Horse vs Human Dataset - Transfer Learning¶
You can get the dataset from here: https://storage.googleapis.com/tensorflow-1-public/course2/week3/validation-horse-or-human.zip¶
In [1]:
import os
import zipfile
import matplotlib.pyplot as plt
import tensorflow as tf
from tensorflow.keras import layers
from tensorflow.keras import Model
from tensorflow.keras.optimizers import RMSprop
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.preprocessing.image import img_to_array, load_img
Preparing the dataset¶
In [3]:
# Define the training and validation base directories
train_dir = '/tmp/training'
validation_dir = '/tmp/validation'
# Directory with training horse pictures
train_horses_dir = os.path.join(train_dir, 'horses')
# Directory with training humans pictures
train_humans_dir = os.path.join(train_dir, 'humans')
# Directory with validation horse pictures
validation_horses_dir = os.path.join(validation_dir, 'horses')
# Directory with validation human pictures
validation_humans_dir = os.path.join(validation_dir, 'humans')
# Check the number of images for each class and set
print(f"There are {len(os.listdir(train_horses_dir))} images of horses for training.\n")
print(f"There are {len(os.listdir(train_humans_dir))} images of humans for training.\n")
print(f"There are {len(os.listdir(validation_horses_dir))} images of horses for validation.\n")
print(f"There are {len(os.listdir(validation_humans_dir))} images of humans for validation.\n")
There are 500 images of horses for training. There are 527 images of humans for training. There are 128 images of horses for validation. There are 128 images of humans for validation.
Check a sample image of each one of the classes:
In [4]:
print("Sample horse image:")
plt.imshow(load_img(f"{os.path.join(train_horses_dir, os.listdir(train_horses_dir)[0])}"))
plt.show()
print("\nSample human image:")
plt.imshow(load_img(f"{os.path.join(train_humans_dir, os.listdir(train_humans_dir)[0])}"))
plt.show()
Sample horse image:
Sample human image:
matplotlib makes it easy to see that these images have a resolution of 300x300 and are colored:
In [5]:
# Load the first example of a horse
sample_image = load_img(f"{os.path.join(train_horses_dir, os.listdir(train_horses_dir)[0])}")
# Convert the image into its numpy array representation
sample_array = img_to_array(sample_image)
print(f"Each image has shape: {sample_array.shape}")
Each image has shape: (300, 300, 3)
Training and Validation Generators¶
In [6]:
# Instantiate the ImageDataGenerator class
train_datagen = ImageDataGenerator(rescale=1.0/255,
rotation_range=40,
width_shift_range=0.2,
height_shift_range=0.2,
shear_range=0.2,
zoom_range=0.2,
horizontal_flip=True,
fill_mode='nearest')
train_generator = train_datagen.flow_from_directory(directory=TRAINING_DIR,
batch_size=32,
class_mode='binary',
target_size=(150, 150))
# Instantiate the ImageDataGenerator class
validation_datagen = ImageDataGenerator( rescale = 1.0/255. )
validation_generator = validation_datagen.flow_from_directory(directory=VALIDATION_DIR,
batch_size=32,
class_mode='binary',
target_size=(150, 150))
Found 1027 images belonging to 2 classes.
Found 256 images belonging to 2 classes.Transfer learning - Create the pre-trained model¶
Download the inception V3 weights into the /tmp/ directory:
You can get it from here: https://storage.googleapis.com/mledu-datasets/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5
Now load the InceptionV3 model and save the path to the weights you just downloaded:
In [9]:
# Import the inception model
from tensorflow.keras.applications.inception_v3 import InceptionV3
# Create an instance of the inception model from the local pre-trained weights
local_weights_file = '/tmp/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5'
Create a pretrained model:¶
In [10]:
from tensorflow.keras.applications.inception_v3 import InceptionV3
from tensorflow.keras import layers
local_weights_file = '/tmp/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5'
pre_trained_model = InceptionV3(input_shape = (150, 150, 3),
include_top = False,
weights = None)
pre_trained_model.load_weights(local_weights_file)
# Make all the layers in the pre-trained model non-trainable
for layer in pre_trained_model.layers:
layer.trainable = False
Check the model summary:
In [11]:
pre_trained_model.summary()
Model: "inception_v3"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_1 (InputLayer) [(None, 150, 150, 3 0 []
)]
conv2d (Conv2D) (None, 74, 74, 32) 864 ['input_1[0][0]']
batch_normalization (BatchNorm (None, 74, 74, 32) 96 ['conv2d[0][0]']
alization)
activation (Activation) (None, 74, 74, 32) 0 ['batch_normalization[0][0]']
conv2d_1 (Conv2D) (None, 72, 72, 32) 9216 ['activation[0][0]']
batch_normalization_1 (BatchNo (None, 72, 72, 32) 96 ['conv2d_1[0][0]']
rmalization)
activation_1 (Activation) (None, 72, 72, 32) 0 ['batch_normalization_1[0][0]']
conv2d_2 (Conv2D) (None, 72, 72, 64) 18432 ['activation_1[0][0]']
batch_normalization_2 (BatchNo (None, 72, 72, 64) 192 ['conv2d_2[0][0]']
rmalization)
activation_2 (Activation) (None, 72, 72, 64) 0 ['batch_normalization_2[0][0]']
max_pooling2d (MaxPooling2D) (None, 35, 35, 64) 0 ['activation_2[0][0]']
conv2d_3 (Conv2D) (None, 35, 35, 80) 5120 ['max_pooling2d[0][0]']
batch_normalization_3 (BatchNo (None, 35, 35, 80) 240 ['conv2d_3[0][0]']
rmalization)
activation_3 (Activation) (None, 35, 35, 80) 0 ['batch_normalization_3[0][0]']
conv2d_4 (Conv2D) (None, 33, 33, 192) 138240 ['activation_3[0][0]']
batch_normalization_4 (BatchNo (None, 33, 33, 192) 576 ['conv2d_4[0][0]']
rmalization)
activation_4 (Activation) (None, 33, 33, 192) 0 ['batch_normalization_4[0][0]']
max_pooling2d_1 (MaxPooling2D) (None, 16, 16, 192) 0 ['activation_4[0][0]']
conv2d_8 (Conv2D) (None, 16, 16, 64) 12288 ['max_pooling2d_1[0][0]']
batch_normalization_8 (BatchNo (None, 16, 16, 64) 192 ['conv2d_8[0][0]']
rmalization)
activation_8 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_8[0][0]']
conv2d_6 (Conv2D) (None, 16, 16, 48) 9216 ['max_pooling2d_1[0][0]']
conv2d_9 (Conv2D) (None, 16, 16, 96) 55296 ['activation_8[0][0]']
batch_normalization_6 (BatchNo (None, 16, 16, 48) 144 ['conv2d_6[0][0]']
rmalization)
batch_normalization_9 (BatchNo (None, 16, 16, 96) 288 ['conv2d_9[0][0]']
rmalization)
activation_6 (Activation) (None, 16, 16, 48) 0 ['batch_normalization_6[0][0]']
activation_9 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_9[0][0]']
average_pooling2d (AveragePool (None, 16, 16, 192) 0 ['max_pooling2d_1[0][0]']
ing2D)
conv2d_5 (Conv2D) (None, 16, 16, 64) 12288 ['max_pooling2d_1[0][0]']
conv2d_7 (Conv2D) (None, 16, 16, 64) 76800 ['activation_6[0][0]']
conv2d_10 (Conv2D) (None, 16, 16, 96) 82944 ['activation_9[0][0]']
conv2d_11 (Conv2D) (None, 16, 16, 32) 6144 ['average_pooling2d[0][0]']
batch_normalization_5 (BatchNo (None, 16, 16, 64) 192 ['conv2d_5[0][0]']
rmalization)
batch_normalization_7 (BatchNo (None, 16, 16, 64) 192 ['conv2d_7[0][0]']
rmalization)
batch_normalization_10 (BatchN (None, 16, 16, 96) 288 ['conv2d_10[0][0]']
ormalization)
batch_normalization_11 (BatchN (None, 16, 16, 32) 96 ['conv2d_11[0][0]']
ormalization)
activation_5 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_5[0][0]']
activation_7 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_7[0][0]']
activation_10 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_10[0][0]']
activation_11 (Activation) (None, 16, 16, 32) 0 ['batch_normalization_11[0][0]']
mixed0 (Concatenate) (None, 16, 16, 256) 0 ['activation_5[0][0]',
'activation_7[0][0]',
'activation_10[0][0]',
'activation_11[0][0]']
conv2d_15 (Conv2D) (None, 16, 16, 64) 16384 ['mixed0[0][0]']
batch_normalization_15 (BatchN (None, 16, 16, 64) 192 ['conv2d_15[0][0]']
ormalization)
activation_15 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_15[0][0]']
conv2d_13 (Conv2D) (None, 16, 16, 48) 12288 ['mixed0[0][0]']
conv2d_16 (Conv2D) (None, 16, 16, 96) 55296 ['activation_15[0][0]']
batch_normalization_13 (BatchN (None, 16, 16, 48) 144 ['conv2d_13[0][0]']
ormalization)
batch_normalization_16 (BatchN (None, 16, 16, 96) 288 ['conv2d_16[0][0]']
ormalization)
activation_13 (Activation) (None, 16, 16, 48) 0 ['batch_normalization_13[0][0]']
activation_16 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_16[0][0]']
average_pooling2d_1 (AveragePo (None, 16, 16, 256) 0 ['mixed0[0][0]']
oling2D)
conv2d_12 (Conv2D) (None, 16, 16, 64) 16384 ['mixed0[0][0]']
conv2d_14 (Conv2D) (None, 16, 16, 64) 76800 ['activation_13[0][0]']
conv2d_17 (Conv2D) (None, 16, 16, 96) 82944 ['activation_16[0][0]']
conv2d_18 (Conv2D) (None, 16, 16, 64) 16384 ['average_pooling2d_1[0][0]']
batch_normalization_12 (BatchN (None, 16, 16, 64) 192 ['conv2d_12[0][0]']
ormalization)
batch_normalization_14 (BatchN (None, 16, 16, 64) 192 ['conv2d_14[0][0]']
ormalization)
batch_normalization_17 (BatchN (None, 16, 16, 96) 288 ['conv2d_17[0][0]']
ormalization)
batch_normalization_18 (BatchN (None, 16, 16, 64) 192 ['conv2d_18[0][0]']
ormalization)
activation_12 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_12[0][0]']
activation_14 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_14[0][0]']
activation_17 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_17[0][0]']
activation_18 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_18[0][0]']
mixed1 (Concatenate) (None, 16, 16, 288) 0 ['activation_12[0][0]',
'activation_14[0][0]',
'activation_17[0][0]',
'activation_18[0][0]']
conv2d_22 (Conv2D) (None, 16, 16, 64) 18432 ['mixed1[0][0]']
batch_normalization_22 (BatchN (None, 16, 16, 64) 192 ['conv2d_22[0][0]']
ormalization)
activation_22 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_22[0][0]']
conv2d_20 (Conv2D) (None, 16, 16, 48) 13824 ['mixed1[0][0]']
conv2d_23 (Conv2D) (None, 16, 16, 96) 55296 ['activation_22[0][0]']
batch_normalization_20 (BatchN (None, 16, 16, 48) 144 ['conv2d_20[0][0]']
ormalization)
batch_normalization_23 (BatchN (None, 16, 16, 96) 288 ['conv2d_23[0][0]']
ormalization)
activation_20 (Activation) (None, 16, 16, 48) 0 ['batch_normalization_20[0][0]']
activation_23 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_23[0][0]']
average_pooling2d_2 (AveragePo (None, 16, 16, 288) 0 ['mixed1[0][0]']
oling2D)
conv2d_19 (Conv2D) (None, 16, 16, 64) 18432 ['mixed1[0][0]']
conv2d_21 (Conv2D) (None, 16, 16, 64) 76800 ['activation_20[0][0]']
conv2d_24 (Conv2D) (None, 16, 16, 96) 82944 ['activation_23[0][0]']
conv2d_25 (Conv2D) (None, 16, 16, 64) 18432 ['average_pooling2d_2[0][0]']
batch_normalization_19 (BatchN (None, 16, 16, 64) 192 ['conv2d_19[0][0]']
ormalization)
batch_normalization_21 (BatchN (None, 16, 16, 64) 192 ['conv2d_21[0][0]']
ormalization)
batch_normalization_24 (BatchN (None, 16, 16, 96) 288 ['conv2d_24[0][0]']
ormalization)
batch_normalization_25 (BatchN (None, 16, 16, 64) 192 ['conv2d_25[0][0]']
ormalization)
activation_19 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_19[0][0]']
activation_21 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_21[0][0]']
activation_24 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_24[0][0]']
activation_25 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_25[0][0]']
mixed2 (Concatenate) (None, 16, 16, 288) 0 ['activation_19[0][0]',
'activation_21[0][0]',
'activation_24[0][0]',
'activation_25[0][0]']
conv2d_27 (Conv2D) (None, 16, 16, 64) 18432 ['mixed2[0][0]']
batch_normalization_27 (BatchN (None, 16, 16, 64) 192 ['conv2d_27[0][0]']
ormalization)
activation_27 (Activation) (None, 16, 16, 64) 0 ['batch_normalization_27[0][0]']
conv2d_28 (Conv2D) (None, 16, 16, 96) 55296 ['activation_27[0][0]']
batch_normalization_28 (BatchN (None, 16, 16, 96) 288 ['conv2d_28[0][0]']
ormalization)
activation_28 (Activation) (None, 16, 16, 96) 0 ['batch_normalization_28[0][0]']
conv2d_26 (Conv2D) (None, 7, 7, 384) 995328 ['mixed2[0][0]']
conv2d_29 (Conv2D) (None, 7, 7, 96) 82944 ['activation_28[0][0]']
batch_normalization_26 (BatchN (None, 7, 7, 384) 1152 ['conv2d_26[0][0]']
ormalization)
batch_normalization_29 (BatchN (None, 7, 7, 96) 288 ['conv2d_29[0][0]']
ormalization)
activation_26 (Activation) (None, 7, 7, 384) 0 ['batch_normalization_26[0][0]']
activation_29 (Activation) (None, 7, 7, 96) 0 ['batch_normalization_29[0][0]']
max_pooling2d_2 (MaxPooling2D) (None, 7, 7, 288) 0 ['mixed2[0][0]']
mixed3 (Concatenate) (None, 7, 7, 768) 0 ['activation_26[0][0]',
'activation_29[0][0]',
'max_pooling2d_2[0][0]']
conv2d_34 (Conv2D) (None, 7, 7, 128) 98304 ['mixed3[0][0]']
batch_normalization_34 (BatchN (None, 7, 7, 128) 384 ['conv2d_34[0][0]']
ormalization)
activation_34 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_34[0][0]']
conv2d_35 (Conv2D) (None, 7, 7, 128) 114688 ['activation_34[0][0]']
batch_normalization_35 (BatchN (None, 7, 7, 128) 384 ['conv2d_35[0][0]']
ormalization)
activation_35 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_35[0][0]']
conv2d_31 (Conv2D) (None, 7, 7, 128) 98304 ['mixed3[0][0]']
conv2d_36 (Conv2D) (None, 7, 7, 128) 114688 ['activation_35[0][0]']
batch_normalization_31 (BatchN (None, 7, 7, 128) 384 ['conv2d_31[0][0]']
ormalization)
batch_normalization_36 (BatchN (None, 7, 7, 128) 384 ['conv2d_36[0][0]']
ormalization)
activation_31 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_31[0][0]']
activation_36 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_36[0][0]']
conv2d_32 (Conv2D) (None, 7, 7, 128) 114688 ['activation_31[0][0]']
conv2d_37 (Conv2D) (None, 7, 7, 128) 114688 ['activation_36[0][0]']
batch_normalization_32 (BatchN (None, 7, 7, 128) 384 ['conv2d_32[0][0]']
ormalization)
batch_normalization_37 (BatchN (None, 7, 7, 128) 384 ['conv2d_37[0][0]']
ormalization)
activation_32 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_32[0][0]']
activation_37 (Activation) (None, 7, 7, 128) 0 ['batch_normalization_37[0][0]']
average_pooling2d_3 (AveragePo (None, 7, 7, 768) 0 ['mixed3[0][0]']
oling2D)
conv2d_30 (Conv2D) (None, 7, 7, 192) 147456 ['mixed3[0][0]']
conv2d_33 (Conv2D) (None, 7, 7, 192) 172032 ['activation_32[0][0]']
conv2d_38 (Conv2D) (None, 7, 7, 192) 172032 ['activation_37[0][0]']
conv2d_39 (Conv2D) (None, 7, 7, 192) 147456 ['average_pooling2d_3[0][0]']
batch_normalization_30 (BatchN (None, 7, 7, 192) 576 ['conv2d_30[0][0]']
ormalization)
batch_normalization_33 (BatchN (None, 7, 7, 192) 576 ['conv2d_33[0][0]']
ormalization)
batch_normalization_38 (BatchN (None, 7, 7, 192) 576 ['conv2d_38[0][0]']
ormalization)
batch_normalization_39 (BatchN (None, 7, 7, 192) 576 ['conv2d_39[0][0]']
ormalization)
activation_30 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_30[0][0]']
activation_33 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_33[0][0]']
activation_38 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_38[0][0]']
activation_39 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_39[0][0]']
mixed4 (Concatenate) (None, 7, 7, 768) 0 ['activation_30[0][0]',
'activation_33[0][0]',
'activation_38[0][0]',
'activation_39[0][0]']
conv2d_44 (Conv2D) (None, 7, 7, 160) 122880 ['mixed4[0][0]']
batch_normalization_44 (BatchN (None, 7, 7, 160) 480 ['conv2d_44[0][0]']
ormalization)
activation_44 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_44[0][0]']
conv2d_45 (Conv2D) (None, 7, 7, 160) 179200 ['activation_44[0][0]']
batch_normalization_45 (BatchN (None, 7, 7, 160) 480 ['conv2d_45[0][0]']
ormalization)
activation_45 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_45[0][0]']
conv2d_41 (Conv2D) (None, 7, 7, 160) 122880 ['mixed4[0][0]']
conv2d_46 (Conv2D) (None, 7, 7, 160) 179200 ['activation_45[0][0]']
batch_normalization_41 (BatchN (None, 7, 7, 160) 480 ['conv2d_41[0][0]']
ormalization)
batch_normalization_46 (BatchN (None, 7, 7, 160) 480 ['conv2d_46[0][0]']
ormalization)
activation_41 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_41[0][0]']
activation_46 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_46[0][0]']
conv2d_42 (Conv2D) (None, 7, 7, 160) 179200 ['activation_41[0][0]']
conv2d_47 (Conv2D) (None, 7, 7, 160) 179200 ['activation_46[0][0]']
batch_normalization_42 (BatchN (None, 7, 7, 160) 480 ['conv2d_42[0][0]']
ormalization)
batch_normalization_47 (BatchN (None, 7, 7, 160) 480 ['conv2d_47[0][0]']
ormalization)
activation_42 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_42[0][0]']
activation_47 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_47[0][0]']
average_pooling2d_4 (AveragePo (None, 7, 7, 768) 0 ['mixed4[0][0]']
oling2D)
conv2d_40 (Conv2D) (None, 7, 7, 192) 147456 ['mixed4[0][0]']
conv2d_43 (Conv2D) (None, 7, 7, 192) 215040 ['activation_42[0][0]']
conv2d_48 (Conv2D) (None, 7, 7, 192) 215040 ['activation_47[0][0]']
conv2d_49 (Conv2D) (None, 7, 7, 192) 147456 ['average_pooling2d_4[0][0]']
batch_normalization_40 (BatchN (None, 7, 7, 192) 576 ['conv2d_40[0][0]']
ormalization)
batch_normalization_43 (BatchN (None, 7, 7, 192) 576 ['conv2d_43[0][0]']
ormalization)
batch_normalization_48 (BatchN (None, 7, 7, 192) 576 ['conv2d_48[0][0]']
ormalization)
batch_normalization_49 (BatchN (None, 7, 7, 192) 576 ['conv2d_49[0][0]']
ormalization)
activation_40 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_40[0][0]']
activation_43 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_43[0][0]']
activation_48 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_48[0][0]']
activation_49 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_49[0][0]']
mixed5 (Concatenate) (None, 7, 7, 768) 0 ['activation_40[0][0]',
'activation_43[0][0]',
'activation_48[0][0]',
'activation_49[0][0]']
conv2d_54 (Conv2D) (None, 7, 7, 160) 122880 ['mixed5[0][0]']
batch_normalization_54 (BatchN (None, 7, 7, 160) 480 ['conv2d_54[0][0]']
ormalization)
activation_54 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_54[0][0]']
conv2d_55 (Conv2D) (None, 7, 7, 160) 179200 ['activation_54[0][0]']
batch_normalization_55 (BatchN (None, 7, 7, 160) 480 ['conv2d_55[0][0]']
ormalization)
activation_55 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_55[0][0]']
conv2d_51 (Conv2D) (None, 7, 7, 160) 122880 ['mixed5[0][0]']
conv2d_56 (Conv2D) (None, 7, 7, 160) 179200 ['activation_55[0][0]']
batch_normalization_51 (BatchN (None, 7, 7, 160) 480 ['conv2d_51[0][0]']
ormalization)
batch_normalization_56 (BatchN (None, 7, 7, 160) 480 ['conv2d_56[0][0]']
ormalization)
activation_51 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_51[0][0]']
activation_56 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_56[0][0]']
conv2d_52 (Conv2D) (None, 7, 7, 160) 179200 ['activation_51[0][0]']
conv2d_57 (Conv2D) (None, 7, 7, 160) 179200 ['activation_56[0][0]']
batch_normalization_52 (BatchN (None, 7, 7, 160) 480 ['conv2d_52[0][0]']
ormalization)
batch_normalization_57 (BatchN (None, 7, 7, 160) 480 ['conv2d_57[0][0]']
ormalization)
activation_52 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_52[0][0]']
activation_57 (Activation) (None, 7, 7, 160) 0 ['batch_normalization_57[0][0]']
average_pooling2d_5 (AveragePo (None, 7, 7, 768) 0 ['mixed5[0][0]']
oling2D)
conv2d_50 (Conv2D) (None, 7, 7, 192) 147456 ['mixed5[0][0]']
conv2d_53 (Conv2D) (None, 7, 7, 192) 215040 ['activation_52[0][0]']
conv2d_58 (Conv2D) (None, 7, 7, 192) 215040 ['activation_57[0][0]']
conv2d_59 (Conv2D) (None, 7, 7, 192) 147456 ['average_pooling2d_5[0][0]']
batch_normalization_50 (BatchN (None, 7, 7, 192) 576 ['conv2d_50[0][0]']
ormalization)
batch_normalization_53 (BatchN (None, 7, 7, 192) 576 ['conv2d_53[0][0]']
ormalization)
batch_normalization_58 (BatchN (None, 7, 7, 192) 576 ['conv2d_58[0][0]']
ormalization)
batch_normalization_59 (BatchN (None, 7, 7, 192) 576 ['conv2d_59[0][0]']
ormalization)
activation_50 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_50[0][0]']
activation_53 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_53[0][0]']
activation_58 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_58[0][0]']
activation_59 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_59[0][0]']
mixed6 (Concatenate) (None, 7, 7, 768) 0 ['activation_50[0][0]',
'activation_53[0][0]',
'activation_58[0][0]',
'activation_59[0][0]']
conv2d_64 (Conv2D) (None, 7, 7, 192) 147456 ['mixed6[0][0]']
batch_normalization_64 (BatchN (None, 7, 7, 192) 576 ['conv2d_64[0][0]']
ormalization)
activation_64 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_64[0][0]']
conv2d_65 (Conv2D) (None, 7, 7, 192) 258048 ['activation_64[0][0]']
batch_normalization_65 (BatchN (None, 7, 7, 192) 576 ['conv2d_65[0][0]']
ormalization)
activation_65 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_65[0][0]']
conv2d_61 (Conv2D) (None, 7, 7, 192) 147456 ['mixed6[0][0]']
conv2d_66 (Conv2D) (None, 7, 7, 192) 258048 ['activation_65[0][0]']
batch_normalization_61 (BatchN (None, 7, 7, 192) 576 ['conv2d_61[0][0]']
ormalization)
batch_normalization_66 (BatchN (None, 7, 7, 192) 576 ['conv2d_66[0][0]']
ormalization)
activation_61 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_61[0][0]']
activation_66 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_66[0][0]']
conv2d_62 (Conv2D) (None, 7, 7, 192) 258048 ['activation_61[0][0]']
conv2d_67 (Conv2D) (None, 7, 7, 192) 258048 ['activation_66[0][0]']
batch_normalization_62 (BatchN (None, 7, 7, 192) 576 ['conv2d_62[0][0]']
ormalization)
batch_normalization_67 (BatchN (None, 7, 7, 192) 576 ['conv2d_67[0][0]']
ormalization)
activation_62 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_62[0][0]']
activation_67 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_67[0][0]']
average_pooling2d_6 (AveragePo (None, 7, 7, 768) 0 ['mixed6[0][0]']
oling2D)
conv2d_60 (Conv2D) (None, 7, 7, 192) 147456 ['mixed6[0][0]']
conv2d_63 (Conv2D) (None, 7, 7, 192) 258048 ['activation_62[0][0]']
conv2d_68 (Conv2D) (None, 7, 7, 192) 258048 ['activation_67[0][0]']
conv2d_69 (Conv2D) (None, 7, 7, 192) 147456 ['average_pooling2d_6[0][0]']
batch_normalization_60 (BatchN (None, 7, 7, 192) 576 ['conv2d_60[0][0]']
ormalization)
batch_normalization_63 (BatchN (None, 7, 7, 192) 576 ['conv2d_63[0][0]']
ormalization)
batch_normalization_68 (BatchN (None, 7, 7, 192) 576 ['conv2d_68[0][0]']
ormalization)
batch_normalization_69 (BatchN (None, 7, 7, 192) 576 ['conv2d_69[0][0]']
ormalization)
activation_60 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_60[0][0]']
activation_63 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_63[0][0]']
activation_68 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_68[0][0]']
activation_69 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_69[0][0]']
mixed7 (Concatenate) (None, 7, 7, 768) 0 ['activation_60[0][0]',
'activation_63[0][0]',
'activation_68[0][0]',
'activation_69[0][0]']
conv2d_72 (Conv2D) (None, 7, 7, 192) 147456 ['mixed7[0][0]']
batch_normalization_72 (BatchN (None, 7, 7, 192) 576 ['conv2d_72[0][0]']
ormalization)
activation_72 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_72[0][0]']
conv2d_73 (Conv2D) (None, 7, 7, 192) 258048 ['activation_72[0][0]']
batch_normalization_73 (BatchN (None, 7, 7, 192) 576 ['conv2d_73[0][0]']
ormalization)
activation_73 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_73[0][0]']
conv2d_70 (Conv2D) (None, 7, 7, 192) 147456 ['mixed7[0][0]']
conv2d_74 (Conv2D) (None, 7, 7, 192) 258048 ['activation_73[0][0]']
batch_normalization_70 (BatchN (None, 7, 7, 192) 576 ['conv2d_70[0][0]']
ormalization)
batch_normalization_74 (BatchN (None, 7, 7, 192) 576 ['conv2d_74[0][0]']
ormalization)
activation_70 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_70[0][0]']
activation_74 (Activation) (None, 7, 7, 192) 0 ['batch_normalization_74[0][0]']
conv2d_71 (Conv2D) (None, 3, 3, 320) 552960 ['activation_70[0][0]']
conv2d_75 (Conv2D) (None, 3, 3, 192) 331776 ['activation_74[0][0]']
batch_normalization_71 (BatchN (None, 3, 3, 320) 960 ['conv2d_71[0][0]']
ormalization)
batch_normalization_75 (BatchN (None, 3, 3, 192) 576 ['conv2d_75[0][0]']
ormalization)
activation_71 (Activation) (None, 3, 3, 320) 0 ['batch_normalization_71[0][0]']
activation_75 (Activation) (None, 3, 3, 192) 0 ['batch_normalization_75[0][0]']
max_pooling2d_3 (MaxPooling2D) (None, 3, 3, 768) 0 ['mixed7[0][0]']
mixed8 (Concatenate) (None, 3, 3, 1280) 0 ['activation_71[0][0]',
'activation_75[0][0]',
'max_pooling2d_3[0][0]']
conv2d_80 (Conv2D) (None, 3, 3, 448) 573440 ['mixed8[0][0]']
batch_normalization_80 (BatchN (None, 3, 3, 448) 1344 ['conv2d_80[0][0]']
ormalization)
activation_80 (Activation) (None, 3, 3, 448) 0 ['batch_normalization_80[0][0]']
conv2d_77 (Conv2D) (None, 3, 3, 384) 491520 ['mixed8[0][0]']
conv2d_81 (Conv2D) (None, 3, 3, 384) 1548288 ['activation_80[0][0]']
batch_normalization_77 (BatchN (None, 3, 3, 384) 1152 ['conv2d_77[0][0]']
ormalization)
batch_normalization_81 (BatchN (None, 3, 3, 384) 1152 ['conv2d_81[0][0]']
ormalization)
activation_77 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_77[0][0]']
activation_81 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_81[0][0]']
conv2d_78 (Conv2D) (None, 3, 3, 384) 442368 ['activation_77[0][0]']
conv2d_79 (Conv2D) (None, 3, 3, 384) 442368 ['activation_77[0][0]']
conv2d_82 (Conv2D) (None, 3, 3, 384) 442368 ['activation_81[0][0]']
conv2d_83 (Conv2D) (None, 3, 3, 384) 442368 ['activation_81[0][0]']
average_pooling2d_7 (AveragePo (None, 3, 3, 1280) 0 ['mixed8[0][0]']
oling2D)
conv2d_76 (Conv2D) (None, 3, 3, 320) 409600 ['mixed8[0][0]']
batch_normalization_78 (BatchN (None, 3, 3, 384) 1152 ['conv2d_78[0][0]']
ormalization)
batch_normalization_79 (BatchN (None, 3, 3, 384) 1152 ['conv2d_79[0][0]']
ormalization)
batch_normalization_82 (BatchN (None, 3, 3, 384) 1152 ['conv2d_82[0][0]']
ormalization)
batch_normalization_83 (BatchN (None, 3, 3, 384) 1152 ['conv2d_83[0][0]']
ormalization)
conv2d_84 (Conv2D) (None, 3, 3, 192) 245760 ['average_pooling2d_7[0][0]']
batch_normalization_76 (BatchN (None, 3, 3, 320) 960 ['conv2d_76[0][0]']
ormalization)
activation_78 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_78[0][0]']
activation_79 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_79[0][0]']
activation_82 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_82[0][0]']
activation_83 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_83[0][0]']
batch_normalization_84 (BatchN (None, 3, 3, 192) 576 ['conv2d_84[0][0]']
ormalization)
activation_76 (Activation) (None, 3, 3, 320) 0 ['batch_normalization_76[0][0]']
mixed9_0 (Concatenate) (None, 3, 3, 768) 0 ['activation_78[0][0]',
'activation_79[0][0]']
concatenate (Concatenate) (None, 3, 3, 768) 0 ['activation_82[0][0]',
'activation_83[0][0]']
activation_84 (Activation) (None, 3, 3, 192) 0 ['batch_normalization_84[0][0]']
mixed9 (Concatenate) (None, 3, 3, 2048) 0 ['activation_76[0][0]',
'mixed9_0[0][0]',
'concatenate[0][0]',
'activation_84[0][0]']
conv2d_89 (Conv2D) (None, 3, 3, 448) 917504 ['mixed9[0][0]']
batch_normalization_89 (BatchN (None, 3, 3, 448) 1344 ['conv2d_89[0][0]']
ormalization)
activation_89 (Activation) (None, 3, 3, 448) 0 ['batch_normalization_89[0][0]']
conv2d_86 (Conv2D) (None, 3, 3, 384) 786432 ['mixed9[0][0]']
conv2d_90 (Conv2D) (None, 3, 3, 384) 1548288 ['activation_89[0][0]']
batch_normalization_86 (BatchN (None, 3, 3, 384) 1152 ['conv2d_86[0][0]']
ormalization)
batch_normalization_90 (BatchN (None, 3, 3, 384) 1152 ['conv2d_90[0][0]']
ormalization)
activation_86 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_86[0][0]']
activation_90 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_90[0][0]']
conv2d_87 (Conv2D) (None, 3, 3, 384) 442368 ['activation_86[0][0]']
conv2d_88 (Conv2D) (None, 3, 3, 384) 442368 ['activation_86[0][0]']
conv2d_91 (Conv2D) (None, 3, 3, 384) 442368 ['activation_90[0][0]']
conv2d_92 (Conv2D) (None, 3, 3, 384) 442368 ['activation_90[0][0]']
average_pooling2d_8 (AveragePo (None, 3, 3, 2048) 0 ['mixed9[0][0]']
oling2D)
conv2d_85 (Conv2D) (None, 3, 3, 320) 655360 ['mixed9[0][0]']
batch_normalization_87 (BatchN (None, 3, 3, 384) 1152 ['conv2d_87[0][0]']
ormalization)
batch_normalization_88 (BatchN (None, 3, 3, 384) 1152 ['conv2d_88[0][0]']
ormalization)
batch_normalization_91 (BatchN (None, 3, 3, 384) 1152 ['conv2d_91[0][0]']
ormalization)
batch_normalization_92 (BatchN (None, 3, 3, 384) 1152 ['conv2d_92[0][0]']
ormalization)
conv2d_93 (Conv2D) (None, 3, 3, 192) 393216 ['average_pooling2d_8[0][0]']
batch_normalization_85 (BatchN (None, 3, 3, 320) 960 ['conv2d_85[0][0]']
ormalization)
activation_87 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_87[0][0]']
activation_88 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_88[0][0]']
activation_91 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_91[0][0]']
activation_92 (Activation) (None, 3, 3, 384) 0 ['batch_normalization_92[0][0]']
batch_normalization_93 (BatchN (None, 3, 3, 192) 576 ['conv2d_93[0][0]']
ormalization)
activation_85 (Activation) (None, 3, 3, 320) 0 ['batch_normalization_85[0][0]']
mixed9_1 (Concatenate) (None, 3, 3, 768) 0 ['activation_87[0][0]',
'activation_88[0][0]']
concatenate_1 (Concatenate) (None, 3, 3, 768) 0 ['activation_91[0][0]',
'activation_92[0][0]']
activation_93 (Activation) (None, 3, 3, 192) 0 ['batch_normalization_93[0][0]']
mixed10 (Concatenate) (None, 3, 3, 2048) 0 ['activation_85[0][0]',
'mixed9_1[0][0]',
'concatenate_1[0][0]',
'activation_93[0][0]']
==================================================================================================
Total params: 21,802,784
Trainable params: 0
Non-trainable params: 21,802,784
__________________________________________________________________________________________________
Check that all the layers in the model were set to be non-trainable:
In [12]:
total_params = pre_trained_model.count_params()
num_trainable_params = sum([w.shape.num_elements() for w in pre_trained_model.trainable_weights])
print(f"There are {total_params:,} total parameters in this model.")
print(f"There are {num_trainable_params:,} trainable parameters in this model.")
There are 21,802,784 total parameters in this model. There are 0 trainable parameters in this model.
Creating callbacks for later¶
The callback to stop training once an accuracy of 99.9% is reached:
In [13]:
class myCallback(tf.keras.callbacks.Callback):
def on_epoch_end(self, epoch, logs={}):
if(logs.get('accuracy')>0.999):
print("\nReached 99.9% accuracy so cancelling training!")
self.model.stop_training = True
Pipelining the pre-trained model with our own:¶
First, let's get the last desired layer (last_output):
In [16]:
last_desired_layer = pre_trained_model.get_layer('mixed7')
print('last layer output shape: ', last_desired_layer.output_shape)
last_output = last_desired_layer.output
print('last layer output: ', last_output)
Check that everything works as expected:
last layer output shape: (None, 7, 7, 768)
last layer output: KerasTensor(type_spec=TensorSpec(shape=(None, 7, 7, 768), dtype=tf.float32, name=None), name='mixed7/concat:0', description="created by layer 'mixed7'")Now, we will create the final model by adding some additional layers on top of the pre-trained model.
In [18]:
# Print the type of the pre-trained model
print(f"The pretrained model has type: {type(pre_trained_model)}")
The pretrained model has type: <class 'keras.engine.functional.Functional'>
To create the final model, you will use Keras' Model class by defining the appropriate inputs and outputs.
Note that you can get the input from any existing model by using its input attribute and by using the Funcional API you can use the last layer directly as output when creating the final model.
In [19]:
from tensorflow.keras.optimizers import RMSprop
# Flatten the output layer to 1 dimension
x = layers.Flatten()(last_output)
# Add a fully connected layer with 1,024 hidden units and ReLU activation
x = layers.Dense(1024, activation='relu')(x)
# Add a dropout rate of 0.2
x = layers.Dropout(0.2)(x)
# Add a final sigmoid layer for classification
x = layers.Dense (1, activation='sigmoid')(x)
# Create the complete model by using the Model class
model = Model(pre_trained_model.input, x)
# Compile the model
model.compile(optimizer = RMSprop(learning_rate=0.0001),
loss = 'binary_crossentropy',
metrics = ['accuracy'])
In [20]:
# Inspect parameters
total_params = model.count_params()
num_trainable_params = sum([w.shape.num_elements() for w in model.trainable_weights])
print(f"There are {total_params:,} total parameters in this model.")
print(f"There are {num_trainable_params:,} trainable parameters in this model.")
There are 47,512,481 total parameters in this model. There are 38,537,217 trainable parameters in this model.
Let's now train the model¶
In [21]:
callbacks = myCallback()
history = model.fit(train_generator,
validation_data = validation_generator,
epochs = 100,
verbose = 2,
callbacks=callbacks)
Epoch 1/100 33/33 - 26s - loss: 0.1755 - accuracy: 0.9357 - val_loss: 0.0277 - val_accuracy: 0.9805 - 26s/epoch - 802ms/step Epoch 2/100 33/33 - 11s - loss: 0.0311 - accuracy: 0.9893 - val_loss: 0.0382 - val_accuracy: 0.9766 - 11s/epoch - 333ms/step Epoch 3/100 33/33 - 11s - loss: 0.0270 - accuracy: 0.9903 - val_loss: 0.0203 - val_accuracy: 0.9922 - 11s/epoch - 332ms/step Epoch 4/100 33/33 - 11s - loss: 0.0210 - accuracy: 0.9942 - val_loss: 0.0246 - val_accuracy: 0.9922 - 11s/epoch - 335ms/step Epoch 5/100 33/33 - 12s - loss: 0.0201 - accuracy: 0.9903 - val_loss: 0.0026 - val_accuracy: 1.0000 - 12s/epoch - 359ms/step Epoch 6/100 33/33 - 11s - loss: 0.0148 - accuracy: 0.9951 - val_loss: 0.0078 - val_accuracy: 0.9961 - 11s/epoch - 334ms/step Epoch 7/100 33/33 - 11s - loss: 0.0094 - accuracy: 0.9971 - val_loss: 0.0162 - val_accuracy: 0.9922 - 11s/epoch - 336ms/step Epoch 8/100 33/33 - 11s - loss: 0.0246 - accuracy: 0.9903 - val_loss: 0.0067 - val_accuracy: 0.9961 - 11s/epoch - 335ms/step Epoch 9/100 Reached 99.9% accuracy so cancelling training! 33/33 - 11s - loss: 0.0028 - accuracy: 0.9990 - val_loss: 0.0248 - val_accuracy: 0.9922 - 11s/epoch - 337ms/step
Now let's take a quick look at the training and validation accuracies for each epoch of training:
In [22]:
# Plot the training and validation accuracies for each epoch
acc = history.history['accuracy']
val_acc = history.history['val_accuracy']
loss = history.history['loss']
val_loss = history.history['val_loss']
epochs = range(len(acc))
plt.plot(epochs, acc, 'r', label='Training accuracy')
plt.plot(epochs, val_acc, 'b', label='Validation accuracy')
plt.title('Training and validation accuracy')
plt.legend(loc=0)
plt.figure()
plt.show()
<Figure size 432x288 with 0 Axes>