Вы не можете выбрать более 25 тем
Темы должны начинаться с буквы или цифры, могут содержать дефисы(-) и должны содержать не более 35 символов.
442 KiB
442 KiB
# импорт модулей
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 ConfusionMatrixDisplay# загрузка датасета
from keras.datasets import mnist
(X_train, y_train), (X_test, y_test) = mnist.load_data()Downloading data from https://storage.googleapis.com/tensorflow/tf-keras-datasets/mnist.npz
11490434/11490434 ━━━━━━━━━━━━━━━━━━━━ 2s 0us/step
# создание своего разбиения датасета
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 = 23)
# 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"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ┃ Layer (type) ┃ Output Shape ┃ Param # ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ │ conv2d (Conv2D) │ (None, 26, 26, 32) │ 320 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ max_pooling2d (MaxPooling2D) │ (None, 13, 13, 32) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_1 (Conv2D) │ (None, 11, 11, 64) │ 18,496 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ max_pooling2d_1 (MaxPooling2D) │ (None, 5, 5, 64) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dropout (Dropout) │ (None, 5, 5, 64) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ flatten (Flatten) │ (None, 1600) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dense (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 ━━━━━━━━━━━━━━━━━━━━ 9s 40ms/step - accuracy: 0.6012 - loss: 1.2802 - val_accuracy: 0.9483 - val_loss: 0.1785
Epoch 2/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 11ms/step - accuracy: 0.9392 - loss: 0.2017 - val_accuracy: 0.9675 - val_loss: 0.1072
Epoch 3/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 12ms/step - accuracy: 0.9605 - loss: 0.1282 - val_accuracy: 0.9728 - val_loss: 0.0857
Epoch 4/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 11ms/step - accuracy: 0.9684 - loss: 0.1051 - val_accuracy: 0.9792 - val_loss: 0.0735
Epoch 5/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9721 - loss: 0.0891 - val_accuracy: 0.9815 - val_loss: 0.0629
Epoch 6/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9759 - loss: 0.0796 - val_accuracy: 0.9823 - val_loss: 0.0598
Epoch 7/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9779 - loss: 0.0708 - val_accuracy: 0.9845 - val_loss: 0.0574
Epoch 8/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9810 - loss: 0.0614 - val_accuracy: 0.9850 - val_loss: 0.0526
Epoch 9/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9809 - loss: 0.0618 - val_accuracy: 0.9845 - val_loss: 0.0534
Epoch 10/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9815 - loss: 0.0565 - val_accuracy: 0.9847 - val_loss: 0.0497
Epoch 11/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9828 - loss: 0.0538 - val_accuracy: 0.9845 - val_loss: 0.0502
Epoch 12/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9840 - loss: 0.0504 - val_accuracy: 0.9860 - val_loss: 0.0481
Epoch 13/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 10ms/step - accuracy: 0.9841 - loss: 0.0483 - val_accuracy: 0.9872 - val_loss: 0.0468
Epoch 14/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 12ms/step - accuracy: 0.9856 - loss: 0.0464 - val_accuracy: 0.9868 - val_loss: 0.0434
Epoch 15/15
106/106 ━━━━━━━━━━━━━━━━━━━━ 1s 12ms/step - accuracy: 0.9861 - loss: 0.0432 - val_accuracy: 0.9868 - val_loss: 0.0438
<keras.src.callbacks.history.History at 0x784f00d33ec0># 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 ━━━━━━━━━━━━━━━━━━━━ 2s 4ms/step - accuracy: 0.9900 - loss: 0.0318
Loss on test data: 0.03606646880507469
Accuracy on test data: 0.9894999861717224
# вывод двух тестовых изображений и результатов распознавания
for n in [3,26]:
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 401ms/step
NN output: [[3.3479900e-08 4.7786756e-14 9.9999976e-01 2.1413245e-08 1.2324374e-11
1.0817477e-09 6.3094833e-12 4.4141193e-10 2.3786414e-07 8.6100585e-11]]
Real mark: 2
NN answer: 2
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 30ms/step
NN output: [[2.7364138e-09 7.8356831e-13 1.9827752e-08 7.4724680e-06 1.7721488e-06
2.3284849e-07 1.7225671e-11 2.4611420e-05 6.3226136e-05 9.9990261e-01]]
Real mark: 9
NN answer: 9
# 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 1.00 0.99 997
1 1.00 0.99 1.00 1164
2 0.99 0.98 0.99 1030
3 0.99 0.99 0.99 1031
4 0.99 0.99 0.99 967
5 0.98 0.99 0.99 860
6 0.99 1.00 0.99 977
7 0.98 0.99 0.99 1072
8 0.99 0.98 0.99 939
9 0.99 0.98 0.98 963
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
for name_image in ['test3.png', 'test5.png']:
file_data = Image.open(name_image)
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 29ms/step
I think it's 3
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 31ms/step
I think it's 5
model_lr1 = keras.models.load_model("best_model.keras")
model_lr1.summary()Model: "sequential_10"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ┃ Layer (type) ┃ Output Shape ┃ Param # ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ │ dense_23 (Dense) │ (None, 100) │ 78,500 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dense_24 (Dense) │ (None, 50) │ 5,050 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dense_25 (Dense) │ (None, 10) │ 510 │ └─────────────────────────────────┴────────────────────────┴───────────────┘
Total params: 84,062 (328.37 KB)
Trainable params: 84,060 (328.36 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 = 23)
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_lr1.evaluate(X_test, y_test)
print('Loss on test data:', scores[0])
print('Accuracy on test data:', scores[1])313/313 ━━━━━━━━━━━━━━━━━━━━ 3s 5ms/step - accuracy: 0.9576 - loss: 0.1293
Loss on test data: 0.13758081197738647
Accuracy on test data: 0.9567000269889832
# загрузка датасета
from keras.datasets import cifar10
(X_train, y_train), (X_test, y_test) = cifar10.load_data()Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz
170498071/170498071 ━━━━━━━━━━━━━━━━━━━━ 13s 0us/step
# создание своего разбиения датасета
# 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 = 50000,
random_state = 23)
# 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: (50000, 32, 32, 3)
Shape of y train: (50000, 1)
Shape of X test: (10000, 32, 32, 3)
Shape of y test: (10000, 1)
class_names = ['airplane', 'automobile', 'bird', 'cat', 'deer',
'dog', 'frog', 'horse', 'ship', 'truck']
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])
plt.xlabel(class_names[y_train[i][0]])
plt.show()
# Зададим параметры данных и модели
num_classes = 10
input_shape = (32, 32, 3)
# Приведение входных данных к диапазону [0, 1]
X_train = X_train / 255
X_test = X_test / 255
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: (50000, 32, 32, 3)
Shape of transformed X test: (10000, 32, 32, 3)
Shape of transformed y train: (50000, 10)
Shape of transformed y test: (10000, 10)
# создаем модель
model = Sequential()
# Блок 1
model.add(layers.Conv2D(32, (3, 3), padding="same",
activation="relu", input_shape=input_shape))
model.add(layers.BatchNormalization())
model.add(layers.Conv2D(32, (3, 3), padding="same", activation="relu"))
model.add(layers.BatchNormalization())
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Dropout(0.25))
# Блок 2
model.add(layers.Conv2D(64, (3, 3), padding="same", activation="relu"))
model.add(layers.BatchNormalization())
model.add(layers.Conv2D(64, (3, 3), padding="same", activation="relu"))
model.add(layers.BatchNormalization())
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Dropout(0.25))
# Блок 3
model.add(layers.Conv2D(128, (3, 3), padding="same", activation="relu"))
model.add(layers.BatchNormalization())
model.add(layers.Conv2D(128, (3, 3), padding="same", activation="relu"))
model.add(layers.BatchNormalization())
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Dropout(0.4))
model.add(layers.Flatten())
model.add(layers.Dense(128, activation='relu'))
model.add(layers.Dropout(0.5))
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, 32, 32, 32) │ 896 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization │ (None, 32, 32, 32) │ 128 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_3 (Conv2D) │ (None, 32, 32, 32) │ 9,248 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization_1 │ (None, 32, 32, 32) │ 128 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ max_pooling2d_2 (MaxPooling2D) │ (None, 16, 16, 32) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dropout_1 (Dropout) │ (None, 16, 16, 32) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_4 (Conv2D) │ (None, 16, 16, 64) │ 18,496 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization_2 │ (None, 16, 16, 64) │ 256 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_5 (Conv2D) │ (None, 16, 16, 64) │ 36,928 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization_3 │ (None, 16, 16, 64) │ 256 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ max_pooling2d_3 (MaxPooling2D) │ (None, 8, 8, 64) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dropout_2 (Dropout) │ (None, 8, 8, 64) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_6 (Conv2D) │ (None, 8, 8, 128) │ 73,856 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization_4 │ (None, 8, 8, 128) │ 512 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_7 (Conv2D) │ (None, 8, 8, 128) │ 147,584 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization_5 │ (None, 8, 8, 128) │ 512 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ max_pooling2d_4 (MaxPooling2D) │ (None, 4, 4, 128) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dropout_3 (Dropout) │ (None, 4, 4, 128) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ flatten_1 (Flatten) │ (None, 2048) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dense_1 (Dense) │ (None, 128) │ 262,272 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dropout_4 (Dropout) │ (None, 128) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dense_2 (Dense) │ (None, 10) │ 1,290 │ └─────────────────────────────────┴────────────────────────┴───────────────┘
Total params: 552,362 (2.11 MB)
Trainable params: 551,466 (2.10 MB)
Non-trainable params: 896 (3.50 KB)
# компилируем и обучаем модель
batch_size = 64
epochs = 50
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 1/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 33s 23ms/step - accuracy: 0.2713 - loss: 2.0996 - val_accuracy: 0.4600 - val_loss: 1.4698
Epoch 2/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.4710 - loss: 1.4611 - val_accuracy: 0.5876 - val_loss: 1.2134
Epoch 3/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.5671 - loss: 1.2243 - val_accuracy: 0.6042 - val_loss: 1.2126
Epoch 4/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.6242 - loss: 1.0786 - val_accuracy: 0.6838 - val_loss: 0.8650
Epoch 5/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.6618 - loss: 0.9714 - val_accuracy: 0.7076 - val_loss: 0.8535
Epoch 6/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.7015 - loss: 0.8811 - val_accuracy: 0.7096 - val_loss: 0.8280
Epoch 7/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.7175 - loss: 0.8198 - val_accuracy: 0.7428 - val_loss: 0.7633
Epoch 8/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.7368 - loss: 0.7710 - val_accuracy: 0.7632 - val_loss: 0.6851
Epoch 9/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.7579 - loss: 0.7130 - val_accuracy: 0.7674 - val_loss: 0.6738
Epoch 10/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.7730 - loss: 0.6751 - val_accuracy: 0.8030 - val_loss: 0.5984
Epoch 11/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.7847 - loss: 0.6376 - val_accuracy: 0.7694 - val_loss: 0.7044
Epoch 12/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.7970 - loss: 0.5956 - val_accuracy: 0.7886 - val_loss: 0.6338
Epoch 13/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 10s 9ms/step - accuracy: 0.8038 - loss: 0.5751 - val_accuracy: 0.7992 - val_loss: 0.6106
Epoch 14/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8139 - loss: 0.5532 - val_accuracy: 0.8152 - val_loss: 0.5782
Epoch 15/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.8206 - loss: 0.5221 - val_accuracy: 0.8126 - val_loss: 0.5920
Epoch 16/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8288 - loss: 0.5039 - val_accuracy: 0.8266 - val_loss: 0.5345
Epoch 17/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.8327 - loss: 0.4915 - val_accuracy: 0.8188 - val_loss: 0.5628
Epoch 18/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8415 - loss: 0.4732 - val_accuracy: 0.8248 - val_loss: 0.5356
Epoch 19/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.8486 - loss: 0.4533 - val_accuracy: 0.8232 - val_loss: 0.5351
Epoch 20/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8525 - loss: 0.4387 - val_accuracy: 0.8304 - val_loss: 0.5254
Epoch 21/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8520 - loss: 0.4297 - val_accuracy: 0.7862 - val_loss: 0.7218
Epoch 22/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8589 - loss: 0.4169 - val_accuracy: 0.8384 - val_loss: 0.5035
Epoch 23/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.8636 - loss: 0.4034 - val_accuracy: 0.8192 - val_loss: 0.5844
Epoch 24/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8716 - loss: 0.3834 - val_accuracy: 0.8250 - val_loss: 0.5500
Epoch 25/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.8709 - loss: 0.3822 - val_accuracy: 0.8364 - val_loss: 0.5000
Epoch 26/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8786 - loss: 0.3592 - val_accuracy: 0.8500 - val_loss: 0.4872
Epoch 27/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.8708 - loss: 0.3783 - val_accuracy: 0.8492 - val_loss: 0.5018
Epoch 28/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8819 - loss: 0.3502 - val_accuracy: 0.8498 - val_loss: 0.4784
Epoch 29/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.8866 - loss: 0.3325 - val_accuracy: 0.8434 - val_loss: 0.5009
Epoch 30/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8878 - loss: 0.3299 - val_accuracy: 0.8450 - val_loss: 0.4781
Epoch 31/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.8902 - loss: 0.3252 - val_accuracy: 0.8550 - val_loss: 0.4734
Epoch 32/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8907 - loss: 0.3245 - val_accuracy: 0.8434 - val_loss: 0.5142
Epoch 33/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.8927 - loss: 0.3160 - val_accuracy: 0.8498 - val_loss: 0.5049
Epoch 34/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.8983 - loss: 0.3004 - val_accuracy: 0.8432 - val_loss: 0.5066
Epoch 35/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.8984 - loss: 0.3011 - val_accuracy: 0.8394 - val_loss: 0.5416
Epoch 36/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.9025 - loss: 0.2877 - val_accuracy: 0.8502 - val_loss: 0.4972
Epoch 37/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.9016 - loss: 0.2897 - val_accuracy: 0.8468 - val_loss: 0.5086
Epoch 38/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.9011 - loss: 0.2889 - val_accuracy: 0.8528 - val_loss: 0.4809
Epoch 39/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.9063 - loss: 0.2709 - val_accuracy: 0.8506 - val_loss: 0.5207
Epoch 40/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.9082 - loss: 0.2689 - val_accuracy: 0.8546 - val_loss: 0.4983
Epoch 41/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.9040 - loss: 0.2784 - val_accuracy: 0.8522 - val_loss: 0.5021
Epoch 42/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 9ms/step - accuracy: 0.9055 - loss: 0.2769 - val_accuracy: 0.8520 - val_loss: 0.5108
Epoch 43/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.9098 - loss: 0.2575 - val_accuracy: 0.8470 - val_loss: 0.5368
Epoch 44/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.9134 - loss: 0.2487 - val_accuracy: 0.8542 - val_loss: 0.4839
Epoch 45/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.9147 - loss: 0.2523 - val_accuracy: 0.8494 - val_loss: 0.5282
Epoch 46/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.9162 - loss: 0.2465 - val_accuracy: 0.8538 - val_loss: 0.4990
Epoch 47/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.9169 - loss: 0.2459 - val_accuracy: 0.8584 - val_loss: 0.4915
Epoch 48/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.9178 - loss: 0.2409 - val_accuracy: 0.8602 - val_loss: 0.4881
Epoch 49/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 7s 10ms/step - accuracy: 0.9168 - loss: 0.2516 - val_accuracy: 0.8578 - val_loss: 0.5049
Epoch 50/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 9ms/step - accuracy: 0.9204 - loss: 0.2299 - val_accuracy: 0.8524 - val_loss: 0.4976
<keras.src.callbacks.history.History at 0x784efd0525d0># 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 ━━━━━━━━━━━━━━━━━━━━ 3s 5ms/step - accuracy: 0.8549 - loss: 0.5187
Loss on test data: 0.518433153629303
Accuracy on test data: 0.8521999716758728
# вывод двух тестовых изображений и результатов распознавания
for n in [1,10]:
result = model.predict(X_test[n:n+1])
print('NN output:', result)
plt.imshow(X_test[n].reshape(32,32,3), cmap=plt.get_cmap('gray'))
plt.show()
print('Real mark: ', np.argmax(y_test[n]))
print('NN answer: ', np.argmax(result))1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 193ms/step
NN output: [[5.0897655e-05 1.5590033e-06 1.1971882e-07 5.6473659e-10 3.7635881e-09
1.3418226e-10 1.1458374e-08 1.3129012e-10 9.9994743e-01 1.9529903e-08]]
Real mark: 8
NN answer: 8
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 142ms/step
NN output: [[3.7840240e-05 8.8557690e-06 1.5065701e-02 6.9747168e-01 3.3291127e-03
6.2251734e-03 1.4282507e-02 2.6306707e-01 7.0983755e-05 4.4104352e-04]]
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, target_names=class_names))
# вычисление матрицы ошибок
conf_matrix = confusion_matrix(true_labels, predicted_labels)
# отрисовка матрицы ошибок в виде "тепловой карты"
fig, ax = plt.subplots(figsize=(6, 6))
disp = ConfusionMatrixDisplay(confusion_matrix=conf_matrix,display_labels=class_names)
disp.plot(ax=ax, xticks_rotation=45) # поворот подписей по X и приятная палитра
plt.tight_layout() # чтобы всё влезло
plt.show()313/313 ━━━━━━━━━━━━━━━━━━━━ 5s 9ms/step
precision recall f1-score support
airplane 0.86 0.88 0.87 986
automobile 0.94 0.93 0.94 971
bird 0.75 0.85 0.80 1043
cat 0.83 0.64 0.72 1037
deer 0.78 0.90 0.83 969
dog 0.78 0.75 0.76 979
frog 0.83 0.92 0.87 1025
horse 0.92 0.83 0.87 948
ship 0.93 0.93 0.93 1003
truck 0.94 0.90 0.92 1039
accuracy 0.85 10000
macro avg 0.86 0.85 0.85 10000
weighted avg 0.86 0.85 0.85 10000
