TensorFlow provides a library, can be directly used to automatically download and install MNIST.
MNIST contains three data sets: a first set of training data is (mnist.train.images), the other two being the test data set (mnist.test.images), and validation data set (mnist.validation).
Code one_hot = True, the label indicates the sample into one_hot coding.
Print information is the beginning of the decompressed data collection means. The first time you run, will display related messages downloaded data.
Then print out a training set of picture information, it is a 55,000 line 784 of the matrix. That is, the training set, there are 55,000 pictures.
1 from tensorflow.examples.tutorials.mnist import input_data 2 mnist = input_data.read_data_sets("MNIST_data/",one_hot=True) 3 print ('输入数据:',mnist.train.images) 4 print ('输入打印shape:',mnist.train.images.shape) 5 import pylab 6 im = mnist.train.images[1] 7 im = im.reshape(-1,28) 8 pylab.imshow(im) 9 pylab.show()
FIG output of the code: