Fix up : Configurable attribute "copts" doesn't match this configuration

ERROR: /home/lenger/.cache/bazel/_bazel_lenger/e744db622218e02ee1e2cbf3b8750f17/external/nsync/BUILD:402:13: Configurable attribute "copts" doesn't match this configuration (would a default condition help?).
Conditions checked:
 @nsync//:android_arm
 @nsync//:android_arm64
 @nsync//:android_armeabi
 @nsync//:android_x86_32
 @nsync//:android_x86_64
 @nsync//:clang_macos_x86_64
 @nsync//:gcc_linux_aarch64
 @nsync//:gcc_linux_ppc64
 @nsync//:gcc_linux_x86_64_1
 @nsync//:gcc_linux_x86_64_2
 @nsync//:ios_x86_64
 @nsync//:msvc_windows_x86_64.

Solution:
vim /home/lenger/.cache/bazel/_bazel_lenger/e744db622218e02ee1e2cbf3b8750f17/external/nsync/BUILD


Add a default conditions there as marked as red line below:

NSYNC_OPTS_GENERIC = select({
    # Select the CPU architecture include directory.
    # This select() has no real effect in the C++11 build, but satisfies a
    # #include that would otherwise need a #if.
    ":gcc_linux_x86_64_1": ["-I" + pkg_path_name() + "/platform/x86_64"],
    ":gcc_linux_x86_64_2": ["-I" + pkg_path_name() + "/platform/x86_64"],
    ":gcc_linux_aarch64": ["-I" + pkg_path_name() + "/platform/aarch64"],
    ":gcc_linux_ppc64": ["-I" + pkg_path_name() + "/platform/ppc64"],
    ":clang_macos_x86_64": ["-I" + pkg_path_name() + "/platform/x86_64"],
    ":ios_x86_64": ["-I" + pkg_path_name() + "/platform/x86_64"],
    ":android_x86_32": ["-I" + pkg_path_name() + "/platform/x86_32"],
    ":android_x86_64": ["-I" + pkg_path_name() + "/platform/x86_64"],
    ":android_armeabi": ["-I" + pkg_path_name() + "/platform/arm"],
    ":android_arm": ["-I" + pkg_path_name() + "/platform/arm"],
    ":android_arm64": ["-I" + pkg_path_name() + "/platform/aarch64"],
    ":msvc_windows_x86_64": ["-I" + pkg_path_name() + "/platform/x86_64"],
    "//conditions:default": [],
}) + [
    "-I" + pkg_path_name() + "/public",
    "-I" + pkg_path_name() + "/internal",
    "-I" + pkg_path_name() + "/platform/posix",
] + select({
    ":msvc_windows_x86_64": [
    ],
    "//conditions:default": [
        "-D_POSIX_C_SOURCE=200809L",
        "-pthread",
    ],
})

Cross compile tensorflow for armv7l targets via bazel

Refer:
https://github.com/bazelbuild/bazel/wiki/Building-with-a-custom-toolchain
https://www.tensorflow.org/tutorials/image_recognition

Generate image dataset from one folder, label each image one by one, for machine learning via tensorflow

This is an update for previous post.
http://lengerrong.blogspot.com/2017/04/create-your-own-dataset-for-machine.html

It is hard to collect enough images for your machine traning.
So I update this post to append images once you find some to your existed dataset.
All you need to do is copy your images to a folder and run the python script.

Of cause, you have to label each image while running the python script.
Below python script just have two label, 0 means cat, 1 means dog.
You can add more labels by just modify blue codes.

How to fix tf.nn.in_top_k out of range issue: tensorflow.python.framework.errors_impl.InvalidArgumentError: targets[0] is out of range

tensorflow.python.framework.errors_impl.InvalidArgumentError: targets[0] is out of range
         [[Node: InTopK = InTopK[T=DT_INT32, k=1, _device="/job:localhost/replica:0/task:0/cpu:0"](softmax_linear/softmax_linear, Reshape_1)]]

Caused by op u'InTopK', defined at:

Create your own dataset for machine learning, same format as CIFAR-10 dataset, via PIL and numpy

Refers:
http://www.cs.toronto.edu/~kriz/cifar.html
https://www.tensorflow.org/tutorials/deep_cnn


The CIFAR-10 dataset consists of 60000 32x32 colour images in 10 classes.

ClixSense Click on the cat job use the image that its size is 128 x 96 px, and
the image is cat or dog, no other type.
I want to use CNN to do machine learning for this job. 
I collected some images now.

The key python codes to append one image with label to the dataset is as below:

from PIL import Image
import numpy as np   im = Image.open(filename)
  im = (np.array(im))

  r = im[:,:,0].flatten()
  g = im[:,:,1].flatten()
  b = im[:,:,2].flatten()

  if iscat:
    label = [0]
  else:
    label = [1]

  out  = np.array(list(label) + list(r) + list(g) + list(b), np.uint8)
  out.tofile(dataset)

Installing tensorflow with CPU support only from source on Ubuntu 16.04

Refers :
https://www.tensorflow.org/install/install_sources
https://bazel.build/versions/master/docs/install.html