refactor: rename object_detector to predictor

src/nrtk_explorer/app/transforms.py

@@ -52,6 +52,8 @@
 
 
 class LazyDict(Mapping):
+    """If function provided for value, run function when value is accessed"""
+
     def __init__(self, *args, **kw):
         self._raw_dict = dict(*args, **kw)
 
@@ -244,7 +246,7 @@ def delete_meta_state(old_ids, new_ids):
     def on_server_ready(self, *args, **kwargs):
         self.state.change("inference_model")(self.on_inference_model_change)
         self.state.change("current_dataset")(self._cancel_update_images)
-        self.state.change("current_dataset")(self.reset_detector)
+        self.state.change("current_dataset")(self.reset_predictor)
         self.state.change("confidence_score_threshold")(self._start_update_images)
 
     def on_inference_model_change(self, **kwargs):
@@ -253,7 +255,7 @@ def on_inference_model_change(self, **kwargs):
         self.predictor.set_model(self.state.inference_model)
         self._start_update_images()
 
-    def reset_detector(self, **kwargs):
+    def reset_predictor(self, **kwargs):
         self.predictor.reset()
 
     def set_on_transform(self, fn):
@@ -403,6 +405,13 @@ def _cancel_update_images(self, **kwargs):
             self._update_task.cancel()
 
     def _start_update_images(self, **kwargs):
+        """
+        After updating the images visible in the image list, all other selected
+        images are updated and their scores computed. After images are scored,
+        the table sort may have changed the images that are visible, so
+        ImageList is asked to send visible image IDs again, which may trigger
+        a new _update_all_images if the set of images in view has changed.
+        """
         self._cancel_update_images()
         self._update_task = asynchronous.create_task(
             self._update_all_images(self.visible_dataset_ids)

src/nrtk_explorer/library/multiprocess_predictor.py

@@ -4,12 +4,12 @@
 import logging
 import queue
 import uuid
-from .object_detector import ObjectDetector
+from .predictor import Predictor
 
 
 def _child_worker(request_queue, result_queue, model_name, force_cpu):
     logger = logging.getLogger(__name__)
-    detector = ObjectDetector(model_name=model_name, force_cpu=force_cpu)
+    predictor = Predictor(model_name=model_name, force_cpu=force_cpu)
 
     while True:
         try:
@@ -28,7 +28,7 @@ def _child_worker(request_queue, result_queue, model_name, force_cpu):
 
         if command == "SET_MODEL":
             try:
-                detector = ObjectDetector(
+                predictor = Predictor(
                     model_name=payload["model_name"],
                     force_cpu=payload["force_cpu"],
                 )
@@ -38,14 +38,14 @@ def _child_worker(request_queue, result_queue, model_name, force_cpu):
                 result_queue.put((req_id, {"status": "ERROR", "message": str(e)}))
         elif command == "INFER":
             try:
-                predictions = detector.eval(payload["images"])
+                predictions = predictor.eval(payload["images"])
                 result_queue.put((req_id, {"status": "OK", "result": predictions}))
             except Exception as e:
                 logger.exception("Inference failed.")
                 result_queue.put((req_id, {"status": "ERROR", "message": str(e)}))
         elif command == "RESET":
             try:
-                detector.reset()
+                predictor.reset()
                 result_queue.put((req_id, {"status": "OK"}))
             except Exception as e:
                 logger.exception("Reset failed.")

src/nrtk_explorer/library/object_detector.py → src/nrtk_explorer/library/predictor.py

@@ -17,8 +17,7 @@ class ImageWithId(NamedTuple):
 STARTING_BATCH_SIZE = 32
 
 
-class ObjectDetector:
-    """Object detection using Hugging Face's transformers library"""
+class Predictor:
 
     def __init__(
         self,
@@ -106,7 +105,7 @@ def eval(
                     self.batch_size = self.batch_size // 2
                     self.batch_size = self.batch_size
                     print(
-                        f"Caught out of memory exception:\n{e}\nWas batch_size={previous_batch_size}, setting batch_size={self.batch_size}"
+                        f"Changing pipeline batch_size from {previous_batch_size} to {self.batch_size} because caught out of memory exception:\n{e}"
                     )
                 else:
                     raise

tests/test_object_detector.py → tests/test_predictor.py

@@ -1,15 +1,15 @@
 import pytest
-from nrtk_explorer.library import object_detector
+from nrtk_explorer.library.predictor import Predictor
+from nrtk_explorer.library.multiprocess_predictor import MultiprocessPredictor
 from nrtk_explorer.library.scoring import compute_score
 from nrtk_explorer.library.dataset import get_dataset
 from utils import get_images, DATASET
-from nrtk_explorer.library.multiprocess_predictor import MultiprocessPredictor
 
 
-def test_detector_small():
+def test_predictor_small():
     sample = get_images()
-    detector = object_detector.ObjectDetector(model_name="hustvl/yolos-tiny")
-    img = detector.eval(sample)
+    predictor = Predictor(model_name="hustvl/yolos-tiny")
+    img = predictor.eval(sample)
     assert len(img) == len(sample.keys())
 
 
@@ -46,8 +46,8 @@ def test_set_model(predictor):
 def test_scorer():
     ds = get_dataset(DATASET)
     sample = get_images()
-    detector = object_detector.ObjectDetector(model_name="facebook/detr-resnet-50")
-    predictions = detector.eval(sample)
+    predictor = Predictor(model_name="facebook/detr-resnet-50")
+    predictions = predictor.eval(sample)
 
     dataset_annotations = dict()
     for annotation in ds.anns.values():