raspberry pi 3にTensorFlowを入れて画像認識させる

こいつの続き、ラズパイ３にTensorFlowを入れるところから。

これでわしもきゅうり判別機を作れるだろうかｗ。

・TensorFlowを入れる

GitHub - samjabrahams/tensorflow-on-raspberry-pi: TensorFlow for Raspberry Pi

pi@raspberrypi:~ $ wget https://github.com/samjabrahams/tensorflow-on-raspberry-pi/releases/download/v1.1.0/tensorflow-1.1.0-cp27-none-linux_armv7l.whl

pi@raspberrypi:~ $ sudo pip install tensorflow-1.1.0-cp27-none-linux_armv7l.whl

pi@raspberrypi:~ $ sudo pip uninstall mock

pi@raspberrypi:~ $ sudo pip install mock

・確認

pi@raspberrypi:~ $ python
Python 2.7.9 (default, Sep 17 2016, 20:26:04)
[GCC 4.9.2] on linux2
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)
Hello, TensorFlow!
>>> a = tf.constant(10)
>>> b = tf.constant(32)
>>> print sess.run(a + b)
42
>>>

・kerasを入れる（かなり時間かかった。。）

pi@raspberrypi:~ $ sudo pip install keras

・学習済みのモデルを持ってくる

pi@raspberrypi:~ $ git clone https://github.com/fchollet/deep-learning-models

・実行 → 怒られた

pi@raspberrypi:~/deep-learning-models $ python inception_v3.py
Using TensorFlow backend.
Downloading data from https://github.com/fchollet/deep-learning-models/releases/download/v0.5/inception_v3_weights_tf_dim_ordering_tf_kernels.h5
Traceback (most recent call last):
File "inception_v3.py", line 399, in <module>
    model = InceptionV3(include_top=True, weights='imagenet')
File "inception_v3.py", line 385, in InceptionV3
    model.load_weights(weights_path)
File "/usr/local/lib/python2.7/dist-packages/keras/engine/topology.py", line 2531, in load_weights
    raise ImportError('`load_weights` requires h5py.')
ImportError: `load_weights` requires h5py.
pi@raspberrypi:~/deep-learning-models $

・h5pyっていうのがいるの？ → pipで入れる → 怒られた

pi@raspberrypi:~ $ sudo pip install h5py

※中略

In file included from /tmp/pip-build-1n231R/h5py/h5py/defs.c:470:0:
    /tmp/pip-build-1n231R/h5py/h5py/api_compat.h:27:18: fatal error: hdf5.h: No such file or directory
     #include "hdf5.h"
                      ^
    compilation terminated.
    error: command 'arm-linux-gnueabihf-gcc' failed with exit status 1

・必要なものをいれる

pi@raspberrypi:~ $ sudo apt-get install -y libhdf5-dev

・とりあえずh5pyっていうのが入った

pi@raspberrypi:~ $ sudo pip install h5py
Collecting h5py
Using cached h5py-2.7.0.tar.gz
Requirement already satisfied: numpy>=1.7 in /usr/local/lib/python2.7/dist-packages (from h5py)
Requirement already satisfied: six in /usr/local/lib/python2.7/dist-packages (from h5py)
Building wheels for collected packages: h5py
Running setup.py bdist_wheel for h5py ... done
Stored in directory: /root/.cache/pip/wheels/3e/ff/67/d04f183c668c251c1bc737224584dec995715d486178f77ede
Successfully built h5py
Installing collected packages: h5py
Successfully installed h5py-2.7.0
pi@raspberrypi:~ $

・これで行けるか？ → 失敗。。。

pi@raspberrypi:~/deep-learning-models $ python inception_v3.py
Using TensorFlow backend.
Traceback (most recent call last):
File "inception_v3.py", line 402, in <module>
img = image.load_img(img_path, target_size=(299, 299))
File "/usr/local/lib/python2.7/dist-packages/keras/preprocessing/image.py", line 318, in load_img
raise ImportError('Could not import PIL.Image. '
ImportError: Could not import PIL.Image. The use of `array_to_img` requires PIL.
pi@raspberrypi:~/deep-learning-models $

・これがいる模様 → pillow

pi@raspberrypi:~/deep-learning-models $ sudo pip install pillow

・必要なもの入ったんじゃね？

pi@raspberrypi:~/deep-learning-models $ python
Python 2.7.9 (default, Sep 17 2016, 20:26:04)
[GCC 4.9.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import PIL
>>>

・来たーー！！？？？

pi@raspberrypi:~/deep-learning-models $ python inception_v3.py
Using TensorFlow backend.
Downloading data from https://s3.amazonaws.com/deep-learning-models/image-models/imagenet_class_index.json
Predicted: (u'n02504458', u'African_elephant', 0.58458394), (u'n01871265', u'tusker', 0.35385633), (u'n02504013', u'Indian_elephant', 0.01978549), (u'n02134084', u'ice_bear', 0.0005529598), (u'n02437312', u'Arabian_camel', 0.00053362496)
pi@raspberrypi:~/deep-learning-models $

※この画像

f:id:ken5owata:20170612223613j:plain