Make Using XMP Metadata to Enhance Image Search Optional, Configurable

- Break the compute embeddings method into separate methods:
  compute_image_embeddings and compute_metadata_embeddings

- If image_metadata_embeddings isn't defined, do not use it to enhance
  search results. Given image_metadata_embeddings wouldn't be defined
  if use_xmp_metadata is False, we can avoid unnecessary addition of
  args to query method

sample_config.yml

@@ -12,6 +12,7 @@ content-type:
   image:
     embeddings-file: '.image_embeddings.pt'
     batch-size: 50
+    use-xmp-metadata: 'no'
 
 search-type:
   asymmetric:

src/main.py

@@ -189,6 +189,7 @@ if __name__ == '__main__':
             pathlib.Path(image_config['embeddings-file']),
             batch_size=image_config['batch-size'],
             regenerate=args.regenerate,
+            use_xmp_metadata={'yes': True, 'no': False}[image_config['use-xmp-metadata']],
             verbose=args.verbose)
 
     # Start Application Server

src/search_type/image_search.py

@@ -30,10 +30,17 @@ def extract_entries(image_directory, verbose=0):
     return image_names
 
 
-def compute_embeddings(image_names, model, embeddings_file, batch_size=50, regenerate=False, verbose=0):
+def compute_embeddings(image_names, model, embeddings_file, batch_size=50, regenerate=False, use_xmp_metadata=False, verbose=0):
     "Compute (and Save) Embeddings or Load Pre-Computed Embeddings"
+
+    image_embeddings = compute_image_embeddings(image_names, model, embeddings_file, batch_size, regenerate, verbose)
+    image_metadata_embeddings = compute_metadata_embeddings(image_names, model, embeddings_file, batch_size, use_xmp_metadata, regenerate, verbose)
+
+    return image_embeddings, image_metadata_embeddings
+
+
+def compute_image_embeddings(image_names, model, embeddings_file, batch_size=50, regenerate=False, verbose=0):
     image_embeddings = None
-    image_metadata_embeddings = None
 
     # Load pre-computed image embeddings from file if exists
     if resolve_absolute_path(embeddings_file).exists() and not regenerate:
@@ -41,16 +48,7 @@ def compute_embeddings(image_names, model, embeddings_file, batch_size=50, regen
         if verbose > 0:
             print(f"Loaded pre-computed embeddings from {embeddings_file}")
 
-    # load pre-computed image metadata embedding file if exists
-    if resolve_absolute_path(f"{embeddings_file}_metadata").exists() and not regenerate:
-        image_metadata_embeddings = torch.load(f"{embeddings_file}_metadata")
-        if verbose > 0:
-            print(f"Loaded pre-computed embeddings from {embeddings_file}_metadata")
-
-    if image_embeddings is None or image_metadata_embeddings is None:  # Else compute the image_embeddings from scratch, which can take a while
-        if verbose > 0:
-            print(f"Loading the {len(image_names)} images into memory")
-
+    # Else compute the image embeddings from scratch, which can take a while
     if image_embeddings is None:
         image_embeddings = []
         for index in trange(0, len(image_names), batch_size):
@@ -60,7 +58,20 @@ def compute_embeddings(image_names, model, embeddings_file, batch_size=50, regen
         if verbose > 0:
             print(f"Saved computed embeddings to {embeddings_file}")
 
-    if image_metadata_embeddings is None:
+    return image_embeddings
+
+
+def compute_metadata_embeddings(image_names, model, embeddings_file, batch_size=50, regenerate=False, use_xmp_metadata=False, verbose=0):
+    image_metadata_embeddings = None
+
+    # Load pre-computed image metadata embedding file if exists
+    if use_xmp_metadata and resolve_absolute_path(f"{embeddings_file}_metadata").exists() and not regenerate:
+        image_metadata_embeddings = torch.load(f"{embeddings_file}_metadata")
+        if verbose > 0:
+            print(f"Loaded pre-computed embeddings from {embeddings_file}_metadata")
+
+    # Else compute the image metadata embeddings from scratch, which can take a while
+    if use_xmp_metadata and image_metadata_embeddings is None:
         image_metadata_embeddings = []
         for index in trange(0, len(image_names), batch_size):
             image_metadata = [extract_metadata(image_name, verbose) for image_name in image_names[index:index+batch_size]]
@@ -69,7 +80,7 @@ def compute_embeddings(image_names, model, embeddings_file, batch_size=50, regen
         if verbose > 0:
             print(f"Saved computed metadata embeddings to {embeddings_file}_metadata")
 
-    return image_embeddings, image_metadata_embeddings
+    return image_metadata_embeddings
 
 
 def extract_metadata(image_name, verbose=0):
@@ -102,13 +113,14 @@ def query_images(query, image_embeddings, image_metadata_embeddings, model, coun
                   in util.semantic_search(query_embedding, image_embeddings, top_k=count)[0]}
 
     # Compute top_k ranked images based on cosine-similarity b/w query and all image metadata embeddings.
-    metadata_hits = {result['corpus_id']: result['score']
-                     for result
-                     in util.semantic_search(query_embedding, image_metadata_embeddings, top_k=count)[0]}
+    if image_metadata_embeddings:
+        metadata_hits = {result['corpus_id']: result['score']
+                         for result
+                         in util.semantic_search(query_embedding, image_metadata_embeddings, top_k=count)[0]}
 
-    # Sum metadata, image scores of the highest ranked images
-    for corpus_id, score in metadata_hits.items():
-        image_hits[corpus_id] = image_hits.get(corpus_id, 0) + score
+        # Sum metadata, image scores of the highest ranked images
+        for corpus_id, score in metadata_hits.items():
+            image_hits[corpus_id] = image_hits.get(corpus_id, 0) + score
 
     # Reformat results in original form from sentence transformer semantic_search()
     hits = [{'corpus_id': corpus_id, 'score': score} for corpus_id, score in image_hits.items()]
@@ -138,7 +150,7 @@ def collate_results(hits, image_names, image_directory, count=5):
         in hits[0:count]]
 
 
-def setup(image_directory, embeddings_file, batch_size=50, regenerate=False, verbose=0):
+def setup(image_directory, embeddings_file, batch_size=50, regenerate=False, use_xmp_metadata=False, verbose=0):
     # Initialize Model
     model = initialize_model()
 
@@ -148,7 +160,8 @@ def setup(image_directory, embeddings_file, batch_size=50, regenerate=False, ver
 
     # Compute or Load Embeddings
     embeddings_file = resolve_absolute_path(embeddings_file)
-    image_embeddings, image_metadata_embeddings = compute_embeddings(image_names, model, embeddings_file, batch_size=batch_size, regenerate=regenerate, verbose=verbose)
+    image_embeddings, image_metadata_embeddings = compute_embeddings(image_names, model, embeddings_file,
+                                                                     batch_size=batch_size, regenerate=regenerate, use_xmp_metadata=use_xmp_metadata, verbose=verbose)
 
     return image_names, image_embeddings, image_metadata_embeddings, model
 

src/utils/cli.py

@@ -57,7 +57,8 @@ default_config = {
         'image':
         {
             'embeddings-file': '.image_embeddings.pt',
-            'batch-size': 50
+            'batch-size': 50,
+            'use-xmp-metadata': 'no'
         },
         'music':
         {