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.
321 KiB
321 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.models import Sequential
from tensorflow.keras import layers
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).
# загрузка датасета
from keras.datasets import cifar10
(X_train, y_train), (X_test, y_test) = cifar10.load_data()# 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 = 50000,
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: (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))
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_4"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ┃ Layer (type) ┃ Output Shape ┃ Param # ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ │ conv2d_12 (Conv2D) │ (None, 32, 32, 32) │ 896 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization_6 │ (None, 32, 32, 32) │ 128 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_13 (Conv2D) │ (None, 32, 32, 32) │ 9,248 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization_7 │ (None, 32, 32, 32) │ 128 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ max_pooling2d_9 (MaxPooling2D) │ (None, 16, 16, 32) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dropout_6 (Dropout) │ (None, 16, 16, 32) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_14 (Conv2D) │ (None, 16, 16, 64) │ 18,496 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization_8 │ (None, 16, 16, 64) │ 256 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ conv2d_15 (Conv2D) │ (None, 16, 16, 64) │ 36,928 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ batch_normalization_9 │ (None, 16, 16, 64) │ 256 │ │ (BatchNormalization) │ │ │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ max_pooling2d_10 (MaxPooling2D) │ (None, 8, 8, 64) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dropout_7 (Dropout) │ (None, 8, 8, 64) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ flatten_3 (Flatten) │ (None, 4096) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dense_6 (Dense) │ (None, 128) │ 524,416 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dropout_8 (Dropout) │ (None, 128) │ 0 │ ├─────────────────────────────────┼────────────────────────┼───────────────┤ │ dense_7 (Dense) │ (None, 10) │ 1,290 │ └─────────────────────────────────┴────────────────────────┴───────────────┘
Total params: 592,042 (2.26 MB)
Trainable params: 591,658 (2.26 MB)
Non-trainable params: 384 (1.50 KB)
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 ━━━━━━━━━━━━━━━━━━━━ 22s 18ms/step - accuracy: 0.2869 - loss: 2.0611 - val_accuracy: 0.4912 - val_loss: 1.3440
Epoch 2/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.4823 - loss: 1.4245 - val_accuracy: 0.5306 - val_loss: 1.3164
Epoch 3/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.5579 - loss: 1.2440 - val_accuracy: 0.6416 - val_loss: 0.9811
Epoch 4/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.6049 - loss: 1.1170 - val_accuracy: 0.6588 - val_loss: 0.9544
Epoch 5/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.6420 - loss: 1.0236 - val_accuracy: 0.7096 - val_loss: 0.8580
Epoch 6/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.6799 - loss: 0.9205 - val_accuracy: 0.6866 - val_loss: 0.9137
Epoch 7/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.6960 - loss: 0.8744 - val_accuracy: 0.7164 - val_loss: 0.7933
Epoch 8/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.7182 - loss: 0.8150 - val_accuracy: 0.7492 - val_loss: 0.7311
Epoch 9/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.7266 - loss: 0.7832 - val_accuracy: 0.7552 - val_loss: 0.7084
Epoch 10/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.7434 - loss: 0.7428 - val_accuracy: 0.7456 - val_loss: 0.7759
Epoch 11/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.7521 - loss: 0.7132 - val_accuracy: 0.7662 - val_loss: 0.6787
Epoch 12/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.7676 - loss: 0.6744 - val_accuracy: 0.7694 - val_loss: 0.6662
Epoch 13/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.7746 - loss: 0.6513 - val_accuracy: 0.7818 - val_loss: 0.6265
Epoch 14/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.7857 - loss: 0.6170 - val_accuracy: 0.7862 - val_loss: 0.6263
Epoch 15/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.7898 - loss: 0.6089 - val_accuracy: 0.7670 - val_loss: 0.6986
Epoch 16/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.7923 - loss: 0.5927 - val_accuracy: 0.7836 - val_loss: 0.6244
Epoch 17/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8020 - loss: 0.5676 - val_accuracy: 0.7718 - val_loss: 0.6886
Epoch 18/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8127 - loss: 0.5410 - val_accuracy: 0.7874 - val_loss: 0.6365
Epoch 19/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8199 - loss: 0.5235 - val_accuracy: 0.7916 - val_loss: 0.6148
Epoch 20/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8194 - loss: 0.5110 - val_accuracy: 0.8006 - val_loss: 0.6143
Epoch 21/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.8252 - loss: 0.5031 - val_accuracy: 0.7932 - val_loss: 0.6536
Epoch 22/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8332 - loss: 0.4793 - val_accuracy: 0.8000 - val_loss: 0.6312
Epoch 23/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8351 - loss: 0.4699 - val_accuracy: 0.7950 - val_loss: 0.6454
Epoch 24/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8417 - loss: 0.4573 - val_accuracy: 0.7992 - val_loss: 0.6198
Epoch 25/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8458 - loss: 0.4401 - val_accuracy: 0.7964 - val_loss: 0.6369
Epoch 26/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8467 - loss: 0.4380 - val_accuracy: 0.7906 - val_loss: 0.6787
Epoch 27/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 11s 8ms/step - accuracy: 0.8537 - loss: 0.4269 - val_accuracy: 0.7964 - val_loss: 0.6376
Epoch 28/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8533 - loss: 0.4183 - val_accuracy: 0.8022 - val_loss: 0.6392
Epoch 29/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8561 - loss: 0.4099 - val_accuracy: 0.8194 - val_loss: 0.5746
Epoch 30/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8590 - loss: 0.4008 - val_accuracy: 0.8114 - val_loss: 0.5972
Epoch 31/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.8674 - loss: 0.3831 - val_accuracy: 0.8142 - val_loss: 0.6090
Epoch 32/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.8669 - loss: 0.3852 - val_accuracy: 0.8098 - val_loss: 0.6277
Epoch 33/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8669 - loss: 0.3798 - val_accuracy: 0.8112 - val_loss: 0.5966
Epoch 34/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8685 - loss: 0.3794 - val_accuracy: 0.8194 - val_loss: 0.5824
Epoch 35/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8719 - loss: 0.3688 - val_accuracy: 0.8162 - val_loss: 0.6381
Epoch 36/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8718 - loss: 0.3660 - val_accuracy: 0.8168 - val_loss: 0.5981
Epoch 37/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8785 - loss: 0.3505 - val_accuracy: 0.8054 - val_loss: 0.6218
Epoch 38/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.8812 - loss: 0.3460 - val_accuracy: 0.8158 - val_loss: 0.6058
Epoch 39/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8805 - loss: 0.3441 - val_accuracy: 0.8184 - val_loss: 0.5941
Epoch 40/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8817 - loss: 0.3422 - val_accuracy: 0.8146 - val_loss: 0.6009
Epoch 41/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8766 - loss: 0.3513 - val_accuracy: 0.8166 - val_loss: 0.5945
Epoch 42/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8840 - loss: 0.3354 - val_accuracy: 0.8178 - val_loss: 0.6113
Epoch 43/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8872 - loss: 0.3287 - val_accuracy: 0.8278 - val_loss: 0.5791
Epoch 44/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 8ms/step - accuracy: 0.8890 - loss: 0.3196 - val_accuracy: 0.8126 - val_loss: 0.6564
Epoch 45/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8871 - loss: 0.3257 - val_accuracy: 0.8252 - val_loss: 0.5983
Epoch 46/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8906 - loss: 0.3150 - val_accuracy: 0.8122 - val_loss: 0.6215
Epoch 47/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8908 - loss: 0.3083 - val_accuracy: 0.8226 - val_loss: 0.5767
Epoch 48/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 6s 8ms/step - accuracy: 0.8921 - loss: 0.3090 - val_accuracy: 0.8194 - val_loss: 0.6160
Epoch 49/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8921 - loss: 0.3063 - val_accuracy: 0.8238 - val_loss: 0.6257
Epoch 50/50
704/704 ━━━━━━━━━━━━━━━━━━━━ 5s 7ms/step - accuracy: 0.8962 - loss: 0.3017 - val_accuracy: 0.8230 - val_loss: 0.6093
<keras.src.callbacks.history.History at 0x786494d78620>scores = model.evaluate(X_test, y_test)
print('Loss on test data:', scores[0])
print('Accuracy on test data:', scores[1])313/313 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step - accuracy: 0.8182 - loss: 0.6295
Loss on test data: 0.635892927646637
Accuracy on test data: 0.8194000124931335
for n in [5,17]:
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 72ms/step
NN output: [[0.19707698 0.12492625 0.35796925 0.08262431 0.0519276 0.01125633
0.04046627 0.03933177 0.03143692 0.0629843 ]]
Real mark: 0
NN answer: 2
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 46ms/step
NN output: [[6.0066624e-05 5.0396560e-05 4.3396626e-02 7.0300132e-02 1.8314796e-03
8.7599868e-01 8.3906803e-04 7.5012995e-03 1.6654752e-05 5.6444887e-06]]
Real mark: 5
NN answer: 5
# 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 ━━━━━━━━━━━━━━━━━━━━ 2s 5ms/step
precision recall f1-score support
airplane 0.85 0.86 0.86 1013
automobile 0.93 0.91 0.92 989
bird 0.75 0.75 0.75 1018
cat 0.69 0.66 0.67 1049
deer 0.79 0.78 0.78 1009
dog 0.73 0.68 0.71 978
frog 0.79 0.90 0.84 981
horse 0.88 0.84 0.86 986
ship 0.89 0.92 0.91 1029
truck 0.88 0.91 0.89 948
accuracy 0.82 10000
macro avg 0.82 0.82 0.82 10000
weighted avg 0.82 0.82 0.82 10000
