refactor(object_detector): add return type annotation

src/nrtk_explorer/app/image_meta.py

@@ -1,11 +1,10 @@
 from typing import TypedDict
 from nrtk_explorer.app.trame_utils import delete_state
-from nrtk_explorer.app.image_ids import DatasetId
 
 ImageMetaId = str
 
 
-def image_id_to_meta(image_id: DatasetId) -> ImageMetaId:
+def image_id_to_meta(image_id: str) -> ImageMetaId:
     return f"meta_{image_id}"
 
 
@@ -26,14 +25,14 @@ class DatasetImageMeta(TypedDict):
 }
 
 
-def update_image_meta(state, dataset_id: DatasetId, meta_patch: PartialDatasetImageMeta):
+def update_image_meta(state, dataset_id: str, meta_patch: PartialDatasetImageMeta):
     meta_key = image_id_to_meta(dataset_id)
     current_meta = {}
     if state.has(meta_key) and state[meta_key] is not None:
         current_meta = state[meta_key]
     state[meta_key] = {**IMAGE_META_DEFAULTS, **current_meta, **meta_patch}
 
 
-def delete_image_meta(state, dataset_id: DatasetId):
+def delete_image_meta(state, dataset_id: str):
     meta_key = image_id_to_meta(dataset_id)
     delete_state(state, meta_key)

src/nrtk_explorer/library/object_detector.py

@@ -2,10 +2,12 @@
 import torch
 import transformers
 
-from typing import Optional
+from typing import Optional, Sequence
 
 from nrtk_explorer.library import images_manager
 
+ImageIdToAnnotations = dict[str, Sequence[dict]]
+
 
 class ObjectDetector:
     """Object detection using Hugging Face's transformers library"""
@@ -58,7 +60,7 @@ def eval(
         image_ids: list[str],
         content: Optional[dict] = None,
         batch_size: int = 32,
-    ):
+    ) -> ImageIdToAnnotations:
         """Compute object recognition. Returns Annotations grouped by input image paths."""
 
         # Some models require all the images in a batch to be the same size,