Raspberry PI Tensorflow 설치

라즈베리파이 텐서플로우 설치

라즈베리파이 1 B

메모리 8GB

OS RASPBIAN STRETCH

Python3.5.3

라즈베리파이 2 B

메모리 64GB

OS RASPBIAN STRETCH

Python3.5.3

파이썬 설치

sudo apt-get update

sudo apt-get install python3.5

sudo apt-get install python3-pip

* 라즈베리파이 모델 확인

https://blog.ayukawa.kr/archives/1735

cat /proc/cpuinfo로 cpu정보를 확인

pi2b

pi1b

모델에 맞는 텐서플로우 다운로드

아래의 링크에서 적절한 파일을 다운로드 합니다.

https://github.com/lhelontra/tensorflow-on-arm/releases

라즈베리파이 1 B

tensorflow-1.8.0-cp35-none-linux_armv6l.whl

sudo pip3 install tensorflow-1.8.0-cp35-none-linux_arm6l.whl

라즈베리파이 2 B

tensorflow-1.8.0-cp35-none-linux_armv7l.whl

sudo pip3 install tensorflow-1.8.0-cp35-none-linux_arm7l.whl

numpy 빌드 오류시

https://www.raspberrypi.org/forums/viewtopic.php?t=203775

libatlas3-base 설치

sudo apt-get install libatlas3-base

예제 TEST

$ python3

Python 3.5.3 (default, Jan 19 2017, 14:11:04)

[GCC 6.3.0 20170124] on linux

Type "help", "copyright", "credits" or "license" for more information.

>>> import tensorflow as tf

>>>

>>> hello = tf.constant('Hello, TensorFlow!')

>>> sess = tf.Session()

>>> print(sess.run(hello))

b'Hello, TensorFlow!'

>>> a = tf.constant(10)

>>> b = tf.constant(94)

>>> print(sess.run(a + b))

104

>>>

sudo apt-get install pkg-config zip g++ zlib1g-dev unziphttps://m.blog.naver.com/PostView.nhn?blogId=chandong83&logNo=221118855572&proxyReferer=https%3A%2F%2Fwww.google.co.kr%2F

https://blog.naver.com/chandong83/221115327298

https://github.com/bazelbuild/bazel

https://github.com/samjabrahams/tensorflow-on-raspberry-pi/blob/master/GUIDE.md#2-install-a-memory-drive-as-swap-for-compiling

바젤 릴리즈

https://github.com/bazelbuild/bazel/releases

준비사항

스왑영역을 위한 1GB이상의 메모리

스왑영역 설정

sudo umount /dev/sda1

sudo mkswap

필수패키지설치

바젤

sudo apt-get install pkg-config zip g++ zlib1g-dev unzip

텐서플로우

sudo apt-get install python3-pip python3-numpy swig python3-dev

sudo pip3 install wheel

sudo pip3 uninstall mock

sudo pip3 install mock

바젤 다운로드

 wget https://github.com/bazelbuild/bazel/releases/download/0.12.0/bazel-0.12.0-dist.zip

압축해제

unzip bazel-0.12.0-dist.zip

스크립트 수정

vi scripts/bootstrap/compile.sh

run "${JAVAC}" -classpath "${classpath}" -sourcepath "${sourcepath}" \

      -d "${output}/classes" -source "$JAVA_VERSION" -target "$JAVA_VERSION" \

      -encoding UTF-8 "@${paramfile}" 

뒤에 -J-Xmx500M 추가

run "${JAVAC}" -classpath "${classpath}" -sourcepath "${sourcepath}" \

      -d "${output}/classes" -source "$JAVA_VERSION" -target "$JAVA_VERSION" \

      -encoding UTF-8 "@${paramfile}" -J-Xmx500M

vi tools/cpp/cc_configure.bzl

빌드

sudo ./compile.ph

바이너리 복사

sudo cp output/bazel /usr/local/bin/bazel

설치 확인

$ bazel

텐서플로우 설치

git clone --recurse-submodules https://github.com/tensorflow/tensorflow.git

cd tf

cd tensorflow

git checkout

설정

./configure

bazel build -c opt --copt="-mfpu=neon-vfpv4" --copt="-funsafe-math-optimizations" --copt="-ftree-vectorize" --copt="-fomit-frame-pointer" --local_resources 1024,1.0,1.0 --verbose_failures tensorflow/tools/pip_package:build_pip_package

바질을이용한 텐서플로우 빌드

 !! 몇시간 이상 걸릴수 있습니다.

bazel build -c opt --copt="-mfpu=neon-vfpv4" --copt="-funsafe-math-optimizations" --copt="-ftree-vectorize" --copt="-fomit-frame-pointer" --local_resources 1024,1.0,1.0 --verbose_failures tensorflow/tools/pip_package:build_pip_package

https://github.com/samjabrahams/tensorflow-on-raspberry-pi

https://github.com/samjabrahams/tensorflow-on-raspberry-pi/blob/master/GUIDE.md