarch64-linux-gnu/libturbojpeg.so.0.1.0 /usr/lib/aarch64-linux-gnu/libturbojpeg.so
sudo ldconfig

# libboost-regex-devをインストールする
sudo apt-get -y install libboost-regex-dev
# libopenblas-base libopenblas-devをインストールする
sudo apt-get -y install libopenblas-base libopenblas-dev

cd
cd nvcaffe

# ビルド設定を初期化
make clean

# make all -j4で 4 coreを使ってビルド速度を高速化（約 30分）
time make all -j4

# real    29m21.720s
# user    73m4.716s
# sys     7m26.316s

# make testと make runtestで動作確認
time make test -j4

# real    37m4.786s
# user    73m44.072s
# sys     4m6.148s

time make runtest -j4

# real    31m23.448s
# user    22m35.384s
# sys     6m40.428s

# Python用の Caffeの作成と distributeでビルド内容を反映
make pycaffe
make distribute

# ここまで成功すれば Caffeのビルドは完了

cd
cd nvcaffe

# ビルド設定を初期化
make clean

# make all -j4で 4 coreを使ってビルド速度を高速化（約 82分）
time make all -j4

# real    82m5.378s
# user    264m0.340s
# sys     21m39.764s

# Python用の Caffeの作成と distributeでビルド内容を反映
# ImportError : No module named _caffe
make pycaffe
make distribute

# ここまで成功すれば Caffeのビルドは完了

# 環境変数 CAFFE_HOMEに PATHを設定する
# export CAFFE_HOME=/home/user/nvcaffe
cd
cd nvcaffe
export CAFFE_HOME=$(pwd)

echo ${CAFFE_HOME}
# /home/jetson/nvcaffe

# パスにビルドで作成した Caffeを追加する
export PYTHONPATH=${CAFFE_HOME}/python:$PYTHONPATH
export LD_LIBRARY_PATH=${CAFFE_HOME}/distribute/lib:$LD_LIBRARY_PATH

# 上記手順で NVCaffeをビルドする
# 上記手順で CAFFE_HOME等の環境変数を設定する

echo ${CAFFE_HOME}
# /home/jetson/nvcaffe

# OpenPoseのビルドに必要なファイルをコピーする
# fatal error half_float/half.hpp No such file or directory
cp -r ${CAFFE_HOME}/3rdparty/* ${CAFFE_HOME}/include

# fatal error caffe/proto/caffe.pb.h No such file or directory
cd ${CAFFE_HOME}
protoc src/caffe/proto/caffe.proto --cpp_out=.
mkdir include/caffe/proto
mv src/caffe/proto/caffe.pb.h include/caffe/proto

# NVCaffeの動作確認
# cd /opt/caffe
cd
cd nvcaffe
data/mnist/get_mnist.sh
examples/mnist/create_mnist.sh
examples/mnist/train_lenet.sh

# If everything is running correctly, NVCaffe should download and create a data set, and
# then start training LeNet. If the training is successful, you will see a code similar to the
# following towards the end of the output:
# 下記の様な表示が出れば成功

Creating lmdb...
I0429 12:53:05.703790 11191 db_lmdb.cpp:36] Opened lmdb examples/mnist/mnist_train_lmdb
I0429 12:53:05.704241 11191 convert_mnist_data.cpp:88] A total of 60000 items.
I0429 12:53:05.704269 11191 convert_mnist_data.cpp:89] Rows: 28 Cols: 28
I0429 12:53:22.402772 11191 convert_mnist_data.cpp:108] Processed 60000 files.
I0429 12:53:22.909931 11201 db_lmdb.cpp:36] Opened lmdb examples/mnist/mnist_test_lmdb

...

I0402 15:08:01.016016 33 solver.cpp:431] Iteration 10000, loss = 0.0342847
I0402 15:08:01.016043 33 solver.cpp:453] Iteration 10000, Testing net (#0)
I0402 15:08:01.085050 38 data_reader.cpp:128] Restarting data pre-fetching
I0402 15:08:01.087720 33 solver.cpp:543] Test net output #0: accuracy = 0.9587
I0402 15:08:01.087751 33 solver.cpp:543] Test net output #1: loss = 0.130223 (* 1 = 0.130223 loss)
I0402 15:08:01.087767 33 caffe.cpp:239] Solver performance on device 0: 498.3 * 64 = 3.189e+04 img/sec
I0402 15:08:01.087780 33 caffe.cpp:242] Optimization Done in 24s

# Python版の Google Deep Dream kesara/deepdreamerをダウンロードする
cd
git clone https://github.com/kesara/deepdreamer
cd deepdreamer

# 学習データ bvlc_googlenet.caffemodelをダウンロードする
wget http://dl.caffe.berkeleyvision.org/bvlc_googlenet.caffemodel

# deploy.prototxtをダウンロードする
# https://github.com/BVLC/caffe/tree/master/models/bvlc_googlenet
wget https://raw.githubusercontent.com/BVLC/caffe/master/models/bvlc_googlenet/deploy.prototxt

# deploy.prototxtの一番最後に force_backward: trueの行を追加する
# nano deploy.prototxt
# force_backward: true

# コマンドラインでサクッと force_backward: trueの行を追加する
echo "force_backward: true" >> deploy.prototxt

# tailコマンドで最後に追加された事を確認する
tail deploy.prototxt

nano ./deepdreamer/deepdreamer.py

● 12行目をコメントにする。
# from deepdreamer.images2gif import writeGif

● 196～198行目をコメントにする。
#         writeGif(
#             "{}.gif".format(img_path), frames, duration=duration,
#             repeat=loop)

# コマンドラインでサクッと deepdreamer/deepdreamer.pyを編集する
# 196～198行目を削除する
sed '196,198d' -i ./deepdreamer/deepdreamer.py
# 12行目を削除する
sed '12d' -i ./deepdreamer/deepdreamer.py

cd
cd deepdreamer
wget https://github.com/google/deepdream/raw/master/flowers.jpg
cp flowers.jpg nvf.jpg
# NVIDIA版 Caffe NVCaffeは --gpuid 0は不要
python deepdreamer.py nvf.jpg

user@user-desktop:~/deepdreamer$ ls -l nvf*
-rw-rw-r-- 1 user user 30570  5月  7 17:14 nvf.jpg
-rw-rw-r-- 1 user user 22700  5月  7 20:44 nvf.jpg_0.jpg
-rw-rw-r-- 1 user user 22305  5月  7 20:45 nvf.jpg_1.jpg
-rw-rw-r-- 1 user user 21920  5月  7 20:45 nvf.jpg_2.jpg
-rw-rw-r-- 1 user user 21845  5月  7 20:45 nvf.jpg_3.jpg
-rw-rw-r-- 1 user user 21878  5月  7 20:45 nvf.jpg_4.jpg
-rw-rw-r-- 1 user user 21774  5月  7 20:45 nvf.jpg_5.jpg
-rw-rw-r-- 1 user user 21788  5月  7 20:45 nvf.jpg_6.jpg
-rw-rw-r-- 1 user user 21807  5月  7 20:45 nvf.jpg_7.jpg
-rw-rw-r-- 1 user user 21742  5月  7 20:45 nvf.jpg_8.jpg
-rw-rw-r-- 1 user user 21780  5月  7 20:45 nvf.jpg_9.jpg
-rw-rw-r-- 1 user user 21654  5月  7 20:46 nvf.jpg_10.jpg
-rw-rw-r-- 1 user user 21655  5月  7 20:46 nvf.jpg_11.jpg
-rw-rw-r-- 1 user user 21431  5月  7 20:46 nvf.jpg_12.jpg

CXX src/caffe/util/io.cpp
src/caffe/util/io.cpp:17:10: fatal error: turbojpeg.h: No such file or directory
 #include <turbojpeg.h>
          ^~~~~~~~~~~~~
compilation terminated.
Makefile:610: recipe for target '.build_release/src/caffe/util/io.o' failed
make: *** [.build_release/src/caffe/util/io.o] Error 1
make: *** Waiting for unfinished jobs....

# libturbojpegをインストールする
sudo apt-get -y install libturbojpeg libturbojpeg-dev

# Useful notes
# Libturbojpeg library is used since 0.16.5. It has a packaging bug.
# Please execute the following (required for Makefile, optional for CMake):
ls -l /usr/lib/aarch64-linux-gnu/libturbojpeg.*
ls -l /usr/lib/aarch64-linux-gnu/libturbojpeg.so
# No such file or directory

# Solution （解決方法）
sudo ln -s /usr/lib/aarch64-linux-gnu/libturbojpeg.so.0.1.0 /usr/lib/aarch64-linux-gnu/libturbojpeg.so
sudo ldconfig

CXX examples/mnist/convert_mnist_data.cpp
AR -o .build_release/lib/libcaffe-nv.a
LD -o .build_release/lib/libcaffe-nv.so.0.17.3
/usr/bin/ld: cannot find -lboost_regex
/usr/bin/ld: cannot find -lopenblas
collect2: error: ld returned 1 exit status
Makefile:600: recipe for target '.build_release/lib/libcaffe-nv.so.0.17.3' failed
make: *** [.build_release/lib/libcaffe-nv.so.0.17.3] Error 1

# Solution
# libboost-regex-devをインストールする
sudo apt-get -y install libboost-regex-dev
# libopenblas-base libopenblas-devをインストールする
sudo apt-get -y install libopenblas-base libopenblas-dev

user@user-desktop:~/deepdreamer$ python deepdreamer.py nvf.jpg
Traceback (most recent call last):
  File "deepdreamer.py", line 9, in <module>
    from deepdreamer.deepdreamer import deepdream, deepdream_video, list_layers
  File "/home/user/deepdreamer/deepdreamer/deepdreamer.py", line 11, in <module>
    from caffe import Classifier, set_device, set_mode_gpu
  File "/home/user/N/caffe/python/caffe/__init__.py", line 1, in <module>
    from .pycaffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, RMSPropSolver, \
  File "/home/user/N/caffe/python/caffe/pycaffe.py", line 13, in <module>
    from ._caffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, \
ImportError: No module named _caffe

# Solution
# Python用の Caffeの作成と distributeでビルド内容を反映
make pycaffe
make distribute

# パスにビルドで作成した Caffeを追加する
export PYTHONPATH=${CAFFE_HOME}/python:$PYTHONPATH
export LD_LIBRARY_PATH=${CAFFE_HOME}/distribute/lib:$LD_LIBRARY_PATH

user@user-desktop:~/deepdreamer$ python deepdreamer.py nvf.jpg
Traceback (most recent call last):
  File "deepdreamer.py", line 9, in <module>
    from deepdreamer.deepdreamer import deepdream, deepdream_video, list_layers
  File "/home/user/deepdreamer/deepdreamer/deepdreamer.py", line 11, in <module>
    from caffe import Classifier, set_device, set_mode_gpu
  File "/home/user/N/caffe/python/caffe/__init__.py", line 1, in <module>
    from .pycaffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, RMSPropSolver, \
  File "/home/user/N/caffe/python/caffe/pycaffe.py", line 13, in <module>
    from ._caffe import Net, SGDSolver, NesterovSolver, AdaGradSolver, \
ImportError: /home/user/N/caffe/python/caffe/_caffe.so: undefined symbol: _ZN5caffe21PyErrReportAndForwardEv

# Solution
# WITH_PYTHON_LAYER := 1のコメント文字＃を取って有効にする
# Uncomment to support layers written in Python (will link against Python libs)
WITH_PYTHON_LAYER := 1

user@user-desktop:~/N/caffe$ ./.build_release/tools/caffe -version
WARNING: Logging before InitGoogleLogging() is written to STDERR
I0505 19:57:53.678395 17696 parallel.cpp:48] P2PManager::Init @ user-desktop
caffe version 0.17.3

WARNINGなので、とりあえず無視する。

NVIDIA Jetson Nanoで OpenPoseをビルドする方法、動画から人体の骨格検出

【2020年版】NVIDIA Jetson Nano対応の OpenPoseをビルドする手順

【2020年版】NVIDIA Jetson用に最新の CMake 3.17.3をビルドしてインストールする方法

# -- CUDA detected: 10.2
# -- Found cuDNN: ver. ??? found (include: /usr/include, library: /usr/lib/aarch64-linux-gnu/libcudnn.so)
# CMake Error at cmake/Cuda.cmake:256 (message):
#   cuDNN version >3 is required.
# Call Stack (most recent call first):
#   cmake/Cuda.cmake:284 (detect_cuDNN)
#   cmake/Dependencies.cmake:85 (include)
#   CMakeLists.txt:49 (include)

# Caffe Configuration
# 読み込むファイルを cudnn.hから cudnn_version.hに変更する
# file(READ ${CUDNN_INCLUDE}/cudnn.h CUDNN_VERSION_FILE_CONTENTS)
sed -i -e "s/cudnn.h/cudnn_version.h/g" ../cmake/Cuda.cmake

cat /usr/include/cudnn_version.h | grep CUDNN
# #ifndef CUDNN_VERSION_H_
# #define CUDNN_VERSION_H_
# #define CUDNN_MAJOR 8
# #define CUDNN_MINOR 0
# #define CUDNN_PATCHLEVEL 0
# #define CUDNN_VERSION (CUDNN_MAJOR * 1000 + CUDNN_MINOR * 100 + CUDNN_PATCHLEVEL)
# #endif /* CUDNN_VERSION_H */

NVIDIA Jetson Nano 開発者キットを買ってみた。メモリ容量 4GB LPDDR4 RAM

Raspberry Piでメモリを馬鹿食いするアプリ用に不要なサービスを停止してフリーメモリを増やす方法

【成功版】最新版の Darknetに digitalbrain79版の Darknet with NNPACKの NNPACK処理を適用する

【成功版】Raspberry Piで NNPACK対応版の Darknet Neural Network Frameworkをビルドする方法

【成功版】Raspberry Piで Darknet Neural Network Frameworkをビルドする方法

【成功版】Raspberry Piに TensorFlow Deep Learning Frameworkをインストールする方法

Raspberry Piで TensorFlow Deep Learning Frameworkを自己ビルドする方法

Raspberry Piで Caffe Deep Learning Frameworkで物体認識を行なってみるテスト

【ビルド版】Raspberry Piで DeepDreamを動かしてキモイ絵をモリモリ量産 Caffe Deep Learning Framework

【インストール版】Raspberry Piで DeepDreamを動かしてキモイ絵をモリモリ量産 Caffe Deep Learning

Raspberry Piで Caffe2 Deep Learning Frameworkをソースコードからビルドする方法

Orange Pi PC 2の 64bitのチカラで DeepDreamしてキモイ絵を高速でモリモリ量産してみるテスト

Raspberry Piに Jupyter Notebookをインストールして拡張子 ipynb形式の IPythonを動かす

Raspberry Piで Deep Learningフレームワーク Chainerをインストールしてみる

Raspberry Piで DeepBeliefSDKをビルドして画像認識フレームワークを動かす方法

Raspberry Piで Microsoftの ELLをビルドする方法

Raspberry Piで MXNet port of SSD Single Shot MultiBoxを動かして画像の物体検出をする方法

Raspberry Piで Apache MXNet Incubatingをビルドする方法

Raspberry Piで OpenCVの Haar Cascade Object Detectionでリアルタイムにカメラ映像の顔検出を行なってみる

Raspberry Piで NNPACKをビルドする方法

Raspberry Pi 3の Linuxコンソール上で使用する各種コマンドまとめ