logo 2.png

Cara Menggunakan Program

Buka Google Colaboratory menggunakan akun anda
Install library bila diperlukan
Gunakan script code program di bawah menggunakan Google Colaboratory
Modifikasi sesuai materi yang dipelajari

Deskripsi Program

Impor library yang akan digunakan

# Library u/ ekstrak dan atur lokasi

import zipfile

import os

# Library u/ machine learning Tensorflow

from tensorflow.keras.preprocessing.image import ImageDataGenerator

from tensorflow.keras.models import Sequential

from tensorflow.keras.layers import Conv2D, MaxPooling2D, Activation, Dense, Flatten

from tensorflow.keras.layers import Dropout

from tensorflow.keras.optimizers import Adam

from tensorflow.keras.callbacks import EarlyStopping

# Library untuk upload file gcolabs dan plot

import keras.utils as image

from google.colab import files

import numpy as np

import matplotlib.pyplot as plt

import matplotlib.image as mpimg

%matplotlib inline

Upload Dataset Rock Paper Scissors

#Import data dari local drive

from google.colab import files

uploaded = files.upload()

Unzip / Ekstrak Dataset

# Ekstrak file tipe zip

ekstrak_zip = 'rockpaperscissors.zip'

out_zip = zipfile.ZipFile(ekstrak_zip, 'r')

out_zip.extractall('/dataset')

out_zip.close()

Atur Lokasi Dataset RockPaperScissors

# Mengatur Lokasi dataset yang akan digunakan

dir_dataset = "/dataset/rockpaperscissors/rps-cv-images"

dir_paper = os.path.join("/dataset/rockpaperscissors/rps-cv-images/paper")

dir_rock = os.path.join("/dataset/rockpaperscissors/rps-cv-images/rock")

dir_scissors = os.path.join("/dataset/rockpaperscissors/rps-cv-images/scissors")

# Mengecek jumlah data rock paper scissor

total_paper = len(os.listdir(dir_paper))

total_rock = len(os.listdir(dir_rock))

total_scissors = len(os.listdir(dir_scissors))

# Mencetak jumlah data rock paper scissor

print("Total Data Paper Image : ",total_paper)

print("Total Data Rock Image : ",total_rock)

print("Total Data Scissors Image : ",total_scissors)

Implementasi Image Data Generator Tensorflow untuk pengolahan gambar

# Pembuatan Train dan Validation data generator

# Untuk augmentasi gambar

# Ukuran validation 20% dari dataset

val_size = 0.2

Train_datagen = ImageDataGenerator(

rotation_range = 30,

brightness_range = [0.2,1.0],

shear_range = 0.2,

zoom_range = 0.2,

horizontal_flip = True,

fill_mode = "nearest",

rescale = 1./255,

validation_split = val_size

)

Validation_datagen = ImageDataGenerator(

rotation_range = 30,

brightness_range = [0.2,1.0],

shear_range = 0.2,

zoom_range = 0.2,

horizontal_flip = True,

fill_mode = "nearest",

rescale = 1./255,

validation_split = val_size

)

Data Training(80%) dan Data Validation(20%)

# Mengatur lebar dan tinggi gambar

img_width = 150

img_height = 150

# Train dan Validation generator dengan mode categorical

Train_generator = Train_datagen.flow_from_directory(

dir_dataset,

target_size = (img_width,img_height),

color_mode = "rgb",

class_mode = "categorical",

batch_size = 16,

shuffle = True,

subset = "training"

)

Validation_generator = Validation_datagen.flow_from_directory(

dir_dataset,

target_size = (img_width,img_height),

color_mode = "rgb",

class_mode = "categorical",

batch_size = 16,

shuffle = False,

subset = "validation"

)

Model Sequential dan Dropout Regularization

# Pembuatan Model dengan tipe Sequential

Model = Sequential(

[

# Activation menggunakan relu + softmax

Conv2D(32, (3,3), strides = (1,1), activation = 'relu' , input_shape = (img_width,img_height,3)),

MaxPooling2D(pool_size = (2,2), padding = 'valid'),

Conv2D(64, (3,3), strides = (1,1), activation = 'relu' ),

MaxPooling2D(pool_size = (2,2), padding = 'valid'),

Conv2D(128, (3,3), strides = (1,1), activation = 'relu' ),

MaxPooling2D(pool_size = (2,2), padding = 'valid'),

Flatten(),

# Droupout

Dropout(0.2),

Dense(128, activation = 'relu'),

Dense(3, activation='softmax')

]

)

Model Compile menggunakan Optimizer

# Penggunaan Optimizer 'Adam' dengan learning rate 0.00146

Adam(learning_rate=0.00146, name='Adam')

Model.compile(optimizer = 'Adam',loss = 'categorical_crossentropy',metrics = ['accuracy'])

Proses Training Dataset dan Penggunaan Callbacks

# Proses Training

batch_size = 16

# Mengatur Jumlah Validation Steps

Validation_steps = Validation_generator.samples / Validation_generator.batch_size -1

# Penggunaan Callbacks dengan fungsi EarlyStopping dengan memonitoring Val_Loss

callback = EarlyStopping(

monitor = 'val_loss',

min_delta = 0.001,

patience = 3,

verbose = 1,

mode = 'auto'

)

Model.fit(

Train_generator,

steps_per_epoch = 30,

epochs = 20,

verbose = 1,

validation_data = Validation_generator,

validation_steps = Validation_steps,

callbacks=[callback]

)

Prediksi Hasil Model gambar antara Rock Paper Scissors

# Upload File ke google colabs

uploaded = files.upload()

# pengkondisian

for file_upload in uploaded.keys():

path = file_upload

img = image.load_img(path, target_size=(150,150))

imgplot = plt.imshow(img)

x = image.img_to_array(img)

x = np.expand_dims(x, axis=0)

# Membuat numpy vstack array untuk hasil prediksi

images = np.vstack([x])

classes = Model.predict(images, batch_size=16)

print("\n")

print('Hasil Prediksi : ',classes[0],'\n')

# Pengkondisian

# Ketika array ke 0 = 1 maka tergolong Paper

# Ketika array ke 1 = 1 maka tergolong Rock

# Ketika array ke 2 = 1 maka tergolong Scissors

if classes[0][0] == 1:

print('Kategori Gambar : Paper')

elif classes[0][1] == 1:

print('Kategori Gambar : Rock')

else:

print('Kategori Gambar : Scissor')

Paket 3 (CNN)

Deskripsi Program