Update app.py

app.py

@@ -2,92 +2,99 @@ import gradio as gr
 from transformers import pipeline
 from PIL import Image
 import numpy as np
-import cv2
 
-# Model utama AI-detector
+# Model utama
 detector = pipeline("image-classification", model="umm-maybe/AI-image-detector")
-# Model tambahan general classifier
 general = pipeline("image-classification", model="google/vit-base-patch16-224")
 
-def variance_of_laplacian(image):
-    # Menghitung blur menggunakan Laplacian
-    return cv2.Laplacian(image, cv2.CV_64F).var()
-
-def estimate_noise(image):
-    # Menghitung noise sederhana
-    img_gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
-    m = np.mean(img_gray)
-    noise = np.mean((img_gray - m)**2)
+# Estimasi blur & noise tanpa OpenCV
+def estimate_blur(img):
+    gray = img.convert("L")
+    arr = np.array(gray)
+    gx, gy = np.gradient(arr)
+    laplacian_var = np.var(gx + gy)
+    return laplacian_var
+
+def estimate_noise(img):
+    gray = img.convert("L")
+    arr = np.array(gray)
+    m = np.mean(arr)
+    noise = np.mean((arr - m) ** 2)
     return noise
 
+# Pengecekan pola photorealistic AI
+def photorealistic_ai_adjustment(img, ai_score):
+    # Cek metadata, blur, dan noise
+    has_metadata = bool(img.info)
+    blur = estimate_blur(img)
+    noise = estimate_noise(img)
+
+    # Logika sensitif:
+    # Foto asli biasanya ada metadata kamera, blur & noise tertentu
+    # Jika metadata kosong, blur sangat rendah, dan noise aneh => kemungkinan AI lebih tinggi
+    adjustment = 0
+    if not has_metadata:
+        adjustment += 10
+    if blur < 100:  # sangat tajam
+        adjustment += 10
+    if noise < 50:  # terlalu bersih
+        adjustment += 10
+
+    ai_score += adjustment
+    ai_score = min(ai_score, 100)  # maksimal 100%
+    return ai_score
+
 def detect_image(img):
     try:
-        # Convert PIL ke OpenCV format
-        img_cv = cv2.cvtColor(np.array(img), cv2.COLOR_RGB2BGR)
-
-        # Blur dan Noise
-        blur_score = variance_of_laplacian(img_cv)
-        noise_score = estimate_noise(img_cv)
-
-        # Metadata kamera
-        metadata_camera = img.info.get("exif") is not None
-
-        # Prediksi AI
-        result_ai = detector(img)
-        label_ai = result_ai[0]['label'].lower()
-        conf_ai = result_ai[0]['score'] * 100
-
-        # Prediksi general
-        result_gen = general(img)
-        label_gen = result_gen[0]['label']
-        conf_gen = result_gen[0]['score'] * 100
-
-        # Hybrid scoring
-        if "artificial" in label_ai or "fake" in label_ai:
-            ai_score = max(conf_ai, 70)
-        elif "human" in label_ai:
-            ai_score = 100 - conf_ai * 0.7
+        # Prediksi AI detector
+        result1 = detector(img)
+        label1 = result1[0]['label'].lower()
+        conf1 = round(result1[0]['score'] * 100, 2)
+
+        # Prediksi model general
+        result2 = general(img)
+        label2 = result2[0]['label']
+        conf2 = round(result2[0]['score'] * 100, 2)
+
+        # Hitung AI % awal
+        if "artificial" in label1 or "fake" in label1:
+            ai_percent = conf1
+        elif "human" in label1:
+            ai_percent = 100 - conf1
         else:
-            ai_score = 50
+            ai_percent = conf1
 
-        # Tambah bobot metadata kamera
-        if metadata_camera:
-            ai_score -= 10
+        # Adjust untuk photorealistic AI
+        ai_percent = photorealistic_ai_adjustment(img, ai_percent)
+        ai_percent = round(ai_percent, 2)
+        real_percent = round(100 - ai_percent, 2)
 
-        # Penalti blur rendah untuk AI photorealistic
-        if blur_score < 100:
-            ai_score += 10  # foto asli biasanya lebih tajam
-        # Penalti noise sangat rendah (AI sering terlalu bersih)
-        if noise_score < 30:
-            ai_score += 5
+        # Blur & Noise untuk info
+        blur_score = round(estimate_blur(img), 2)
+        noise_score = round(estimate_noise(img), 2)
+        has_metadata = "Ada" if img.info else "Tidak"
 
-        # Clamp nilai
-        ai_score = max(0, min(100, ai_score))
-        real_score = 100 - ai_score
-
-        # Output
         output = f"""
 ### Hasil Deteksi:
-🖼️ Gambar ini {round(ai_score,2)}% AI / {round(real_score,2)}% Asli
+🖼️ Gambar ini {ai_percent}% AI / {real_percent}% Asli
 
-**Model AI-detector:** {result_ai[0]['label']} ({round(conf_ai,2)}%)  
-**Model General (ViT):** {label_gen} ({round(conf_gen,2)}%)  
-Blur Score: {round(blur_score,2)}  
-Noise Score: {round(noise_score,2)}  
-Metadata Kamera: {'Ada' if metadata_camera else 'Tidak'}
+**Model AI-detector:** {label1} ({conf1}%)
+**Model General (ViT):** {label2} ({conf2}%)
+**Blur Score:** {blur_score}
+**Noise Score:** {noise_score}
+**Metadata Kamera:** {has_metadata}
 """
         return output
-
     except Exception as e:
         return f"Terjadi error: {str(e)}"
 
-# Gradio interface
+# UI Gradio
 iface = gr.Interface(
     fn=detect_image,
     inputs=gr.Image(type="pil"),
     outputs="markdown",
-    title="AI vs Real Image Detector (Advanced Hybrid)",
-    description="Gratis, mendeteksi AI photorealistic dengan blur & noise analisis, semua persentase."
+    title="Hybrid AI vs Real Image Detector (Sensitif Photorealistic)",
+    description="Upload foto untuk mendeteksi kemungkinan besar gambar asli atau hasil AI. Lebih sensitif terhadap AI photorealistic. 100% gratis."
 )
 
 if __name__ == "__main__":