Intro to CNNs

Introduction to Deep Learning with Keras

Miguel Esteban

Data Scientist & Founder

# Import Conv2D layer and Flatten from tensorflow keras layers
from tensorflow.keras.layers import Dense, Conv2D, Flatten

# Instantiate your model as usual
model = Sequential()

# Add a convolutional layer with 32 filters of size 3x3
model.add(Conv2D(filters=32,
                 kernel_size=3,
                 input_shape=(28, 28, 1),
                 activation='relu'))

# Add another convolutional layer
model.add(Conv2D(8, kernel_size=3, activation='relu'))

# Flatten the output of the previous layer
model.add(Flatten())

# End this multiclass model with 3 outputs and softmax
model.add(Dense(3, activation='softmax'))

Pre-processing images for ResNet50

# Import image from keras preprocessing
from tensorflow.keras.preprocessing import image

# Import preprocess_input from tensorflow keras applications resnet50
from tensorflow.keras.applications.resnet50 import preprocess_input

# Load the image with the right target size for your model
img = image.load_img(img_path, target_size=(224, 224))

# Turn it into an array
img = image.img_to_array(img)

# Expand the dimensions so that it's understood by our network:
# img.shape turns from (224, 224, 3) into (1, 224, 224, 3)
img = np.expand_dims(img, axis=0)

# Pre-process the img in the same way training images were
img = preprocess_input(img)

Using the ResNet50 model in Keras

# Import ResNet50 and decode_predictions from tensorflow.keras.applications.resnet50
from tensorflow.keras.applications.resnet50 import ResNet50, decode_predictions

# Instantiate a ResNet50 model with imagenet weights
model = ResNet50(weights='imagenet')

# Predict with ResNet50 on our img
preds = model.predict(img)

# Decode predictions and print it
print('Predicted:', decode_predictions(preds, top=1)[0])

Predicted: [('n07697313', 'cheeseburger', 0.9868016)]

Let's experiment!

Introduction to Deep Learning with Keras