- Semantic Search in Frigate allows you to find tracked objects within
- your review items using either the image itself, a user-defined text
- description, or an automatically generated one. This feature works
- by creating embeddings — numerical vector representations — for both
- the images and text descriptions of your tracked objects. By
- comparing these embeddings, Frigate assesses their similarities to
- deliver relevant search results.
-
+
+
+ Semantic Search
+
+
+
+
+ Semantic Search in Frigate allows you to find tracked objects
+ within your review items using either the image itself, a
+ user-defined text description, or an automatically generated one.
+ This feature works by creating embeddings — numerical vector
+ representations — for both the images and text descriptions of
+ your tracked objects. By comparing these embeddings, Frigate
+ assesses their similarities to deliver relevant search results.
+
-
-
- Read the semantic search docs
-
-
+
+
+ Read the Documentation
+
+
+
-
-
+
+
@@ -210,23 +216,24 @@ export default function SearchSettingsView({
/>
-
+
-
Model Size
+
Model Size
- Configure the size of the model used for semantic search
- embeddings:
-
-
- • Configuring the small model employs a quantized
- version of the model that uses much less RAM and runs faster
- on CPU with a very negligible difference in embedding quality.
-
-
- • Configuring the large model employs the full Jina
- model and will automatically run on the GPU if applicable.
+ The size of the model used for semantic search embeddings.
+
+
+ Using small employs a quantized version of the
+ model that uses much less RAM and runs faster on CPU with a
+ very negligible difference in embedding quality.
+
+
+ Using large employs the full Jina model and will
+ automatically run on the GPU if applicable.
+