Added image embedding as playground and roll backed for create_album end point to not use k mean clustring

cloudzy/embedding/image_embedding.py

@@ -0,0 +1,53 @@
+from transformers import AutoModel, AutoProcessor
+from PIL import Image
+import requests
+import numpy as np
+import torch
+from io import BytesIO
+
+# Load model and processor directly
+model = AutoModel.from_pretrained("jinaai/jina-clip-v2", trust_remote_code=True)
+processor = AutoProcessor.from_pretrained("jinaai/jina-clip-v2", trust_remote_code=True)
+
+texts = ["Woman taking pictures on a road trip.", "delicious fruits glowing under sunlight"]
+# Process and encode text
+text_inputs = processor(text=texts, return_tensors="pt", padding=True)
+with torch.no_grad():
+    text_embeddings = model.get_text_features(**text_inputs)
+text_embeddings = text_embeddings.cpu().numpy()
+print("Text embeddings shape:", text_embeddings.shape)
+
+image_paths = [
+    "/Users/komeilfathi/Documents/hf_deploy_test/cloudzy_ai_challenge/uploads/img_1_20251026_014959_886.jpg",
+    "/Users/komeilfathi/Documents/hf_deploy_test/cloudzy_ai_challenge/uploads/img_9_20251024_185602_319.webp"
+]
+images = []
+for path in image_paths:
+    try:
+        img = Image.open(path).convert("RGB")
+        images.append(img)
+        print(f"✓ Loaded image from {path}")
+    except Exception as e:
+        print(f"✗ Failed to load image from {path}: {e}")
+
+# Process and encode images
+if images:
+    image_inputs = processor(images=images, return_tensors="pt")
+    with torch.no_grad():
+        image_embeddings = model.get_image_features(**image_inputs)
+    image_embeddings = image_embeddings.cpu().numpy()
+    print("Image embeddings shape:", image_embeddings.shape)
+else:
+    print("⚠ No images loaded successfully")
+    image_embeddings = np.array([])
+
+def cosine_similarity(a, b):
+    return np.dot(a, b) / (np.linalg.norm(a) * np.linalg.norm(b))
+
+if len(image_embeddings) > 0:
+    for i, t_emb in enumerate(text_embeddings):
+        for j, i_emb in enumerate(image_embeddings):
+            sim = cosine_similarity(t_emb, i_emb)
+            print(f"Similarity between text {i} and image {j}: {sim:.4f}")
+else:
+    print("No images to compare similarity with")

cloudzy/models.py

@@ -12,6 +12,7 @@ class Photo(SQLModel, table=True):
     filepath: str  # Full path to stored image
     tags: str = Field(default="[]")  # JSON string of tags
     caption: str = Field(default="")
+    description: str = Field(default="")
     embedding: Optional[str] = Field(default=None)  # JSON string of embedding vector
     created_at: datetime = Field(default_factory=datetime.utcnow)
     

cloudzy/routes/photo.py

@@ -37,7 +37,7 @@ async def get_photo(
         image_url = f"{APP_DOMAIN}uploads/{photo.filename}",
         tags=photo.get_tags(),
         caption=photo.caption,
-        # embedding=photo.get_embedding(),
+        description=photo.description,
         created_at=photo.created_at,
     )
 
@@ -72,7 +72,7 @@ async def list_photos(
             image_url = f"{APP_DOMAIN}uploads/{photo.filename}",
             tags=photo.get_tags(),
             caption=photo.caption,
-            # embedding=photo.get_embedding(),
+            description=photo.description,
             created_at=photo.created_at,
         )
         for photo in photos
@@ -89,7 +89,7 @@ async def get_albums(
     """
 
     search_engine = SearchEngine()
-    albums_ids = search_engine.create_albums_kmeans(top_k=top_k)
+    albums_ids = search_engine.create_albums(top_k=top_k)
     APP_DOMAIN = os.getenv("APP_DOMAIN") or "http://127.0.0.1:8000/"
     summarizer = TextSummarizer()
 
@@ -127,6 +127,7 @@ async def get_albums(
                     image_url=f"{APP_DOMAIN}uploads/{photo.filename}",
                     tags=photo.get_tags(),
                     caption=photo.caption,
+                    description=photo.description,
                     distance=float(distance_val),
                 )
             )

cloudzy/routes/search.py

@@ -58,6 +58,7 @@ async def search_photos(
                     image_url=f"{APP_DOMAIN}uploads/{photo.filename}",
                     tags=photo.get_tags(),
                     caption=photo.caption,
+                    description=photo.description,
                     distance=distance,
                 )
             )

cloudzy/routes/upload.py

@@ -127,6 +127,7 @@ def process_image_in_background(photo_id: int, filepath: str):
             photo = session.exec(select(Photo).where(Photo.id == photo_id)).first()
             if photo:
                 photo.caption = caption
+                photo.description = description
                 photo.set_tags(tags)
                 photo.set_embedding(embedding.tolist())
                 session.add(photo)

cloudzy/schemas.py

@@ -9,6 +9,7 @@ class PhotoResponse(BaseModel):
     id: int
     filename: str
     image_url: str
+    description: str
     tags: List[str]
     caption: str
     created_at: datetime
@@ -31,6 +32,7 @@ class SearchResult(BaseModel):
     image_url: str
     tags: List[str]
     caption: str
+    description: str
     distance: float  # L2 distance (lower is more similar)
     
     class Config:
@@ -60,6 +62,7 @@ class PhotoItem(BaseModel):
     image_url: str
     tags: List[str]
     caption: str
+    description: str
     distance: float
 
 class AlbumItem(BaseModel):