Update app.py

app.py

@@ -74,14 +74,14 @@ def load_index(index_path):
     """Load FAISS index once and cache it"""
     return faiss.read_index(index_path)
 
-def distance_to_similarity(distances, temp=1e-4):
+def distance_to_similarity(distance, temp=1e-4):
     """Convert distance to similarity"""
-    for ii in range(len(distances)):
-        contribs = distances[ii].max() - distances[ii]
+    for ii in range(len(distance)):
+        contribs = distance[ii].max() - distance[ii]
         contribs = contribs / temp
         sum_contribs = np.exp(contribs).sum()
-        distances[ii] = np.exp(contribs) / sum_contribs
-    return distances
+        distance[ii] = np.exp(contribs) / sum_contribs
+    return distance
 
 def calculate_rewards(subscription, num_generations, author_share, ro_share, num_users_k, similarities, num_authors=1800):
     """Calculate rewards based on user inputs and similarities"""
@@ -110,7 +110,7 @@ def calculate_rewards(subscription, num_generations, author_share, ro_share, num
             'attribution': f"{sim * 100:.0f}%",
             'author_month_reward': f"{author_month_reward:.0f}€",
             'ro_month_reward': f"{ro_month_reward:.0f}€",
-            'raw_similarity': sim_value
+            'raw_similarity': distances
         })
 
     return rewards
@@ -159,12 +159,32 @@ def select_preset(preset_id):
         params = DEFAULT_PARAMS.copy()
         
         # Get features and search
-        features = get_ft(model, image)        #extrait les features, soit le vecteur représentant l’image
-        distances, indices = get_topk(index, features, topk=params['num_neighbors'])
+        #features = get_ft(model, image)        #extrait les features, soit le vecteur représentant l’image
+        #distances, indices = get_topk(index, features, topk=params['num_neighbors'])
                                                #utilise l’index pour trouver les k voisins de l’image
                                                #retourne 
                                                # - distances avec les voisins
                                                # - indices : les positions (dans l'index) des voisins
+        # Process image
+        #features = get_ft(model, image)                                                ########   extrait le vecteur de l'image
+        _, indices = get_topk(index, features, topk=params['num_neighbors'])   ########   extrait les distances avec les premiers voisins
+        
+        # Supposons que vous ayez une fonction pour charger une image
+        def load_image(path):
+            return Image.open(path).convert('RGB')
+        
+        # Charger les deux images
+        image1 = load_image(PRESET_IMAGES[1])
+        image2 = load_image(PRESET_IMAGES[2])
+        
+        # Extraire les features
+        features1 = get_ft(model, image1)
+        features2 = get_ft(model, image2)
+        
+        # Calculer la distance euclidienne
+        distances = np.linalg.norm(features1 - features2)
+        
+        print(f"Distance euclidienne entre l'image 1 et l'image 2 : {distances}")
         
         # Collect valid results first
         valid_results = []