image_recognition1.py

import tensorflow as tf
from tensorflow import keras
import matplotlib.pyplot as plt
import numpy as np
import random

# --- 1. Load dataset MNIST ---
mnist = keras.datasets.mnist
(x_train, y_train), (x_test, y_test) = mnist.load_data()

# Normalisasi data (ubah nilai pixel dari 0–255 menjadi 0–1)
x_train, x_test = x_train / 255.0, x_test / 255.0

# Ambil subset data 
subset_train = 10000  
subset_test = 2000     
x_train, y_train = x_train[:subset_train], y_train[:subset_train]
x_test, y_test = x_test[:subset_test], y_test[:subset_test]

# model Neural Network sederhana 
# Model terdiri dari 3 lapisan:
model = keras.Sequential([
    keras.layers.Flatten(input_shape=(28, 28)),
    keras.layers.Dense(64, activation='relu'),
    keras.layers.Dense(10, activation='softmax')
])

# Kompilasi model
model.compile(
    optimizer='adam',                              # algoritma optimasi
    loss='sparse_categorical_crossentropy',        #  label angka
    metrics=['accuracy']                           #  akurasi
)

# Latih model
print("Training sedang berjalan...\n")

model.fit(
    x_train, y_train,
    epochs=3,
    batch_size=128,
    validation_split=0.1,
    verbose=2
)

#  6. Evaluasi hasil pada data uji
test_loss, test_acc = model.evaluate(x_test, y_test, verbose=2)
print(f"\nAkurasi pada data test: {test_acc * 100:.2f}%")

# Prediksi satu gambar acak
# Pilih satu gambar dari data uji
index = random.randint(0, len(x_test) - 1)
plt.imshow(x_test[index], cmap='gray')
plt.title(f"Asli: {y_test[index]} (index {index})")
plt.axis('off')
plt.show()

# Model membuat prediksi untuk gambar tersebut
pred = model.predict(np.expand_dims(x_test[index], axis=0), verbose=0)
print(f"Prediksi model: {np.argmax(pred)}")

# Tampilkan 5 gambar acak dengan hasil prediksi 
print("\nMenampilkan 5 gambar acak dari dataset test:")
for i in range(5):
    idx = random.randint(0, len(x_test) - 1)

    # Tampilkan gambar
    plt.imshow(x_test[idx], cmap='gray')
    plt.title(f"Asli: {y_test[idx]} (index {idx})")
    plt.axis('off')
    plt.show()

    # Prediksi dan tampilkan hasilnyo
    pred = model.predict(np.expand_dims(x_test[idx], axis=0), verbose=0)
    print(f"Gambar ke-{i+1} → Asli: {y_test[idx]}, Prediksi: {np.argmax(pred)}")