Upload deepfake_model.py

deepfake_model.py

@@ -0,0 +1,100 @@
+# Imports
+import sys
+import tensorflow as tf
+import cv2
+import pandas as pd
+import numpy as np
+import matplotlib.pyplot as plt
+
+
+batch_size = 32
+num_epochs = 4
+
+# Load the dataset
+data_dir = '/Users/daisy./Documents/data'
+dataset = tf.keras.preprocessing.image_dataset_from_directory(data_dir)
+
+# Split the dataset into train, validation, and test sets
+train_size = int(len(dataset) * 0.7)
+val_size = int(len(dataset) * 0.2)
+test_size = int(len(dataset) * 0.1)
+
+train = dataset.take(train_size)
+val = dataset.skip(train_size).take(val_size)
+test = dataset.skip(train_size + val_size).take(test_size)
+
+# Preprocess the data for the Xception model
+tf.keras.backend.clear_session()
+
+preprocess = tf.keras.applications.xception.preprocess_input
+
+# Apply preprocessing before batching
+train = train.map(lambda x, y: (preprocess(x), y))
+val = val.map(lambda x, y: (preprocess(x), y))
+test = test.map(lambda x, y: (preprocess(x), y))
+
+# Shuffle and batch the datasets, leaving the last batch incomplete
+train_set = train.shuffle(buffer_size=len(train)).unbatch().batch(batch_size).repeat(num_epochs).prefetch(1)
+valid_set = val.unbatch().batch(batch_size).repeat(num_epochs)
+test_set = test.unbatch().batch(batch_size).repeat(num_epochs)
+
+# Check the element specifications
+# print("Train set element spec:", train_set.element_spec)
+# print("Validation set element spec:", valid_set.element_spec)
+# print("Test set element spec:", test_set.element_spec)
+
+tf.random.set_seed(42)  # extra code – ensures reproducibility
+
+input = tf.keras.Input(shape=(256, 256, 3))
+inp_resized = tf.keras.layers.Lambda(lambda X: tf.image.resize(X, (299, 299)))(input)
+
+base_model = tf.keras.applications.xception.Xception(weights="imagenet", include_top=False, input_tensor=inp_resized)
+
+avg = tf.keras.layers.GlobalAveragePooling2D()(base_model.output)
+output = tf.keras.layers.Dense(1, activation="sigmoid")(avg)
+model = tf.keras.Model(inputs=base_model.input, outputs=output)
+
+# freezing layers
+for layer in base_model.layers:
+    layer.trainable = False
+
+class PrintShapeCallback(tf.keras.callbacks.Callback):
+    def on_epoch_begin(self, epoch, logs=None):
+        print(f"Epoch {epoch+1}: Input shape:", self.model.input_shape)
+
+print_shape_callback = PrintShapeCallback()
+
+# Set the number of steps per epoch
+steps_per_epoch = 5
+
+optimizer = tf.keras.optimizers.SGD(learning_rate=0.1, momentum=0.9)
+model.compile(loss="binary_crossentropy", optimizer=optimizer, metrics=["accuracy"])
+history = model.fit(train_set, steps_per_epoch=steps_per_epoch, validation_data=valid_set, epochs=10, callbacks=[print_shape_callback])
+
+# plot model performance
+acc = history.history['accuracy']
+val_acc = history.history['val_accuracy']
+loss = history.history['loss']
+val_loss = history.history['val_loss']
+epochs_range = range(1, len(history.epoch) + 1)
+
+plt.figure(figsize=(15,5))
+
+plt.subplot(1, 2, 1)
+plt.plot(epochs_range, acc, label='Train Set')
+plt.plot(epochs_range, val_acc, label='Val Set')
+plt.legend(loc="best")
+plt.xlabel('Epochs')
+plt.ylabel('Accuracy')
+plt.title('Model Accuracy')
+
+plt.subplot(1, 2, 2)
+plt.plot(epochs_range, loss, label='Train Set')
+plt.plot(epochs_range, val_loss, label='Val Set')
+plt.legend(loc="best")
+plt.xlabel('Epochs')
+plt.ylabel('Loss')
+plt.title('Model Loss')
+
+plt.tight_layout()
+plt.show()