scripts for model creation reflect new vector store abstraction

README.md

@@ -71,6 +71,25 @@ Please [see its documentation](https://github.com/NERC-CEH/object_store_api) for
 
 `python src/os_api/api.py`
 
+## Feature extraction API
+
+FastAPI wrapper around different models - POST an image URL, get back embeddings 
+
+## Label Studio ML backend
+
+Pre-annotation backend for Label Studio following their standard pattern.
+
+Build an image embedding model which will assign a likely-detritus tag:
+
+```
+cd scripts
+dvc repro
+```
+
+Application is in `src/label_studio_cyto_ml`
+
+[Setup documentation](src/label_studio_cyto_ml/README.md)
+
 ## Pipelines
 
 ### DVC 

scripts/cluster.py

@@ -6,7 +6,7 @@
 import yaml
 
 from sklearn.cluster import KMeans
-from cyto_ml.data.vectorstore import embeddings, vector_store
+from cyto_ml.data.vectorstore import vector_store
 
 
 def main() -> None:
@@ -25,8 +25,8 @@ def main() -> None:
         n_clusters = 5
 
     kmeans = KMeans(n_clusters=n_clusters, random_state=42)
-    store = vector_store(collection_name)
-    X = embeddings(store)
+    store = vector_store("sqlite", collection_name)
+    X = store.embeddings()
     kmeans.fit(X)
 
     # We supply a -o for output directory - this doesn't ensure we write there.