Multiclass Classifications using CNN and Tensorflow
We will identify whether an image is Rock, Paper or Scissors.
The dataset
Rock Paper Scissors is a dataset containing about 3,000 computer-generated images from a variety of different hands, from different races, ages and genders, posed into Rock, Paper or Scissors and labelled as such. Each image is 300 X 300 pixels in 24-bit color. The images have all been generated using CGI techniques as an experiment in determining if a CGI-based dataset can be used for classification against real images. You can download the dataset here.
In this blog post, we will identify whether an image is Rock, Paper or Scissors.
# download the training and test set zip files
!wget --no-check-certificate \
https://storage.googleapis.com/laurencemoroney-blog.appspot.com/rps.zip \
-O /tmp/rps.zip
!wget --no-check-certificate \
https://storage.googleapis.com/laurencemoroney-blog.appspot.com/rps-test-set.zip \
-O /tmp/rps-test-set.zip
import os
import zipfile
# unzip the data into the tmp directory
local_zip = '/tmp/rps.zip'
zip_ref = zipfile.ZipFile(local_zip, 'r')
zip_ref.extractall('/tmp/')
zip_ref.close()
local_zip = '/tmp/rps-test-set.zip'
zip_ref = zipfile.ZipFile(local_zip, 'r')
zip_ref.extractall('/tmp/')
zip_ref.close()
rock_dir = os.path.join('/tmp/rps/rock')
paper_dir = os.path.join('/tmp/rps/paper')
scissors_dir = os.path.join('/tmp/rps/scissors')
print('total training rock images:', len(os.listdir(rock_dir)))
print('total training paper images:', len(os.listdir(paper_dir)))
print('total training scissors images:', len(os.listdir(scissors_dir)))
rock_files = os.listdir(rock_dir)
print(rock_files[:10])
paper_files = os.listdir(paper_dir)
print(paper_files[:10])
scissors_files = os.listdir(scissors_dir)
print(scissors_files[:10])
There are 840 images of each class.
Let's see some examples of the images.
%matplotlib inline
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
pic_index = 2
next_rock = [os.path.join(rock_dir, fname)
for fname in rock_files[pic_index-2:pic_index]]
next_paper = [os.path.join(paper_dir, fname)
for fname in paper_files[pic_index-2:pic_index]]
next_scissors = [os.path.join(scissors_dir, fname)
for fname in scissors_files[pic_index-2:pic_index]]
for i, img_path in enumerate(next_rock+next_paper+next_scissors):
print(img_path)
img = mpimg.imread(img_path)
plt.imshow(img)
plt.axis('Off')
plt.show()
import tensorflow as tf
import keras_preprocessing
from keras_preprocessing import image
from keras_preprocessing.image import ImageDataGenerator
# set up the image generator
TRAINING_DIR = "/tmp/rps/"
training_datagen = ImageDataGenerator(
rescale = 1./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')
VALIDATION_DIR = "/tmp/rps-test-set/"
validation_datagen = ImageDataGenerator(rescale = 1./255)
train_generator = training_datagen.flow_from_directory(
TRAINING_DIR,
target_size=(150,150),
class_mode='categorical',
batch_size=126
)
validation_generator = validation_datagen.flow_from_directory(
VALIDATION_DIR,
target_size=(150,150),
class_mode='categorical',
batch_size=126
)
model = tf.keras.models.Sequential([
# Note the input shape is the desired size of the image 150x150 with 3 bytes color
# the first convolution
tf.keras.layers.Conv2D(64, (3,3), activation='relu', input_shape=(150, 150, 3)),
tf.keras.layers.MaxPooling2D(2, 2),
# the second convolution
tf.keras.layers.Conv2D(64, (3,3), activation='relu'),
tf.keras.layers.MaxPooling2D(2,2),
# the third convolution
tf.keras.layers.Conv2D(128, (3,3), activation='relu'),
tf.keras.layers.MaxPooling2D(2,2),
# the fourth convolution
tf.keras.layers.Conv2D(128, (3,3), activation='relu'),
tf.keras.layers.MaxPooling2D(2,2),
# flatten the results to feed into a DNN
tf.keras.layers.Flatten(),
tf.keras.layers.Dropout(0.5),
# 512 neuron hidden layer
tf.keras.layers.Dense(512, activation='relu'),
tf.keras.layers.Dense(3, activation='softmax')
])
model.summary()
model.compile(loss = 'categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])
history = model.fit(train_generator, epochs=25, steps_per_epoch=20, validation_data = validation_generator, verbose = 1, validation_steps=3)
model.save("rps.h5")
import matplotlib.pyplot as plt
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()
We can see that the training accuracy improved over time and trends towards 1.0. The validation accuracy was unstable in the beginning but has a value between 0.9 and 1.0 over time.
Model evaluation
Let us now test the model with some images that it hasn't previously seen. This new dataset with 33 images can be downloaded here.
import numpy as np
from google.colab import files
from keras.preprocessing import image
uploaded = files.upload()
for fn in uploaded.keys():
# predicting images
path = fn
img = image.load_img(path, target_size=(150, 150))
x = image.img_to_array(img)
x = np.expand_dims(x, axis=0)
images = np.vstack([x])
classes = model.predict(images, batch_size=10)
print(fn)
print(classes)
When using the image generator, the classes come from directories and thus were sorted in alphabetical order. So the first value is for paper, then rock, and then scissors.
The model guessed 32 out of 33 images correctly. It got only 1 image wrong, the paper9.png. If you download the images for yourself, you can see why the model got confused.
We can conclude that the model is highly accurate.