In [1]: import numpy as np # linear algebra

import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)

import tensorflow as tf
import matplotlib.pyplot as plt

In [4]: digits = tf.keras.datasets.mnist.load_data()

digits

((array([[[0, 0, 0, ..., 0, 0, 0],

Out[4]:
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

...,

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]]], dtype=uint8),
array([5, 0, 4, ..., 5, 6, 8], dtype=uint8)),
(array([[[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

...,

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]],

[[0, 0, 0, ..., 0, 0, 0],

[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
...,
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0],
[0, 0, 0, ..., 0, 0, 0]]], dtype=uint8),
array([7, 2, 1, ..., 4, 5, 6], dtype=uint8)))

In [6]: print(len(digits[0][1]))

(x_train, y_train), (x_test, y_test) = digits

60000

In [7]: print(len(x_train),len(y_train))
print(len(x_test),len(y_test))

print(x_train[0].shape)
print(x_test[0].shape)

60000 60000
10000 10000
(28, 28)
(28, 28)

In [8]: for i in range(4):

plt.figure()
plt.imshow(x_train[i])
plt.colorbar()
plt.grid(False)
plt.show()

In [9]: x_train = x_train/255.0

x_test = x_test/255.0

In [10]: class_names = ['0','1', '2', '3', '4','5','6','7','8','9']

In [11]: plt.figure(figsize=(10,10))
for i in range(25):
plt.subplot(5,5,i+1)
plt.xticks([])
plt.yticks([])
plt.grid(False)
plt.imshow(x_train[i], cmap=plt.cm.binary)
plt.show()

In [12]: model = tf.keras.Sequential([

tf.keras.layers.Flatten(input_shape=(28, 28)),
tf.keras.layers.Dense(128, activation='relu'),
tf.keras.layers.Dense(10)
])

In [13]: model.compile(optimizer='adam',
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])

In [15]: model.fit(x_train,y_train,epochs = 15)

Epoch 1/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0136 - accuracy: 0.9956
Epoch 2/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0116 - accuracy: 0.9965
Epoch 3/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0095 - accuracy: 0.9971
Epoch 4/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0097 - accuracy: 0.9969
Epoch 5/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0079 - accuracy: 0.9977
Epoch 6/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0067 - accuracy: 0.9979
Epoch 7/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0074 - accuracy: 0.9975
Epoch 8/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0058 - accuracy: 0.9982
Epoch 9/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0046 - accuracy: 0.9986
Epoch 10/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0070 - accuracy: 0.9975
Epoch 11/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0037 - accuracy: 0.9989
Epoch 12/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0061 - accuracy: 0.9980
Epoch 13/15
1875/1875 [==============================] - 3s 1ms/step - loss: 0.0039 - accuracy: 0.9988
Epoch 14/15
1875/1875 [==============================] - 3s 2ms/step - loss: 0.0048 - accuracy: 0.9985
Epoch 15/15
1875/1875 [==============================] - 3s 1ms/step - loss: 0.0049 - accuracy: 0.9984
<keras.callbacks.History at 0x16be8530370>
Out[15]:

In [16]: test_loss, test_acc = model.evaluate(x_test, y_test, verbose=2)

print('\nTest accuracy:', test_acc)

313/313 - 0s - loss: 0.1110 - accuracy: 0.9802 - 467ms/epoch - 1ms/step

Test accuracy: 0.9801999926567078

In [18]: pred = tf.keras.Sequential([model,tf.keras.layers.Softmax()])

predict = pred.predict(x_test)

313/313 [==============================] - 0s 986us/step

In [19]: plt.matshow(x_test[0])

<matplotlib.image.AxesImage at 0x16be8533490>
Out[19]:

In [20]: np.argmax(predict[0])

7
Out[20]:

In [21]: def plot_image(i, predictions_array, true_label, img):

true_label, img = true_label[i], img[i]
plt.grid(False)
plt.xticks([])
plt.yticks([])

plt.imshow(img, cmap=plt.cm.binary)

predicted_label = np.argmax(predictions_array)
if predicted_label == true_label:
color = 'blue'
else:
color = 'red'

plt.xlabel("{} {:2.0f}% ({})".format(class_names[predicted_label],

100*np.max(predictions_array),
class_names[true_label]),
color=color)

def plot_value_array(i, predictions_array, true_label):

true_label = true_label[i]
plt.grid(False)
plt.xticks(range(10))
plt.yticks([])
thisplot = plt.bar(range(10), predictions_array, color="#777777")
plt.ylim([0, 1])
predicted_label = np.argmax(predictions_array)

thisplot[predicted_label].set_color('red')
thisplot[true_label].set_color('blue')

In [22]: i = 0
plt.figure(figsize=(6,3))
plt.subplot(1,2,1)
plot_image(i, predict[i], y_test, x_test)
plt.subplot(1,2,2)
plot_value_array(i, predict[i], y_test)
plt.show()

In [23]: i = 3
plt.figure(figsize=(6,3))
plt.subplot(1,2,1)
plot_image(i, predict[i], y_test, x_test)
plt.subplot(1,2,2)
plot_value_array(i, predict[i], y_test)
plt.show()

In [24]: i = 12
plt.figure(figsize=(6,3))
plt.subplot(1,2,1)
plot_image(i, predict[i], y_test, x_test)
plt.subplot(1,2,2)
plot_value_array(i, predict[i], y_test)
plt.show()7

In [25]: num_rows = 5
num_cols = 3
num_images = num_rows*num_cols
plt.figure(figsize=(2*2*num_cols, 2*num_rows))
for i in range(num_images):
plt.subplot(num_rows, 2*num_cols, 2*i+1)
plot_image(i, predict[i], y_test, x_test)
plt.subplot(num_rows, 2*num_cols, 2*i+2)
plot_value_array(i, predict[i], y_test)
plt.tight_layout()
plt.show()

In [26]: model.save("Digits_Recognize.h5")