форкнуто от main/is_dnn
Вы не можете выбрать более 25 тем
Темы должны начинаться с буквы или цифры, могут содержать дефисы(-) и должны содержать не более 35 символов.
122 KiB
122 KiB
from google.colab import drive
drive.mount('/content/drive')
import os
os.chdir('/content/drive/MyDrive/Colab Notebooks/is_lab3')
from tensorflow import keras
from tensorflow.keras import layers
from tensorflow.keras.models import Sequential
import matplotlib.pyplot as plt
import numpy as np
from sklearn.metrics import classification_report, confusion_matrix
from sklearn.metrics import ConfusionMatrixDisplayDrive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True).
from keras.datasets import mnist
(X_train, y_train), (X_test, y_test) = mnist.load_data()from sklearn.model_selection import train_test_split
# объединяем в один набор
X = np.concatenate((X_train, X_test))
y = np.concatenate((y_train, y_test))
# разбиваем по вариантам
X_train, X_test, y_train, y_test = train_test_split(X, y,
test_size = 10000,
train_size = 60000,
random_state = 7)
# вывод размерностей
print('Shape of X train:', X_train.shape)
print('Shape of y train:', y_train.shape)
print('Shape of X test:', X_test.shape)
print('Shape of y test:', y_test.shape)Shape of X train: (60000, 28, 28)
Shape of y train: (60000,)
Shape of X test: (10000, 28, 28)
Shape of y test: (10000,)
# Зададим параметры данных и модели
num_classes = 10
input_shape = (28, 28, 1)
# Приведение входных данных к диапазону [0, 1]
X_train = X_train / 255
X_test = X_test / 255
# Расширяем размерность входных данных, чтобы каждое изображение имело
# размерность (высота, ширина, количество каналов)
X_train = np.expand_dims(X_train, -1)
X_test = np.expand_dims(X_test, -1)
print('Shape of transformed X train:', X_train.shape)
print('Shape of transformed X test:', X_test.shape)
# переведем метки в one-hot
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)
print('Shape of transformed y train:', y_train.shape)
print('Shape of transformed y test:', y_test.shape)Shape of transformed X train: (60000, 28, 28, 1)
Shape of transformed X test: (10000, 28, 28, 1)
Shape of transformed y train: (60000, 10)
Shape of transformed y test: (10000, 10)
# создаем модель
model = Sequential()
model.add(layers.Conv2D(32, kernel_size=(3, 3), activation="relu", input_shape=input_shape))
model.add(layers.MaxPooling2D(pool_size=(2, 2)))
model.add(layers.Conv2D(64, kernel_size=(3, 3), activation="relu"))
model.add(layers.MaxPooling2D(pool_size=(2, 2)))
model.add(layers.Dropout(0.5))
model.add(layers.Flatten())
model.add(layers.Dense(num_classes, activation="softmax"))
model.summary()/usr/local/lib/python3.12/dist-packages/keras/src/layers/convolutional/base_conv.py:113: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
super().__init__(activity_regularizer=activity_regularizer, **kwargs)
Model: "sequential_1"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ┃ Layer (type) ┃ Output Shape ┃ Param # ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ │ conv2d_2 (Conv2D) │ (None, 26, 26, 32) │ 320 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ max_pooling2d_2 (MaxPooling2D) │ (None, 13, 13, 32) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_3 (Conv2D) │ (None, 11, 11, 64) │ 18,496 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ max_pooling2d_3 (MaxPooling2D) │ (None, 5, 5, 64) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dropout_1 (Dropout) │ (None, 5, 5, 64) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ flatten_1 (Flatten) │ (None, 1600) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dense_1 (Dense) │ (None, 10) │ 16,010 │ └─────────────────────────────────┴────────────────────────┴───────────────┘
Total params: 34,826 (136.04 KB)
Trainable params: 34,826 (136.04 KB)
Non-trainable params: 0 (0.00 B)
# компилируем и обучаем модель
batch_size = 512
epochs = 15
model.compile(loss="categorical_crossentropy", optimizer="adam", metrics=["accuracy"])
model.fit(X_train, y_train, batch_size=batch_size, epochs=epochs, validation_split=0.1)Epoch 1/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 13s 61ms/step - accuracy: 0.6299 - loss: 1.2590 - val_accuracy: 0.9473 - val_loss: 0.1893
Epoch 2/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 11ms/step - accuracy: 0.9374 - loss: 0.2104 - val_accuracy: 0.9685 - val_loss: 0.1111
Epoch 3/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9607 - loss: 0.1312 - val_accuracy: 0.9757 - val_loss: 0.0825
Epoch 4/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9687 - loss: 0.1022 - val_accuracy: 0.9788 - val_loss: 0.0674
Epoch 5/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9702 - loss: 0.0913 - val_accuracy: 0.9812 - val_loss: 0.0597
Epoch 6/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9756 - loss: 0.0814 - val_accuracy: 0.9843 - val_loss: 0.0522
Epoch 7/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9783 - loss: 0.0720 - val_accuracy: 0.9837 - val_loss: 0.0488
Epoch 8/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9802 - loss: 0.0644 - val_accuracy: 0.9865 - val_loss: 0.0447
Epoch 9/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9812 - loss: 0.0596 - val_accuracy: 0.9862 - val_loss: 0.0430
Epoch 10/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9818 - loss: 0.0568 - val_accuracy: 0.9858 - val_loss: 0.0432
Epoch 11/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 12ms/step - accuracy: 0.9843 - loss: 0.0515 - val_accuracy: 0.9863 - val_loss: 0.0406
Epoch 12/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 12ms/step - accuracy: 0.9839 - loss: 0.0523 - val_accuracy: 0.9890 - val_loss: 0.0372
Epoch 13/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 11ms/step - accuracy: 0.9832 - loss: 0.0538 - val_accuracy: 0.9892 - val_loss: 0.0346
Epoch 14/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9856 - loss: 0.0456 - val_accuracy: 0.9892 - val_loss: 0.0337
Epoch 15/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9856 - loss: 0.0446 - val_accuracy: 0.9893 - val_loss: 0.0335
<keras.src.callbacks.history.History at 0x7f3a0efe6ae0># Оценка качества работы модели на тестовых данных
scores = model.evaluate(X_test, y_test)
print('Loss on test data:', scores[0])
print('Accuracy on test data:', scores[1])313/313 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9865 - loss: 0.0462
Loss on test data: 0.04353996366262436
Accuracy on test data: 0.9876000285148621
# вывод тестового изображения и результата распознавания
for n in [222,333]:
result = model.predict(X_test[n:n+1])
print('NN output:', result)
plt.imshow(X_test[n].reshape(28,28), cmap=plt.get_cmap('gray'))
plt.show()
print('Real mark: ', np.argmax(y_test[n]))
print('NN answer: ', np.argmax(result))1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 487ms/step
NN output: [[4.8054830e-04 4.6004916e-05 9.9381435e-01 4.3485279e-04 1.4458959e-04
4.9780807e-05 2.4334698e-03 8.0680138e-06 2.3864885e-03 2.0190493e-04]]
Real mark: 2
NN answer: 2
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 131ms/step
NN output: [[2.1822505e-04 1.3631233e-02 2.7562436e-02 8.6995798e-01 6.5456913e-03
2.4161236e-02 1.1778784e-03 3.9153613e-04 4.9376119e-02 6.9777733e-03]]
Real mark: 3
NN answer: 3
# истинные метки классов
true_labels = np.argmax(y_test, axis=1)
# предсказанные метки классов
predicted_labels = np.argmax(model.predict(X_test), axis=1)
# отчет о качестве классификации
print(classification_report(true_labels, predicted_labels))
# вычисление матрицы ошибок
conf_matrix = confusion_matrix(true_labels, predicted_labels)
# отрисовка матрицы ошибок в виде "тепловой карты"
display = ConfusionMatrixDisplay(confusion_matrix=conf_matrix)
display.plot()
plt.show()313/313 ━━━━━━━━━━━━━━━━━━━━ 1s 2ms/step
precision recall f1-score support
0 0.99 0.99 0.99 968
1 1.00 0.99 0.99 1087
2 0.99 0.99 0.99 1000
3 0.99 0.98 0.99 1039
4 0.99 0.99 0.99 966
5 0.98 0.99 0.99 908
6 0.99 0.99 0.99 972
7 0.98 0.99 0.98 1060
8 0.98 0.98 0.98 1015
9 0.98 0.98 0.98 985
accuracy 0.99 10000
macro avg 0.99 0.99 0.99 10000
weighted avg 0.99 0.99 0.99 10000
# загрузка собственного изображения
from PIL import Image
file_data = Image.open('7.png')
file_data = file_data.convert('L') # перевод в градации серого
test_img = np.array(file_data)
# вывод собственного изображения
plt.imshow(test_img, cmap=plt.get_cmap('gray'))
plt.show()
# предобработка
test_img = test_img / 255
test_img = np.reshape(test_img, (1,28,28,1))
# распознавание
result = model.predict(test_img)
print('I think it\'s ', np.argmax(result))
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 28ms/step
I think it's 7
# загрузка собственного изображения
from PIL import Image
file_data = Image.open('5.png')
file_data = file_data.convert('L') # перевод в градации серого
test_img = np.array(file_data)
# вывод собственного изображения
plt.imshow(test_img, cmap=plt.get_cmap('gray'))
plt.show()
# предобработка
test_img = test_img / 255
test_img = np.reshape(test_img, (1,28,28,1))
# распознавание
result = model.predict(test_img)
print('I think it\'s ', np.argmax(result))
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 28ms/step
I think it's 5
model = keras.models.load_model("best_model.keras")
model.summary()Model: "sequential_3"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ┃ Layer (type) ┃ Output Shape ┃ Param # ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ │ dense_4 (Dense) │ (None, 300) │ 235,500 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dense_5 (Dense) │ (None, 10) │ 3,010 │ └─────────────────────────────────┴────────────────────────┴───────────────┘
Total params: 238,512 (931.69 KB)
Trainable params: 238,510 (931.68 KB)
Non-trainable params: 0 (0.00 B)
Optimizer params: 2 (12.00 B)
# развернем каждое изображение 28*28 в вектор 784
X_train, X_test, y_train, y_test = train_test_split(X, y,
test_size = 10000,
train_size = 60000,
random_state = 7)
num_pixels = X_train.shape[1] * X_train.shape[2]
X_train = X_train.reshape(X_train.shape[0], num_pixels) / 255
X_test = X_test.reshape(X_test.shape[0], num_pixels) / 255
print('Shape of transformed X train:', X_train.shape)
print('Shape of transformed X train:', X_test.shape)
# переведем метки в one-hot
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)
print('Shape of transformed y train:', y_train.shape)
print('Shape of transformed y test:', y_test.shape)Shape of transformed X train: (60000, 784)
Shape of transformed X train: (10000, 784)
Shape of transformed y train: (60000, 10)
Shape of transformed y test: (10000, 10)
# Оценка качества работы модели на тестовых данных
scores = model.evaluate(X_test, y_test)
print('Loss on test data:', scores[0])
print('Accuracy on test data:', scores[1])313/313 ━━━━━━━━━━━━━━━━━━━━ 1s 3ms/step - accuracy: 0.9016 - loss: 0.3667
Loss on test data: 0.37091827392578125
Accuracy on test data: 0.9013000130653381