app.py

import os
import cv2
import numpy as np
import gradio as gr
import tensorflow as tf
from tensorflow.keras.models import load_model
from tensorflow.keras.applications.xception import preprocess_input as xcp_pre
from tensorflow.keras.applications.efficientnet import preprocess_input as eff_pre
from huggingface_hub import hf_hub_download

# Download and load models
xcp_path = hf_hub_download(repo_id="Zeyadd-Mostaffa/deepfake-image-detector", filename="xception_model.h5")
eff_path = hf_hub_download(repo_id="Zeyadd-Mostaffa/deepfake-image-detector", filename="efficientnet_model.h5")
xcp_model = load_model(xcp_path)
eff_model = load_model(eff_path)

def predict(image):
    # Use the full image directly (no face extraction)
    xcp_img = cv2.resize(image, (299, 299))
    eff_img = cv2.resize(image, (224, 224))

    xcp_tensor = xcp_pre(xcp_img.astype(np.float32))[np.newaxis, ...]
    eff_tensor = eff_pre(eff_img.astype(np.float32))[np.newaxis, ...]

    xcp_pred = xcp_model.predict(xcp_tensor, verbose=0).flatten()[0]
    eff_pred = eff_model.predict(eff_tensor, verbose=0).flatten()[0]

    avg_pred = (xcp_pred + eff_pred) / 2
    label = "Real" if avg_pred > 0.5 else "Fake"

    return  label

iface = gr.Interface(
    fn=predict,
    inputs=gr.Image(type="filepath"),
    outputs=gr.JSON(),  # <-- very important
    live=False
)

iface.launch()