keras_flowers.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""Train model using transfer learning on InceptionV3."""
from keras.applications.inception_v3 import InceptionV3
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Model
from keras.layers import Dense, GlobalAveragePooling2D

CLASSES = "daisy dandelion roses sunflowers tulips".split()

# create the base pre-trained model
base_model = InceptionV3(weights='imagenet', include_top=False)

# add a global spatial average pooling layer
x = base_model.output
x = GlobalAveragePooling2D()(x)
# let's add a fully-connected layer
x = Dense(1024, activation='relu')(x)
# and a logistic layer -- let's say we have 2 classes
predictions = Dense(len(CLASSES), activation='softmax')(x)

# this is the model we will train
model = Model(input=base_model.input, output=predictions)

# train only the top layer
for layer in model.layers[:-1]:
    layer.trainable = False

# compile the model (should be done *after* setting layers to non-trainable)
model.compile(optimizer='rmsprop', loss='categorical_crossentropy',
              metrics=['accuracy'])

train_datagen = ImageDataGenerator(
    rescale=1. / 255,
    shear_range=0.2,
    horizontal_flip=True,
    rotation_range=10.,
    width_shift_range=0.2,
    height_shift_range=0.2)

test_datagen = ImageDataGenerator(rescale=1. / 255)

train_generator = train_datagen.flow_from_directory(
    './flower_photos/train/',
    target_size=(299, 299),
    batch_size=32,
    classes=CLASSES,
    class_mode='categorical')

validation_generator = test_datagen.flow_from_directory(
    './flower_photos/test/',
    target_size=(299, 299),
    batch_size=32,
    classes=CLASSES,
    class_mode='categorical')

model.fit_generator(train_generator,
                    samples_per_epoch=625,
                    validation_data=validation_generator,
                    nb_val_samples=100,
                    epochs=10)