Update app.py

app.py

@@ -89,14 +89,10 @@ def segment_clothing(img, clothes=["Hat", "Upper-clothes", "Skirt", "Pants", "Dr
 def find_similar_images(query_embedding, collection, top_k=5):
     # 모든 임베딩을 가져옴
     all_embeddings = collection.get(include=['embeddings'])['embeddings']
-    if len(all_embeddings) == 0:
-        st.error("No embeddings found in the collection.")
-        return []
-
     database_embeddings = np.array(all_embeddings)
     
     # 유사도 계산
-    similarities = cosine_similarity(database_embeddings, query_embedding.reshape(1, -1)).squeeze()
+    similarities = np.dot(database_embeddings, query_embedding.T).squeeze()
     top_indices = np.argsort(similarities)[::-1][:top_k]
     
     # 메타데이터 가져옴
@@ -108,23 +104,13 @@ def find_similar_images(query_embedding, collection, top_k=5):
     # 벡터 ID를 함께 가져와서 확인
     vector_data = collection.get(include=['embeddings', 'metadatas', 'ids'])
 
-    for idx, vector_id in enumerate(vector_data['ids']):
-        st.write(f"ID: {vector_id}, Embedding: {vector_data['embeddings'][idx]}")
-    if len(all_data) == 0:
-        st.error("No metadatas found in the collection.")
-        return []
-    
     top_metadatas = [all_data[idx] for idx in top_indices]
     
     results = []
     for idx, metadata in enumerate(top_metadatas):
-        image_urls = metadata['image_url'].split(',')
-        representative_image_url = image_urls[0] if image_urls else None
-        
         results.append({
             'info': metadata,
-            'similarity': similarities[top_indices[idx]],
-            'image_url': representative_image_url
+            'similarity': similarities[top_indices[idx]]
         })
     return results