怎樣使用tensorflow導入已經下載好的mnist數據集?

我在學習tensorflow入門教程的時候，由於網路原因沒有從腳本下載mnist手寫識別數據集，於是我就手動下載的數據集，並把它放在相應的目錄上。但是我看極客學院的教程上沒有提供相應的從文件夾當中插入數據的方法。那麼應該怎樣直接導入數據集呢？

下載好的文件放到代碼目錄下就可以了呀，調用read_data_sets()的時候第一個路徑參數寫"."，表示當前目錄，mnist.py會判斷是否下載，因為源碼里調用的函數是maybe_download()，這個函數會判斷本地是否已經有文件了。

如果還不行，就直接閱讀tensorflow/contrib/learn/python/learn/datasets/mnist.py源文件，將相應的讀取代碼稍加重組織粘貼到程序中。

from tensorflow.examples.tutorials.mnist import input_data

MNIST_data_folder="" mnist data 的文件夾""

mnist=input_data.read_data_sets(MNIST_data_folder,one_hot=True)

print mnist.train.next_batch(1)

import tensorflow.examples.tutorials.mnist.input_data as input_data

mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)


from tensorflow.examples.tutorials.mnist import input_data

MNIST_data_folder="D:\TestTensorflow\……\MNIST_data"

mnist=input_data.read_data_sets(MNIST_data_folder,one_hot=True)

把本地下載好的數據路徑加進去，要用\

The Environment: Ubuntu + Anaconda + Jupyter notebook

My Anaconda is installed in default in the directory of the "/home"

When run the code below in the reference documentation:

from tensorflow.examples.tutorials.mnist import input_data
import tensorflow as tf
mnist = input_data.read_data_sets("MNIST_data", one_hot=True)
...


In default, the program would download the dataset from the official website.

Since you have downloaded the MNIST dataset, named (they all are not unpacked) :

• t10k-images-idx3-ubyte.gz
• t10k-labels-idx1-ubyte.gz
• train-images-idx3-ubyte.gz
• train-labels-idx1-ubyte.gz

You should put them in the "/home/anaconda/"MNIST_data""

Then the input_data.py would recognize then folder "MNIST_data" have dataset or not, if yes, you load the downloading dataset successfully.