Introduction to CNNs

Image Modeling with Keras

Ariel Rokem

Senior Data Scientist, University of Washington

Software and pre-requisites

• DataCamp's Deep Learning course
• Machine learning:
  • Overfitting
  • Model evaluation
  • Cross-validation

Images as data

import matplotlib.pyplot as plt
data = plt.imread('stop_sign.jpg')
plt.imshow(data)
plt.show()

Images as data

data.shape

(2832, 4256, 3)

Images as data

data[1000, 1500]

array([0.73333333, 0.07843137, 0.14509804])

Images as data

data[250, 3500]

array([0.25882353, 0.43921569, 0.77254902])

Modifying image data

data[:, :, 1] = 0
data[:, :, 2] = 0
plt.imshow(data)
plt.show()

Changing an image

data[200:1200, 200:1200, :] = [0, 1, 0]

plt.imshow(data)
plt.show()

Black and white images

Black and white images

Black and white images

tshirt[10:20, 15:25] = 1

plt.imshow(tshirt)
plt.show()

Let's practice!

Image Modeling with Keras