predict.py

import numpy as np
import cv2
import os
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing.image import img_to_array

model = load_model('deepfake_detection_model.h5')

def preprocess_image(image_path):
    image = cv2.imread(image_path)
    if image is None:
        raise FileNotFoundError(f"Image nahi mili: {image_path}")
    image = cv2.resize(image, (96, 96))
    image = img_to_array(image)
    image = np.expand_dims(image, axis=0)
    image = image / 255.0
    return image

def predict_image(image_path):
    image = preprocess_image(image_path)
    prediction = model.predict(image, verbose=0)
    print(f"Raw prediction: {prediction}")
    class_label = np.argmax(prediction, axis=1)[0]
    confidence = float(np.max(prediction)) * 100
    # Class 1 = Fake, Class 0 = Real (dataset mapping)
    label = "Fake" if class_label == 1 else "Real"
    print(f"Prediction  : {label}")
    print(f"Confidence  : {confidence:.1f}%")
    return label

# Fake image test
image_path = "real_and_fake_face_detection/real_vs_fake/real-vs-fake/train/fake/001DDU0NI4.jpg"
predict_image(image_path)