I verified with TensorFlow whether AI can beat Hikakinbunbun Rock-paper-scissors

1/23/2023

If you're losing to him in terms of annual income and personality,
I want him to at least win rock-paper-scissors.

So, this time,
we will use TensorFlow to create an AI that makes Hikakin Bun Bun win with
an image classifier.

What to prepare

Google Colaboratory (*Jupyter Notebook is also OK)
Distribution source code (*Not required if you want to start from scratch)

How to create AI that makes Hikakin Bunbun win

Let AI learn rock-paper-scissors

First, download the zip file containing the rock-paper-scissors image.

!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

`
–2022-09-19 05:51:56– https://storage.googleapis.com/laurencemoroney-blog.appspot.com/rps.zip
Resolving storage.googleapis.com (storage.googleapis.com)… 172.217.194.128, 142.251.10.128, 142.251.12.128, …
Connecting to storage.googleapis.com (storage.googleapis.com)|172.217.194.128|:443… connected.
HTTP request sent, awaiting response… 200 OK
Length: 200682221 (191M) [application/zip]
Saving to: ‘/tmp/rps.zip’

/tmp/rps.zip 100%[===================>] 191.38M 40.7MB/s in 4.7s

2022-09-19 05:52:02 (40.7 MB/s) – ‘/tmp/rps.zip’ saved [200682221/200682221]

–2022-09-19 05:52:02– https://storage.googleapis.com/laurencemoroney-blog.appspot.com/rps-test-set.zip
Resolving storage.googleapis.com (storage.googleapis.com)… 172.217.194.128, 142.251.10.128, 142.251.12.128, …
Connecting to storage.googleapis.com (storage.googleapis.com)|172.217.194.128|:443… connected.
HTTP request sent, awaiting response… 200 OK
Length: 29516758 (28M) [application/zip]
Saving to: ‘/tmp/rps-test-set.zip’

/tmp/rps-test-set.z 100%[===================>] 28.15M –.-KB/s in 0.1s

2022-09-19 05:52:02 (204 MB/s) – ‘/tmp/rps-test-set.zip’ saved [29516758/29516758]
`

Unzip the zip and save it in the /tmp/ directory.

import os
import zipfile

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()

Categorize rock-paper-scissors images for learning into
goo, choki, and par.

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])

total training rock images: 840 total training paper images: 840 total training scissors images: 840 ['rock01-027.png', 'rock05ck01-108.png', ' rock05ck01-090.png', 'rock01-040.png', 'rock02-108.png', 'rock02-022.png', 'rock01-060.png', 'rock07-k03-061.png', 'rock06ck02-032.png', 'rock03-069.png'] ['paper05-114.png', 'paper03-099.png', 'paper06-011.png', 'paper07-053.png ', 'paper03-040.png', 'paper07-040.png', 'paper05-082.png', 'paper07-116.png', 'paper01-041.png', ' paper07-035.png'] ['scissors04-034.png', 'testscissors01-033.png', 'testscissors03-058.png', 'testscissors01-087.png', ' testscissors03-044.png', 'testscissors01-040.png', 'scissors04-080.png', 'scissors03-069.png', 'testscissors03-088.png', 'scissors04-038.png']

Plot.

%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):
  img = mpimg.imread(img_path)
  plt.imshow(img)
  plt.axis('Off')
  plt.show()

Learn.

import tensorflow as tf
import keras_preprocessing
from keras_preprocessing import image
from keras_preprocessing.image import ImageDataGenerator

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)

# Create training data
train_generator = training_datagen.flow_from_directory(
    TRAINING_DIR,
    target_size=(150,150),
    class_mode='categorical',
  batch_size=126
)

# Create verification data
validation_generator = validation_datagen.flow_from_directory(
    VALIDATION_DIR,
    target_size=(150,150),
    class_mode='categorical',
  batch_size=126
)

# Define a neural network
model = tf.keras.models.Sequential([
    tf.keras.layers.Conv2D(64, (3,3), activation='relu', input_shape=(150, 150, 3)),
    tf.keras.layers.MaxPooling2D(2, 2),
    tf.keras.layers.Conv2D(64, (3,3), activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),
    tf.keras.layers.Conv2D(128, (3,3), activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),
    tf.keras.layers.Conv2D(128, (3,3), activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dropout(0.5),
    tf.keras.layers.Dense(512, activation='relu'),
    tf.keras.layers.Dense(3, activation='softmax')
])

# Summary
model.summary()

# Compiling the neural network
model.compile(loss = 'categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])

# Learn
history = model.fit(train_generator, epochs=25, steps_per_epoch=20, validation_data = validation_generator, verbose = 1, validation_steps=3)

model.save("rps.h5")

The study is over.

Play rock-paper-scissors with Hikakin and let the AI win

Now upload the file and output the winning hand to Hikakin.

import numpy as np
from google.colab import files
from keras.preprocessing import image

uploaded = files.upload()

for fn in uploaded.keys():
  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)
  hikakin_result = np.array(classes[0])
  paper = np.array([1, 0, 0])
  rock =  np.array([0, 1, 0])
  scissors =  np.array([0, 0, 1])
  if hikakin_result[0] == paper[0]:
    print("✌")
  if hikakin_result[1] == rock[1]:
    print("🖐")
  if hikakin_result[2] == scissors[2]:
    print("✊")

Let's do the above.
I will try to play rock-paper-scissors in the 10th round.