Update app.py

app.py

@@ -1,51 +1,77 @@
 import gradio as gr
 import torch
 import numpy as np
-from PIL import Image
 from transformers import AutoImageProcessor, AutoModel
-from sklearn.cluster import KMeans
 
-# Load model pretrained (tanpa training ulang)
-processor = AutoImageProcessor.from_pretrained("google/vit-base-patch16-224")
-model = AutoModel.from_pretrained("google/vit-base-patch16-224")
+# =========================
+# Load model pretrained
+# =========================
+processor = AutoImageProcessor.from_pretrained(
+    "google/vit-base-patch16-224"
+)
+model = AutoModel.from_pretrained(
+    "google/vit-base-patch16-224"
+)
 model.eval()
 
-def extract_features(images):
-    features = []
+# =========================
+# Fungsi ekstraksi fitur
+# =========================
+def get_embedding(image):
+    inputs = processor(images=image, return_tensors="pt")
+    with torch.no_grad():
+        outputs = model(**inputs)
+    embedding = outputs.last_hidden_state.mean(dim=1)
+    return embedding.squeeze().numpy()
+
+# =========================
+# Cosine similarity
+# =========================
+def cosine_similarity(a, b):
+    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))
+
+# =========================
+# Fungsi utama aplikasi
+# =========================
+def unsupervised_image_similarity(img1, img2):
+    if img1 is None or img2 is None:
+        return "❌ Silakan upload DUA gambar."
 
-    for img in images:
-        inputs = processor(images=img, return_tensors="pt")
-        with torch.no_grad():
-            outputs = model(**inputs)
-        embedding = outputs.last_hidden_state.mean(dim=1).squeeze().numpy()
-        features.append(embedding)
+    emb1 = get_embedding(img1)
+    emb2 = get_embedding(img2)
 
-    return np.array(features)
+    similarity = cosine_similarity(emb1, emb2)
 
-def cluster_images(images, n_cluster):
-    if len(images) < n_cluster:
-        return "Jumlah gambar harus ≥ jumlah cluster"
+    hasil = f"""
+HASIL UNSUPERVISED LEARNING
 
-    X = extract_features(images)
-    kmeans = KMeans(n_clusters=n_cluster, random_state=42)
-    labels = kmeans.fit_predict(X)
+Nilai kemiripan gambar : {similarity:.2f}
 
-    hasil = ""
-    for i in range(n_cluster):
-        idx = np.where(labels == i)[0]
-        hasil += f"\n📁 Cluster {i+1}: {len(idx)} gambar\n"
+Interpretasi:
+- Mendekati 1  → Gambar sangat mirip
+- Mendekati 0  → Gambar berbeda
+- Nilai negatif → Sangat tidak mirip
 
+Catatan:
+Tidak menggunakan dataset, label, atau training ulang.
+"""
     return hasil
 
+# =========================
+# Tampilan aplikasi
+# =========================
 app = gr.Interface(
-    fn=cluster_images,
+    fn=unsupervised_image_similarity,
     inputs=[
-        gr.Gallery(label="Upload beberapa gambar"),
-        gr.Slider(2, 5, step=1, label="Jumlah Cluster")
+        gr.Image(type="pil", label="Gambar 1"),
+        gr.Image(type="pil", label="Gambar 2")
     ],
-    outputs=gr.Textbox(label="Hasil Clustering"),
-    title="🧠 Unsupervised Image Clustering",
-    description="Aplikasi unsupervised learning: mengelompokkan gambar TANPA dataset & TANPA label."
+    outputs=gr.Textbox(label="Hasil Analisis"),
+    title="🧠 Unsupervised Image Similarity",
+    description=(
+        "Aplikasi unsupervised learning untuk mengukur kemiripan dua gambar "
+        "tanpa dataset dan tanpa label."
+    )
 )
 
 app.launch()