Update app.py

app.py

@@ -1,65 +1,85 @@
 import gradio as gr
 from transformers import pipeline
 from PIL import Image
+import cv2
+import numpy as np
 
-# === Load 3 model publik khusus AI vs Real ===
-model1 = pipeline("image-classification", model="prithivMLmods/AI-vs-Deepfake-vs-Real")
-model2 = pipeline("image-classification", model="dima806/ai_vs_real_image_detection")
-model3 = pipeline("image-classification", model="Hemg/AI-VS-REAL-IMAGE-DETECTION")
+# =========================
+# Preprocessing function
+# =========================
+def preprocess_image(img_path):
+    img = cv2.imread(img_path)
+    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+    img = cv2.resize(img, (512, 512))  
+    img = cv2.GaussianBlur(img, (3, 3), 0)  # Kurangi noise
+    return Image.fromarray(img)
 
-def detect_image(img: Image.Image):
-    results = {}
+# =========================
+# Load models
+# =========================
+detector_ai = pipeline("image-classification", model="umm-maybe/AI-image-detector")
+detector_resnet = pipeline("image-classification", model="microsoft/resnet-50")
 
-    # Model 1
-    res1 = model1(img)
-    results["Model 1 (prithivMLmods)"] = res1
+# =========================
+# Detection Function
+# =========================
+def detect_image(img):
+    # Preprocess
+    img = preprocess_image(img)
 
-    # Model 2
-    res2 = model2(img)
-    results["Model 2 (dima806)"] = res2
+    # Run AI detector
+    ai_results = detector_ai(img)
+    ai_dict = {res['label']: res['score'] for res in ai_results}
 
-    # Model 3
-    res3 = model3(img)
-    results["Model 3 (Hemg)"] = res3
+    human_score = ai_dict.get("human", 0)
+    artificial_score = ai_dict.get("artificial", 0)
 
-    # --- Ensemble sederhana ---
-    ai_score, real_score = 0, 0
-    for res in [res1, res2, res3]:
-        top = res[0]
-        label = top["label"].lower()
-        score = top["score"]
-        if "real" in label:
-            real_score += score
-        else:
-            ai_score += score
+    # Run ResNet for natural object cross-check
+    resnet_results = detector_resnet(img)
+    top_resnet = resnet_results[0]  # ambil label dengan skor tertinggi
 
-    # Normalisasi
-    total = ai_score + real_score + 1e-9
-    ai_percent = (ai_score / total) * 100
-    real_percent = (real_score / total) * 100
-
-    # Keputusan akhir
-    if real_percent > ai_percent:
-        verdict = f"✅ Foto Asli ({real_percent:.2f}%)"
+    # =========================
+    # Decision Rule with Threshold
+    # =========================
+    if artificial_score > 0.75:
+        verdict = "🟣 AI-generated"
+    elif human_score > 0.65:
+        verdict = "🟢 Foto Asli"
     else:
-        verdict = f"⚠️ AI Generated ({ai_percent:.2f}%)"
+        verdict = "⚠️ Tidak Pasti"
+
+    # Extra check: kalau AI-detector bilang AI tapi ResNet yakin objek nyata
+    if verdict == "🟣 AI-generated" and top_resnet['score'] > 0.70:
+        verdict = "⚠️ Tidak Pasti (deteksi objek nyata tinggi)"
+
+    # =========================
+    # Output
+    # =========================
+    summary = f"""
+📊 Ringkasan Deteksi:
 
-    # Format output
-    output = f"## 📊 Ringkasan Deteksi\n"
-    for name, res in results.items():
-        output += f"\n🔹 **{name}**: {res}\n"
+🔹 AI Detector:
+   - Human: {human_score:.2%}
+   - Artificial: {artificial_score:.2%}
 
-    output += f"\n=== 🧠 ENSEMBLE HASIL AKHIR ===\n{verdict}\n"
+🔹 ResNet50:
+   - Top Label: {top_resnet['label']} ({top_resnet['score']:.2%})
 
-    return output
+=== HASIL AKHIR ===
+{verdict}
+"""
+    return summary
 
-iface = gr.Interface(
+# =========================
+# Gradio App
+# =========================
+demo = gr.Interface(
     fn=detect_image,
-    inputs=gr.Image(type="pil"),
-    outputs="markdown",
-    title="Deteksi AI vs Foto Asli (Ensemble 3 Model Publik)",
-    description="Menggunakan 3 model publik Hugging Face (prithivMLmods, dima806, Hemg) untuk mendeteksi apakah gambar AI-generated atau asli."
+    inputs=gr.Image(type="filepath"),
+    outputs="text",
+    title="Deteksi AI vs Foto Asli",
+    description="Upload foto untuk mendeteksi apakah gambar AI-generated atau Foto Asli. Menggunakan threshold + preprocessing."
 )
 
 if __name__ == "__main__":
-    iface.launch()
+    demo.launch()