Fix: deduplicate images

app.py

@@ -3,7 +3,7 @@ from qdrant_client import QdrantClient
 
 from config import api_key, article, collection_name, description, examples, host_url
 from encoder import encode_text
-from utils import get_file
+from utils import get_images
 
 client = QdrantClient(host=host_url,
                       port=443,
@@ -14,10 +14,9 @@ def search_images(query, modality):
     query_vector = encode_text(query)
     vector_name = "image" if modality == "images" else "text"
     results = client.search(
-        collection_name, (vector_name, query_vector), limit=6, with_payload=True)
+        collection_name, (vector_name, query_vector), limit=20, with_payload=True)
 
-    images = [(get_file(result.payload['url']), result.payload['caption'] +
-               " - " + str(result.score)) for result in results]
+    images = get_images(results)
 
     return images
 

utils.py

@@ -8,3 +8,17 @@ from PIL import Image
 @lru_cache
 def get_file(url):
     return Image.open(io.BytesIO(requests.get(url).content))
+
+
+def get_images(results):
+    seen_urls = set()
+    unique_results = []
+    for result in results:
+        if not result.payload['url'] in seen_urls:
+            seen_urls.add(result.payload['url'])
+            unique_results.append(result)
+
+    images = [(get_file(result.payload['url']), result.payload['caption'] +
+               " - " + str(result.score)) for result in unique_results]
+
+    return images