phase 12

detect/stages/registry/preprocessing.py

@@ -1,4 +1,4 @@
-"""Registration for preprocessing stages: frame extraction, scene filter."""
+"""Registration for preprocessing stages: frame extraction, scene filter, image preprocessing."""
 
 from detect.stages.base import StageDefinition, StageIO, StageConfigField, register_stage
 from ._serializers import serialize_frames, deserialize_frames
@@ -25,6 +25,17 @@ def _deser_filter(data: dict, job_id: str) -> dict:
     return {"_filtered_sequences": data["filtered_frame_sequences"]}
 
 
+def _ser_preprocess(state: dict, job_id: str) -> dict:
+    # Preprocessed crops are numpy arrays — regenerable from frames + boxes + config
+    crops = state.get("preprocessed_crops", {})
+    return {"crop_keys": list(crops.keys()), "count": len(crops)}
+
+
+def _deser_preprocess(data: dict, job_id: str) -> dict:
+    # Crops are regenerable — no need to restore from checkpoint
+    return {"preprocessed_crops": {}}
+
+
 def register():
     extract = StageDefinition(
         name="extract_frames",
@@ -55,3 +66,22 @@ def register():
         deserialize_fn=_deser_filter,
     )
     register_stage(scene_filter)
+
+    preprocess = StageDefinition(
+        name="preprocess",
+        label="Preprocess",
+        description="Image preprocessing on detected regions before OCR",
+        category="preprocessing",
+        io=StageIO(
+            reads=["filtered_frames", "boxes_by_frame"],
+            writes=["preprocessed_crops"],
+        ),
+        config_fields=[
+            StageConfigField("contrast", "bool", True, "CLAHE contrast enhancement"),
+            StageConfigField("deskew", "bool", False, "Correct slight rotation"),
+            StageConfigField("binarize", "bool", False, "Otsu binarization"),
+        ],
+        serialize_fn=_ser_preprocess,
+        deserialize_fn=_deser_preprocess,
+    )
+    register_stage(preprocess)