TensorFlow & DarkFlow & Yolo v2をDocker化する

Object Detectionでyolo v2を使ってみたいと思って調べていたけど
結構うまく構築できないことがあったのでDockerfile　<仮>にしちゃおうってことにしました。

実際にいろいろいじるのはこれからなので
必要なライブラリとかバージョンの互換性とかは
今後順次修正していきますよ(´・ω・｀)

構築する環境

Python 3.6.0 :: Anaconda 4.3.0 (64-bit)
darkflow
TensorFlow

Dockerfile

FROM  centos:7
LABEL noffuy <3330072-noffuy@users.noreply.gitlab.com>

RUN yum install -y bzip2 bzip2-devel gcc gcc-c++ git make wget curl openssl-devel readline-devel zlib-devel unzip vim libSM-devel libSM libXrender libXrender-devel libXext libXext-devel \
    && wget "https://github.com/pyenv/pyenv/archive/v1.2.8.zip" -O /pyenv.zip \
    && unzip /pyenv.zip \
    && rm /pyenv.zip \
    && mv /pyenv-1.2.8 /pyenv \
    && echo 'export PYENV_ROOT="/pyenv"' >> /etc/bashrc \
    && echo 'export PATH="$PYENV_ROOT/bin:$PATH"' >> /etc/bashrc \
    && echo 'eval "$(pyenv init -)"' >> /etc/bashrc \
    && source /etc/bashrc \
    && export PYENV_ROOT="/pyenv" && export PATH="$PYENV_ROOT/bin:$PATH" && eval "$(pyenv init -)" \
    && pyenv install anaconda3-4.3.0 \
    && pyenv rehash \
    && pyenv global anaconda3-4.3.0 \
    && echo 'export PATH="$PYENV_ROOT/versions/anaconda3-4.3.0/bin/:$PATH"' >> /etc/bashrc \
    && source /etc/bashrc \
    && export PATH="$PYENV_ROOT/versions/anaconda3-4.3.0/bin/:$PATH" \
    && conda update conda \
    && conda create -n tensorflow python=3.6.0 \
    && source activate tensorflow \
    && export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.0.1-cp36-cp36m-linux_x86_64.whl \
    && pip install --ignore-installed --upgrade $TF_BINARY_URL\
    && pip install Cython opencv-python numpy pillow \
    && git clone https://github.com/thtrieu/darkflow.git && cd darkflow && pip install -e .

RUN mkdir /app

ENV LANG ja_JP.UTF-8
ENV LC_CTYPE "ja_JP.UTF-8"
ENV LC_NUMERIC "ja_JP.UTF-8"
ENV LC_TIME "ja_JP.UTF-8"
ENV LC_COLLATE "ja_JP.UTF-8"
ENV LC_MONETARY "ja_JP.UTF-8"
ENV LC_MESSAGES "ja_JP.UTF-8"
ENV LC_PAPER "ja_JP.UTF-8"
ENV LC_NAME "ja_JP.UTF-8"
ENV LC_ADDRESS "ja_JP.UTF-8"
ENV LC_TELEPHONE "ja_JP.UTF-8"
ENV LC_MEASUREMENT "ja_JP.UTF-8"
ENV LC_IDENTIFICATION "ja_JP.UTF-8"

WORKDIR /app

FROM centos:7

LABEL noffuy <3330072-noffuy@users.noreply.gitlab.com>

RUN yum install -y bzip2 bzip2-devel gcc gcc-c++ git make wget curl openssl-devel readline-devel zlib-devel unzip vim libSM-devel libSM libXrender libXrender-devel libXext libXext-devel \

&& wget "https://github.com/pyenv/pyenv/archive/v1.2.8.zip" -O /pyenv.zip \

&& unzip /pyenv.zip \

&& rm /pyenv.zip \

&& mv /pyenv-1.2.8 /pyenv \

&& echo 'export PYENV_ROOT="/pyenv"' >> /etc/bashrc \

&& echo 'export PATH="$PYENV_ROOT/bin:$PATH"' >> /etc/bashrc \

&& echo 'eval "$(pyenv init -)"' >> /etc/bashrc \

&& source /etc/bashrc \

&& export PYENV_ROOT="/pyenv" && export PATH="$PYENV_ROOT/bin:$PATH" && eval "$(pyenv init -)" \

&& pyenv install anaconda3-4.3.0 \

&& pyenv rehash \

&& pyenv global anaconda3-4.3.0 \

&& echo 'export PATH="$PYENV_ROOT/versions/anaconda3-4.3.0/bin/:$PATH"' >> /etc/bashrc \

&& source /etc/bashrc \

&& export PATH="$PYENV_ROOT/versions/anaconda3-4.3.0/bin/:$PATH" \

&& conda update conda \

&& conda create -n tensorflow python=3.6.0 \

&& source activate tensorflow \

&& export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.0.1-cp36-cp36m-linux_x86_64.whl \

&& pip install --ignore-installed --upgrade $TF_BINARY_URL\

&& pip install Cython opencv-python numpy pillow \

&& git clone https://github.com/thtrieu/darkflow.git && cd darkflow && pip install -e .

RUN mkdir /app

ENV LANG ja_JP.UTF-8

ENV LC_CTYPE "ja_JP.UTF-8"

ENV LC_NUMERIC "ja_JP.UTF-8"

ENV LC_TIME "ja_JP.UTF-8"

ENV LC_COLLATE "ja_JP.UTF-8"

ENV LC_MONETARY "ja_JP.UTF-8"

ENV LC_MESSAGES "ja_JP.UTF-8"

ENV LC_PAPER "ja_JP.UTF-8"

ENV LC_NAME "ja_JP.UTF-8"

ENV LC_ADDRESS "ja_JP.UTF-8"

ENV LC_TELEPHONE "ja_JP.UTF-8"

ENV LC_MEASUREMENT "ja_JP.UTF-8"

ENV LC_IDENTIFICATION "ja_JP.UTF-8"

WORKDIR /app

これで、
$ docker build -t yolo .
$ docker run -it yolo /bin/bash
で起動して開発作業ができる感じです。

conda環境を作ってあるので
$ source activate tensorflow
を忘れずにね。

じゃあね〜〜〜〜。