Klasifikasi Gambar Tangan dengan Gaya Batu, Gunting dan Kertas

!wget --no-check-certificate \
  https://d.tau.fan/rockpaperscissors.zip \
  -O /tmp/rockpaperscissors.zip

!pip install split-folders

import tensorflow as tf
from tensorflow.keras.optimizers import RMSprop
from tensorflow.keras.preprocessing.image import ImageDataGenerator

# Extract dataset zip
import zipfile,os
# import splitfolders

local_zip = '/tmp/rockpaperscissors.zip'
zip_ref = zipfile.ZipFile(local_zip, 'r')
zip_ref.extractall('/tmp')
zip_ref.close()

# mendefinisikan nama direktori untuk `data latih`` dan `data validasi``
base_dir = '/tmp/rockpaperscissors/rps-cv-images'
# splitfolders.ratio(base_dir, output=base_dir, ratio=(0.8,0.2))

# train_dir = os.path.join(base_dir, 'train')
# validation_dir = os.path.join(base_dir, 'val')

base_dir

os.listdir(base_dir)

# os.listdir(train_dir)

# os.listdir(validation_dir)

# # membuat direktori batu pada direktori data training
# train_rock_dir = os.path.join(train_dir, 'rock')

# # membuat direktori gunting pada direktori data training
# train_scissor_dir = os.path.join(train_dir, 'scissors')

# # membuat direktori kertas pada direktori data training
# train_paper_dir = os.path.join(train_dir, 'paper')

# # membuat direktori batu pada direktori data validasi
# validation_rock_dir = os.path.join(validation_dir, 'rock')

# # membuat direktori gunting pada direktori data validasi
# validation_scissor_dir = os.path.join(validation_dir, 'scissors')

# # membuat direktori kertas pada direktori data validasi
# validation_paper_dir = os.path.join(validation_dir, 'paper')

# proses augmentasi gambar pada sampel data traininng
train_datagen = ImageDataGenerator(rescale=1./255, 
                                   rotation_range=20, 
                                   shear_range = 0.2, 
                                   horizontal_flip=True, 
                                   fill_mode = 'wrap',
                                   validation_split = 0.4)

# ImageDataGenerator(rescale=1./255, 
                            # rotation_range=20, 
                            # horizontal_flip=True, 
                            # shear_range = 0.2, 
                            # validation_split = 0.4, 
                            # fill_mode = 'nearest')

# proses augmentasi gambar pada sampel data testing
test_datagen = ImageDataGenerator(rescale=1./255, 
                                  zoom_range = 0.2, 
                                  shear_range = 0.2, 
                                  horizontal_flip=True, 
                                  validation_split = 0.4)

# persiapan data training
train_generator = train_datagen.flow_from_directory(
        base_dir,  
        # mengubah resolusi seluruh gambar menjadi 150x150 piksel
        target_size=(100, 150),  
        shuffle = True,
        subset='training',
        class_mode='categorical')

# persiapan data validasi
validation_generator = train_datagen.flow_from_directory(
        base_dir, 
        # mengubah resolusi seluruh gambar menjadi 150x150 piksel
        target_size=(100, 150),  
        subset='validation',
        class_mode='categorical')

model = tf.keras.models.Sequential([
    # Conv2D untuk mengekstrak atribut pada gambar 
    tf.keras.layers.Conv2D(32, (3,3), activation='relu', input_shape=(100, 150, 3)),
    # max pooling untuk mengurangi resolusi gambar
    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(512, (3,3), activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(128, activation='relu'),
    tf.keras.layers.Dense(3, activation='softmax')
])

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

history = model.fit(
      train_generator,
      steps_per_epoch=25,  # berapa batch yang akan dieksekusi pada setiap epoch
      epochs=25,
      validation_data=validation_generator, # menampilkan akurasi pengujian data validasi
      validation_steps=5,  # berapa batch yang akan dieksekusi pada setiap epoch
      verbose=2)

import matplotlib.pyplot as plt

accur = history.history['accuracy']
val_accur = history.history['val_accuracy'] 
loss = history.history['loss']
val_loss = history.history['val_loss']

epochs = range(len(accur))

plt.figure(figsize=(12,8))
plt.plot(epochs, accur, 'r', label='Training accuracy')
plt.plot(epochs, val_accur, 'b', label='Validation accuracy')
plt.title('Training and validation accuracy')
plt.legend(loc=0)
plt.show()

print(train_generator.class_indices)

images = np.vstack([x])
model.predict(images, batch_size=10)

from keras.models import model_from_json

# serialize model to JSON
model_json = model.to_json()

with open("modelImageCNN.json", "w") as json_file:
    json_file.write(model_json)

# serialize weights to HDF5
model.save_weights("modelImageCNN.h5")

import numpy as np
from google.colab import files
from keras.preprocessing import image
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
%matplotlib inline

# mengupload file, trus ditampung disini
uploaded = files.upload()

for fn in uploaded.keys():
 
  # predicting images
  path = fn
  
  # memilih file gambar secara interaktif
  img = image.load_img(path, target_size=(100,150))
  imgplot = plt.imshow(img)
  x = image.img_to_array(img)
  x = np.expand_dims(x, axis=0)

  # resize gambar dan mengubahnya menjadi larik numpy
  images = np.vstack([x])
  classes = model.predict(images, batch_size=10)
  outclass = np.argmax(classes)

  print(fn)
  if outclass == 0:
    print('paper')
  elif outclass == 1:
    print('rock')
  else:
    print('scissors')