feat(visualizer): add cosine similarity computation for image comparison

mini-nav/tests/test_visualizer.py

@@ -5,6 +5,7 @@ import io
 
 import numpy as np
 from PIL import Image
+from sklearn.metrics.pairwise import cosine_similarity
 
 
 class TestImageUploadSimilaritySearch:
@@ -33,3 +34,44 @@ class TestImageUploadSimilaritySearch:
         # Verify the image is valid
         assert parsed_img.size == (224, 224)
         assert parsed_img.mode == "RGB"
+
+
+class TestCosineSimilarity:
+    """Test suite for cosine similarity computation between feature vectors."""
+
+    def test_identical_vectors_return_one(self):
+        """Identical vectors should have cosine similarity of 1.0."""
+        vec = np.random.randn(1024).tolist()
+        similarity = cosine_similarity([vec], [vec])[0][0]
+        assert np.isclose(similarity, 1.0)
+
+    def test_orthogonal_vectors_return_zero(self):
+        """Orthogonal vectors should have cosine similarity of 0.0."""
+        vec_a = [1.0, 0.0]
+        vec_b = [0.0, 1.0]
+        similarity = cosine_similarity([vec_a], [vec_b])[0][0]
+        assert np.isclose(similarity, 0.0)
+
+    def test_opposite_vectors_return_negative_one(self):
+        """Opposite vectors should have cosine similarity of -1.0."""
+        vec_a = [1.0, 0.0, 0.0]
+        vec_b = [-1.0, 0.0, 0.0]
+        similarity = cosine_similarity([vec_a], [vec_b])[0][0]
+        assert np.isclose(similarity, -1.0)
+
+    def test_similarity_range(self):
+        """Cosine similarity should always be within [-1, 1]."""
+        # Random vectors
+        for _ in range(10):
+            vec_a = np.random.randn(1024).tolist()
+            vec_b = np.random.randn(1024).tolist()
+            similarity = cosine_similarity([vec_a], [vec_b])[0][0]
+            assert -1.0 <= similarity <= 1.0
+
+    def test_similarity_with_list_input(self):
+        """Cosine similarity should work with Python list inputs (as stored in dcc.Store)."""
+        # Simulate feature vectors stored as Python lists in dcc.Store
+        vec_a = [0.1, 0.2, 0.3, 0.4, 0.5]
+        vec_b = [0.1, 0.2, 0.3, 0.4, 0.5]
+        similarity = cosine_similarity([vec_a], [vec_b])[0][0]
+        assert np.isclose(similarity, 1.0)

mini-nav/visualizer/app.py

@@ -9,6 +9,7 @@ from dash import Dash, Input, Output, State, callback, dcc, html
 from database import db_manager
 from feature_retrieval import FeatureRetrieval
 from PIL import Image
+from sklearn.metrics.pairwise import cosine_similarity
 from transformers import AutoImageProcessor, AutoModel
 from visualizer.events import CellClickedEvent
 
@@ -114,6 +115,18 @@ class APP(Dash):
                             direction="row",
                             wrap="wrap",
                         ),
+                        # Store feature vectors for cosine similarity computation
+                        dcc.Store(id="store-upload-vector"),
+                        dcc.Store(id="store-select-vector"),
+                        # Cosine similarity display (below image comparison)
+                        html.Div(
+                            id="cosine-similarity-display",
+                            style={
+                                "textAlign": "center",
+                                "padding": "10px",
+                                "margin": "10px",
+                            },
+                        ),
                         dag.AgGrid(id="ag-grid"),
                     ],
                     gap="md",
@@ -130,6 +143,7 @@ class APP(Dash):
     Output("output-image-upload", "children"),
     Output("ag-grid", "rowData"),
     Output("ag-grid", "columnDefs"),
+    Output("store-upload-vector", "data"),
     Input("upload-image", "contents"),
     State("upload-image", "filename"),
     State("upload-image", "last_modified"),
@@ -177,7 +191,7 @@ def update_output(
             dmc.Text(f"{feature_vector[:5]}", size="xs"),
         ]
 
-        return children, row_data, columnDefs
+        return children, row_data, columnDefs, feature_vector
     else:
         # When contents is None
         # Exclude 'vector' and 'binary' columns as they are not JSON serializable
@@ -190,11 +204,12 @@ def update_output(
             for column in df.columns
         ]
 
-        return [], row_data, columnDefs
+        return [], row_data, columnDefs, None
 
 
 @callback(
     Output("output-image-select", "children"),
+    Output("store-select-vector", "data"),
     Input("ag-grid", "cellClicked"),
     State("ag-grid", "rowData"),
 )
@@ -208,18 +223,18 @@ def update_images_comparison(
         or row_data is None
         or len(row_data) == 0
     ):
-        return []
+        return [], None
 
     # Get the selected row's data
     row_index = int(clicked_event["rowIndex"])
     if row_index >= len(row_data):
-        return []
+        return [], None
 
     selected_row = row_data[row_index]
     image_id = selected_row.get("id")
 
     if image_id is None:
-        return []
+        return [], None
 
     # Query database for binary data using the id
     result = (
@@ -231,7 +246,7 @@ def update_images_comparison(
     )
 
     if result.height == 0:
-        return []
+        return [], None
 
     # Get binary data
     binary_data = result.row(0, named=True)["binary"]
@@ -268,7 +283,34 @@ def update_images_comparison(
         dmc.Text(f"{vector[:5]}", size="xs"),
     ]
 
-    return children
+    # Convert vector to list for JSON serialization in dcc.Store
+    select_vector = vector if isinstance(vector, list) else list(vector)
+
+    return children, select_vector
+
+
+@callback(
+    Output("cosine-similarity-display", "children"),
+    Input("store-upload-vector", "data"),
+    Input("store-select-vector", "data"),
+)
+def update_cosine_similarity(
+    upload_vector: Optional[List[float]],
+    select_vector: Optional[List[float]],
+):
+    """Compute and display cosine similarity when both vectors are available."""
+    # Only display when both upload and select images are present
+    if upload_vector is None or select_vector is None:
+        return []
+
+    similarity = cosine_similarity([upload_vector], [select_vector])[0][0]
+
+    return dmc.Text(
+        f"Cosine Similarity: {similarity:.4f}",
+        size="lg",
+        fw=700,
+        ta="center",
+    )
 
 
 app = APP()