如何在 Keras、.h5 保存文件中使用训练有素的 model 预测输入图像？

Question

I'm only beginning with Keras and machine learning in general.

I trained a model to classify images from 9 classes and saved it using model.save(). Here is the code I used:

from keras.layers import Input, Lambda, Dense, Flatten
from keras.models import Model
from keras.applications.resnet50 import ResNet50
from keras.preprocessing import image
from keras.preprocessing.image import ImageDataGenerator
from keras.models import Sequential
import numpy as np
from glob import glob
import matplotlib.pyplot as plt

# re-size all the images to this
IMAGE_SIZE = [224, 224]

train_path = 'Datasets/Train'
valid_path = 'Datasets/Test'

# add preprocessing layer to the front of resnet
resnet = ResNet50(input_shape=IMAGE_SIZE + [3], weights='imagenet', include_top=False)

# don't train existing weights
for layer in resnet.layers:
    layer.trainable = False

    # useful for getting number of classes
folders = glob('Datasets/Train/*')

# our layers - you can add more if you want
x = Flatten()(resnet.output)
# x = Dense(1000, activation='relu')(x)
prediction = Dense(len(folders), activation='softmax')(x)

# create a model object
model = Model(inputs=resnet.input, outputs=prediction)

# view the structure of the model
model.summary()

# tell the model what cost and optimization method to use
model.compile(
    loss='categorical_crossentropy',
    optimizer='adam',
    metrics=['accuracy']
)

from keras.preprocessing.image import ImageDataGenerator

train_datagen = ImageDataGenerator(rescale=1. / 255,
                                   shear_range=0.1,
                                   zoom_range=0.1,
                                   horizontal_flip=True)

test_datagen = ImageDataGenerator(rescale=1. / 255)

training_set = train_datagen.flow_from_directory('Datasets/Train',
                                                 target_size=(224, 224),
                                                 batch_size=32,
                                                 class_mode='categorical')

test_set = test_datagen.flow_from_directory('Datasets/Test',
                                            target_size=(224, 224),
                                            batch_size=32,
                                            class_mode='categorical')


# fit the model
r = model.fit_generator(
    training_set,
    validation_data=test_set,
    epochs=3,
    steps_per_epoch=len(training_set),
    validation_steps=len(test_set)
)
def plot_loss_accuracy(r):
    fig = plt.figure(figsize=(12, 6))
    ax = fig.add_subplot(1, 2, 1)
    ax.plot(r.history["loss"], 'r-x', label="Train Loss")
    ax.plot(r.history["val_loss"], 'b-x', label="Validation Loss")
    ax.legend()
    ax.set_title('cross_entropy loss')
    ax.grid(True)

    ax = fig.add_subplot(1, 2, 2)
    ax.plot(r.history["accuracy"], 'r-x', label="Train Accuracy")
    ax.plot(r.history["val_accuracy"], 'b-x', label="Validation Accuracy")
    ax.legend()
    ax.set_title('acuracy')
    ax.grid(True)

It successfully trained. To load and test this model on new images, I used the below code:

from keras.models import load_model
import cv2
import numpy as np

model = load_model('model.h5')

model.compile(loss='categorical_crossentropy',
              optimizer='adam',
              metrics=['accuracy'])

img = cv2.imread('test.jpg')
img = cv2.resize(img,(320,240))
img = np.reshape(img,[1,320,240,3])

classes = model.predict_classes(img)

print(classes)

It outputs:

AttributeError: 'Model' object has no attribute 'predict_classes'

Why wouldn't it even predict?

Thank you,

Answer 1

predict_classes is only available for sequential api http://faroit.com/keras-docs/1.0.0/models/sequential/

So, you first need to get the probabilities and take the max probability as the class.

from keras.models import load_model
import cv2
import numpy as np


class_names = ['a', 'b', 'c', ...] # fill the rest

model = load_model('model.h5')

model.compile(loss='categorical_crossentropy',
              optimizer='adam',
              metrics=['accuracy'])

img = cv2.imread('test.jpg')
img = cv2.resize(img,(320,240))
img = np.reshape(img,[1,320,240,3])

classes = np.argmax(model.predict(img), axis = -1)

print(classes)

names = [class_names[i] for i in classes]

print(names)